r
1...-.. ....
....J&........~z .
t) 0.16341 1/
a‘, Q I.
«an... 3
1......
.{u .2

3.8mm.
. in 25
x . 3....
. 2......- . .
uranhnwiwflr 345......
2191-... Mg. 51.1.3.9. a?
..o.ll:t-i utfl.
.. . a:
2 2.!
4.1.22-
5.2.5:...
. u 2
6:1!”
. , , . iii]. J
, . . , . . . “34...:
2...... . . .
..

Ia

 

. . 2
. 23:21.... .
m... .91.... ,. hfiunfififiwgf

.

v
.p

m).

t‘nl‘g . ::i:‘ l
x . 5:34.... .
3.39.4

Tu...

in...
5).."vG-‘r.

."s

.

 

 

I r :1..-
‘l:: .
. .1:£..v}!.k!t‘.§3£.nifl...
.3. ”Earn... 11.35....l52 t3!
. . hmiili5lel 5101:.
3:32.... a .a.
I.
.515211
I: trill...
$.13... Eh“:
.. 95..
7-13... 5...
Fig,-
35.‘a$

tehmuwuifl
Styli! -. :

 

i .
1.. .11.: 11v}¢!l2.....! 13...:
.82.! is. .131!

 

 

 

{I‘lritii
rt 5 r 11. .331;
.6.“

_....w..........

.95: ilinlvli‘lti
x \h 5.. z...- .3
y 9..

rit- I):

 

égfifim s.

.1.

¢°°L

This is to certify that the

dissertation entitled

Police Social Capital and Officer
Perfomnnce of Cmmmity Policing

presented by

Pmanda L. Robinson

has been accepted towards fulfillment
of the requirements for

Ph . D. .
degree in

 

Interdisciplinary Social

 

Sciences

 

Date W I 3/ 200/

MS U i: an Affirmative Action/Equal Opportunity Institution

0-12771

 

 

_ LIBRARY
M'Chigah State
University

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE ' DATE DUE

 

MAY 3 0 2005

 

JUL 1 4 2005

 

”.3830 08

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:lClRC/DateDue.p65-p.15

 

POLICE SOCIAL CAPITAL AND OFFICER PERFORMANCE
OF COMMUNITY POLICING
By

Amanda L. Robinson

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
College of Social Science

2002

ABSTRACT
POLICE SOCIAL CAPITAL AND OFFICER PERFORMANCE
OF COMMUNITY POLICIN G
By
Amanda L. Robinson
Social capital is used as a theoretical framework to reveal the importance of
networks of relationships between officers and their supervisors for performing
community policing. Police social capital refers to the quality of officers’ relationships
within the police organization; for example, with their peers and their supervisors. It is
expected that officers with higher levels of social capital will be able to accomplish more
community policing than their peers who have lesser amounts of this resource, controlling
for officer characteristics and features of their work environment. Using data firom the
Project on Policing Neighborhoods (POPN), two measures of community policing were
developed: acts of community policing provided to citizens (comfort, referrals, and
information) per citizen encountered, and time spent on community policing activities
(attending community meetings, problem-solving, and crime prevention) per shift.
Separate models were tested on these two measures to determine the relative influence of
social capital (trust, cooperation, group cohesion, social support), officer characteristics
(sex, race, education, tenure, assignment, training) and work environment (department,
shift, beat problems, organizational support of community policing) on officer
performance of community policing. Interaction models were also tested to determine the
extent to which social capital interacts with characteristics of officers and features of their

work environments.

Results from Negative Binomial and Zero-Inflated Negative Binomial regression
models did not support the central hypothesis of this research: social capital was not a
significant predictor of either measure of community policing. Instead, work
environment characteristics tended to offer more consistent explanations of community
policing performance. Specifically, community policing varied significantly according to
the department in which the officer worked, whether officers were assigned to be
community policing specialists, and their levels of tenure. Implications of these findings
are discussed in terms of their relevance to organizational factors which promote or

hinder the implementation of community policing.

This dissertation is dedicated to the beloved memory of James Eldridge Robinson,

Minnie Lee Camp, and Jenny Lea Simpson.

iv

ACKNOWLEDGMENTS

An ‘acknowledgements’ page implies the ability of one mere (ex)graduate student
to remember and adequately address the multitude of personal and professional favors,
advice, support, social interaction, love and faith that it takes from scads of family
members, friends, peers, and mentors over many, many months for one Ph.D. to be
completed. Given that disclaimer, I would like to mention a few people at Michigan
State University that made my six years there especially interesting. My major advisor,
Peter Kirby Manning, is a first-rate example of the positive and profound impact that a
supervisor can have on a student. The unwavering concern and empathy he has shown
for me and other students over the years speaks volumes about his character. Meghan
Sarah Stroshine there over many beers and many years as a friend, co-author, proof-reader
and editor extraordinnaire, and co-creator of thousands of confusing and usually
incomplete SPSS data files. Long live Chanaman Industries! Many others at the school
(or now elsewhere) deserve thanks for their continuous encouragement, energy, and good
will over the years — Merry Morash, Christopher Damian Maxwell, Kevin Ford, Stephen
Mastrofski, Christina DeJong, Kevin Gray, Sean and Tracy Varano, Sameer Hinduja, and
Roni Mayzer. Finally I would like to extend my gratitude to those in my immediate
family — Daddy D., Shirl, Mamasita, Pablo, Yebes, Matt #1 and of course the Eagle —

whose courage and curiosity help me keep the academic life in greater perspective.

TABLE OF CONTENTS

LIST OF TABLES ...................................................... ix
LIST OF FIGURES ..................................................... x
INTRODUCTION ....................................................... 1
CHAPTER 1
REVIEW OF THE SOCIAL CAPITAL LITERATURE .......................... 5
The Sociological Significance of Social Capital .......................... 5
Historical Background ........................................ 5
Bourdieu’s Theoretical Framework .............................. 6
Existing Literature on Social Capital and Policing ........................ 9
Definition and Dimensions of Social Capital ........................... 11
Previous Measurement of Social Capital ............................... 12
Level of Trust .............................................. l3
Cooperative Exchanges ...................................... 14
Group Cohesion ............................................ 15
Social Support ............................................. 16
Police Social Capital .............................................. 17
The Impact of Social Capital ........................................ 19
Positive Outcomes .......................................... 19
Negative Outcomes ......................................... 22
CHAPTER 2
THE PHILOSOPHY AND PRACTICE OF COMMUNITY POLICING ............ 25
Community Policing as a New Police Mandate .......................... 25
Community Policing Activities ...................................... 27
Community Engagement ..................................... 28
Problem-Solving ........................................... 30
Providing Assistance to Citizens ............................... 31
Marginalization of Community Policing Within the Police Subculture ....... 33
CHAPTER 3
FACTORS AFFECTING PERFORMANCE OF COMMUNITY POLICING ........ 39
Police'Social Capital .............................................. 39
Features of the Officer’s Work Environment ........................... 40
Department ................................................ 40
Indianapolis, Indiana .................................. 41
St. Petersburg, Florida ................................. 42

vi

Beat Characteristics ......................................... 43

Shift and Assignment ........................................ 44
Characteristics of the Officer ........................................ 45
Sex ...................................................... 45
Race ..................................................... 45
Education ................................................. 46
Tenure ................................................... 46
Training .................................................. 47
Moderated Causal Relationships ..................................... 50
Sex and Social Capital ....................................... 52
Race and Social Capital ...................................... 53
Education and Social Capital .................................. 55
Tenure and Social Capital .................................... 55
Work Environment and Social Capital .......................... 56
CHAPTER 4
DATA AND MEASUREMENT OF VARIABLES ............................ S9
The Project on Policing Neighborhoods ............................... 59
Description of Data Collection ...................................... 59
Systematic Social Observation ................................. 59
Structured Interviews of Officers ............................... 61
Sample ......................................................... 61
Measurement of Dependent Variables ................................. 63
Providing Acts of Community Policing to Citizens ................ 72
Time Spent Engaged in Community Policing ..................... 73
Description of Dependent Variables .................................. 74
Measurement of Independent Variables ................................ 77
Officer Characteristics ....................................... 78
Work Environment .......................................... 81
Social Capital Dimensions .................................... 84
Description of Independent Variables ................................. 89
CHAPTER 5
ANALYTIC MODELS AND METHODS ................................... 91
Analytic Models .................................................. 91
Additive Models: Assessing Direct Effects Among Variables ........ 91
Interactive Models: Testing Moderated Causal Relationships ......... 93
Analytic Methods ................................................. 98
Regression Models for Count Data ............................. 99
Negative Binomial Regression Model .................... 100
Zero-Inflated Negative Binomial Regression Model ......... 102

vii

CHAPTER 6

RESULTS OF ANALYSES ............................................. 104
Bivariate Analyses ............................................... 104
Multivariate Analyses ............................................ 109

Additive Models ........................................... 109
Interactive Models ......................................... 111
Summary of Findings ............................................. 117

CHAPTER 7

ISSUES RAISED BY THE RESEARCH FINDINGS ......................... 120
Substantive Issues ............................................... 120

Research Questions Revisited ................................ 120
Leadership ............................................... 125
Organizational Structure .................................... 128
Geographic Responsibility ............................. 128
Decentralization ..................................... 129
Methodological Issues ............................................ 130
Interaction Terms .......................................... 130
Regression Equations ....................................... 132
Causal Order ............................................. 133
Sampling Strategy ......................................... 134
Measurement of Community Policing .......................... 135
Measurement of Social Capital ............................... 136

APPENDIX A

CORRELATION MATRD( OF ALL VARIABLES ........................... 138

REFERENCES ....................................................... 141

viii

Table 1

Table 2

Table 3

Table 4

Table 5

Table 6

Table 7

Table 8

Table 9

Table 10

Table 11

Table 12

Table 13

Table 14

LIST OF TABLES

Summary of Expected Direct Relationships. ...................... 49
Measurement and Descriptive Statistics for Dependent Variables. ..... 75

Measurement and Descriptive Statistics for
Officer Characteristics. ................................... 79-80

Measurement and Descriptive Statistics for
Work Environment. ...................................... 82-83

Measurement and Descriptive Statistics for Social Capital
Dimensions ........................................... 84-85

Principal Component Factor Analyses of Social Capital

Dimensions. ............................................... 88
Measurement and Descriptive Statistics for Interaction Terms. ....... 96
Bivariate Correlates of Community Policing Variables. ............ 105

Bivariate Correlates of Community Policing Variables -
Interaction Terms. ......................................... 108

Additive Models for Community Policing Variables. .............. 110

Interaction Models - Zero-Inflated Negative Binomial
Regression on Community Policing Time. .................. 1 13-114

Interaction Models - Negative Binomial Regression on

Community Policing Acts. ............................... 115-1 16
Summary of Findings. ...................................... 117
Correlation Matrix of All Variables ........................ 138-140

ix

LIST OF FIGURES

Figure 1 Conceptual Model. .......................................... 51

INTRODUCTION

Social capital is expected to provide an important new perspective on the social
organization of policing within the new era of community policing. Like employees in
other work organizations, police employees rely on work relationships for information,
access to opportunities, and support to increase the likelihood of productivity. In this
research, the term “social capital” is used to refer to the quality of officers’ relationships
with their peers and their supervisors. Dimensions that are important to these
relationships include the level of trust, the frequency of cooperative exchanges, the level
of group cohesion, and the amount of social support. Police officers’ work relationships
are considered to be a resource (if social capital is high) or a barrier (if social capital is
low or not present) affecting how often officers will perform community policing
activities. In short, it is expected that police officers are able to draw upon their social
capital in order to “get things done.” Community policing activities may be especially
dependent on police social capital, as this new policing movement is substantially
marginalized within the traditional police subculture. Investigating the relationship
between social capital and the likelihood that officers will engage in community-oriented
activities can provide us with both a broader and deeper understanding of police behavior
in the community policing era.

The application of the concept of social capital (a sociological term) to the field of
criminology has resulted in research that primarily focuses on the social capital of

communities, and how this is an important resource resulting from strong police-

community partnerships (a major tenet of the community policing philosophy). For
example, Greene (1998) refers to the importance of measuring the “changes in the
frequency, duration, and quality of police and community interactions” and “public
service networks created through such efforts” (p. 150). Referring to the quality and
quantity of police-community relationships can be considered an implicit reference to
social capital. Although various efforts made by the police and/or community actors to
enhance social control are dependent in part upon levels of social capital, as of yet no one
has examined the levels of social capital among police officers. If we do not know the
distribution of social capital among police officers, and the barriers preventing and
resources promoting its utilization, then our methods of encouraging strong police-
community partnerships will remain limited. This could have dire consequences for the
success of community policing initiatives, which have in recent years been embraced by
the public, police administrators, many police scholars, and the federal government.
Criminal justice research on police behavior would benefit from the theoretical
perspective of social capital. Limited research exists that incorporates a theoretical
perspective which can help us understand and predict officer engagement in many police
activities, particularly community policing. This issue becomes even more salient as the
police are being evaluated by more audiences in more different ways than ever before. As
expectations of police performance expand, so should our knowledge of the theoretical
underpinnings that guide their behavior. Only then will we be able to understand why
certain officers perform in certain ways, suggest how to facilitate better outcomes from

officers engaged in community policing activities, and describe the types of relationships

IQ

that most effectively increase officers’ stocks of social capital.

Several questions are answered by the current study:

(1) Is social capital related to how police officers perform their jobs?
Specifically, what is the relationship between levels of social capital and officers’
engagement in community policing activities?

(2) What is the relative contribution of police social capital in a model that
also includes characteristics of the individual officer and their work environment?

(3) Do officer characteristics, such as their sex, race, education, or tenure
moderate the relationship between social capital and community policing?

(4) Do features of officers’ work environment, such as their department and
their perceptions of the department’s support for community policing, moderate the
relationship between social capital and community policing?

Although there is an extensive body of sociological literature on social capital,
there is substantial room for improvement. Many researchers have not specifically
defined what they mean by the concept “social capital,” explained their measurements of
this construct, or been able to simultaneously incorporate all of the variables that may
influence the social capital-outcome relationship they are investigating. Using limited
measures of social capital also has precluded a full understanding of this complex
construct. Additionally, the majority of the studies in this area have been qualitative in
nature, usually involving in-depth accounts with small samples or a case study approach.
While informative and particularly important in laying the theoretical groundwork, the

social capital literature suffers from a lack of research on sample sizes large enough to be

quantitatively tested. This study could make a valuable theoretical contribution by
empirically testing a model of police social capital, helping us specify which relationships
are important to the theory and which are not. Furthermore, in his monograph on the
“state of the state” of social capital research, the noted social capital researcher Portes
(1998) concluded that, “the greatest theoretical promise of social capital lies at the
individual level” (p. 21). This study moves social capital research forward by specifically _
defining the term, providing multiple measures of the different dimensions of social
capital, testing a model which includes all of the variables that could potentially affect the
police social capital — community policing relationship, and providing this information at
the officer-level.

A final reason why this research is important is that only a handfiil of police
scholars has linked the social capital idea to policing, and they have only done so in an
indirect or tangential way. This concept is rich enough to be the primary focus in a study
of police behavior using detailed observational and survey data. Only then will we be
able to assess the true value of employing a social capital framework to the study of

policing.

CHAPTER 1

REVIEW OF THE SOCIAL CAPITAL LITERATURE

The Sociological Significance of Social Capital

Historical Backgound

The term that has come to the forefront of current sociological research, social
capital, was originally introduced by French sociologist Pierre Bourdieu in the late 1970's.
He made distinctions between four types of capital: (1) economic, (2) cultural, (3) social,
and (4) symbolic (Bourdieu, 1986). The first type corresponds with material goods or
wealth, and its relationship to the human condition has been under investigation since the
inception of sociology. The second type, cultural capital, refers to goods such as art,
language, or books which are proxies for “the long-lasting dispositions of the mind and
body” (Bourdieu, 1986, p. 243). Bourdieu used this term to understand the differences in
educational attainment of French children originating from different social classes, as
these outcomes were hypothesized to be a function of the cultural capital possessed by the
family. Specifically, when children’s cultural capital mirrors the dominant form of
cultural capital in society (i.e., upper—middle class), scholastic achievement is greater.
Thus, the generative nature of capital was revealed by Bourdieu, as the educational
system reproduces larger social structures in society that favor certain groups over others.

The third form of capital, social capital, was developed by Bourdieu to give a

name to the resource present within communities or groups that facilitates collective

action. Bourdieu defined social capital as “the aggregate of the actual or potential
resources which are linked to possession of a durable network of more or less
institutionalized relationships of mutual acquaintance and recognition” (p. 248).
Bourdieu’s definition of social capital contains two elements: (1) the network of social
relationships which a person can mobilize, and (2) the volume of capital (economic,
cultural, social, symbolic) possessed by the network members. Social capital, then,
represents the quality of social relationships that can be linked to various outcomes.

Bourdieu’s fourth form of capital is symbolic capital, which refers to all types of
capital once they are perceived as legitimate. In other words, ‘Vvhen the possession of
any kind of capital is justified not only in the eyes of those who benefit most from its
distribution, but also in the eyes of those who are most deprived of it” (Peillon, 1998, p.
218). Symbolic capital confers “the power to create the official version of the social
world” (Mahar, Harker, & Wilkes, 1990, p. 13). Symbolic capital may be equated with
legitimacy and prestige. An important commonality of Bourdieu’s four types of capital is
their conceptualization as inherently positive and productive; in his view, people strive to
increase their stocks of all four types of capital.
Bourdieu’s Theoretical Framework

To fully comprehend the importance of Bourdieu’s contribution to our
understanding of the social world, it is first necessary to put his forms of capital into a
broader context. Just as paper money only has value because it can be exchanged,
substituted, or transformed into tangible goods and services in society, concepts of capital

must also be recognized above and beyond their intrinsic value. Capital is just one

element in Bourdieu’s theoretical framework. He sought to make a contribution that
would provide a balance to the (often mutually exclusive) sociological traditions of
individualism and structuralism. He provided the concept of afield to identify areas of
social space where people struggle for position. Their struggle, and resulting position, is
affected by the distribution of various forms of capital.

Refening to an earlier example, the French educational system was considered by
Bourdieu to be afield, and the children in it actors whose struggle and outcome varied
according to their levels of cultural capital, which were dependent on those of their
families. Evident in this example is not only the force of social structure to pattern
actions, but also the potential of individuals to fight for position, to use their agency to the
best of their advantage. Bourdieu used the term habitus to explain the process by which
individuals are molded by social structure; it is a mediating construct between social
structure or fields and people struggling in them for capital. So what is habitus? It can
be thought of as a person’s world view -- the knowledge, beliefs, and dispositions that are
produced by socializing agents such as parents and also the social world in which they
exist. As Bourdieu (1990) explained, habitus consist of “schemes of perception, thought,
and action” (p. 54), or on a more basic level, “things to do or not to do, things to say or
not to say” (p. 53).

But all habitus are not created equal. As “the internalization of extemality,”
(Bourdieu, 1990, p. 55) we can see that some extemalities have more resources and
support compared to others, which affects the development of habitus. “Habitus is

intimately linked to capital in that some habitus (those of dominant social and cultural

factions) act as multipliers of various kinds of capital, and in fact constitute a form of
capital (symbolic) in and of themselves” (Mahar, Harker, & Wilkes, 1990, p. 12).
Habitus may be mental perceptions or attributes, but in the real world, in hierarchical
societies, these also translate into differences along lines of class, race, and gender.
Bourdieu conceptualized habitus as generative; that is, they tend to generate, promote, or
reproduce themselves. Returning to the school example, Bourdieu (1984) stated that
school “transforms social classifications into academic classifications, with every
appearance of neutrality” (p. 387) in part because it is patterned on the habitus of the
upper-class. In short, fields are not level, capital is not evenly distributed, and your
habitus counts for a lot.

The interaction between these various concepts provides a foundation for
understanding the behavior of individuals and groups in society. We are all embedded in
fields of struggle, seeking to acquire and exchange various forms of capital, constrained
by the limits of our habitus (or, conversely, propelled if our habitus is consistent with that
of the dominant group in society). The strategies we develop and incorporate into this
“game” of positioning and struggle in various fields are referred to by Bourdieu as
practice. Capital, as one feature of this game, is an important mechanism which
facilitates the style, content, and success of practice. As the title of Bourdieu’s (arguably)
most important book, The Logic of Practice (1990) indicates, these actions and strategies
are not random, but patterned. Because there is an underlying logic to our practice, we
can expect or predict certain behaviors or outcomes based on the knowledge we have

about the field under investigation and the distribution of capital among the players of the

game.
Existing Literature on Policing and Social Capital

Although the literature linking social capital to the study of policing is limited,
several scholars have joined these two fields, with promising results. One of the first to
do so was Manning (1994), who incorporated one of Bourdieu’s four forms of capital,
symbolic capital, into his theoretical discussion of the police response to domestic
violence. He provided a critical outlook on the exchange of capital in policing fields,
arguing that “police arrests in domestic conflicts are seen by some as enhancing symbolic
capital by ‘empowering women’... [but] for other observers, it creates yet another
intrusion of the state into private relations, an enforcement of class-biased notions about
disputing, and a source of exacerbation of conflict and increased costs to lower—class
domestic units” (p. 86). Because the police reflect and reproduce the habitus of the
upper-classes, their symbolic capital is used to reinforce “the patriarchal order and class-
biased character of policing” (Manning, 1994, p. 89). Consequently, he contended that
police work tends to reduce the capital of the lower-classes.

Lyons (1999) used the concept of social capital in his study of the Seattle police
department to illustrate his point that the contemporary community policing movement is
not achieving its intended aims. He provides the following explanation of the
relationship between social capital and community policing, “the most basic reciprocal
exchange at the heart of stories about community policing is a police/state commitment to
perform their duties in a way that enhances the generation of social capital in

communities and a community commitment to invest a portion of that capital in

cooperative efforts with the police to improve public safety” (p. 28). Lyons (1999) is
critical of community policing because its proponents operate under the assumption that
these new police strategies are supposed to help communities reclaim lost social capital
and use it to improve local, informal social control, yet this assumption is routinely
violated. Instead, the police mandate is broadened as the fight continues “to control
political, economic, and social resources for the power to say what policing is and who
communities are” (Lyons, 1999, p. 4), usually at the expense of the very communities the
community policing movement was originally intended to benefit.

Duffee et al. (1999) also touch upon the importance of community social capital
for the success of community policing initiatives. As they stated, “without sufficient
social capital... policing initiatives to prevent crime in such areas are particularly
problematic — ofien engendering no citizen involvement at all or increasing, rather than
reducing, dissension within the neighborhood” (p. 94). Although Henig (1982) found
that poor perception of police services was related to declining local organization, Duffee
et al. (1999) maintain that the plight of such neighborhoods needs to be understood in
terms of a larger “urban struggle,” where the police are just one institution that should
play a role in “constituency building” (p. 94).

Overall, this literature tends to focus on the social capital of communities, and
how it pertains to their relationships with the police. The one exception is a recent study
by Pino (2001)-who specifically examined social capital and community policing in a
small department in Iowa. Employing a qualitative methodology, he examined

interactions among and between the police, the citizenry, and neighborhood groups. He

10

found that there was a substantial lack of trust between the public and the police, as well
as between community policing and regular patrol officers, and that this had a detrimental
effect on efforts at co-production of safety. This study, while restricted in generalizability
due to the methodology and sample, points to the importance of understanding police
relationships and their subsequent impact on the success or failure of community policing
initiatives.
Definition and Dimensions of Social Capital

In the sociology literature, social capital refers to relationships among individuals,
networks of relationships, and people’s “ability to mobilize a wide range of personal
social contacts” (Newton, 1997, p. 577) to accomplish a particular objective. Coleman
(1988) was one of the first to apply this concept to sociology in America. He extended
Bourdieu’s merger of two streams of thought: (1) a sociological focus on the norms, rules
and obligations which socialize people and subsequently govern their behavior, allowing
for action to be explained in a social context, and (2) an economic focus on the
independent and rational goals of people which they subscribe to purely out of self-
interest. As Bourdieu explicated in his idea of habitus, neither of these views adequately
explains the social world, for people are not just blank slates scribed upon by society, nor
are their actions completely independent of the social context in which they occur. By
“introducing social structure into the rational action paradigm,” Coleman (1988, p. S95)
made a valuable application of social capital to contemporary sociological thought in the
United States.

While research on social capital has suffered due to ambiguous definitions and

11

poor operationalization of this construct, several themes emerge in the literature that I
refer to as dimensions of social capital. These include the level of trust, the frequency of
cooperative exchanges, the level of group cohesion, and the amount of social support
present in relationships. That is, researchers have either used all or part of these
dimensions to explain the formulation and/or utilization of social capital in various
settings. The dimensions relevant to this study are discussed below.1 Particular attention
is paid to how these dimensions have been measured in past research, with implications
for how these dimensions are measured in the current study (discussed in detail in
Chapter 4).
Previous Measurement of Social Capital

The empirical research on social capital often includes measures of the number of
relationships as a proxy for social capital (Bursick, 1999; Burt, 1997; Coleman, 1988;
Frank & Yasumoto, 1998; F urstenberg & Hughes, 1995; Granovetter, 1973; McCarthy &
Hagan, 1995; Molinas, 1998; Robinson & Morash, 2000; Teachman, Paasch, & Carver,
1997; Wellrnan & Wortley, 1990). Because numbers alone tell us nothing about the
quality of the relationship or the potential of relationships to be a social resource for those

in the relationship, other research (less often empirical) has described social capital not

 

1

Civic engagement, for example, has often been referred to as an important component of social capital in
community—level research. This has been measured as the proportion of people volunteering in various
religious, social service, or community-based associations (see Greeley, 1997; Portney & Berry, 1997;
Furstenberg & Hughes, 1995; Brehm & Rahn, 1997). While this dimension is important for macro-level
research on social capital (or research on social capital at the individual-level in a different context), civic
engagement is not an important dimension in my study of police social capital and its affect on officer
performance of corrrrnunity policing activities. It would be important to address, however, if the focus of
the current study was on community social capital and its impact on community participation in corrrrnunity
policing initiatives.

12

only in terms of the number of social relationships, but also in terms of the qualities
present in these relationships which may enhance or constrain the potential of those
relationships to be a resource. The literature has identified level of trust, cooperative
exchanges, group cohesion, and social support as important qualities to assess when
studying social capital. In other words, it is assumed that pe0ple who have relationships
that are high in these qualities have more social capital than people whose relationships
do not possess these qualities.
Level of Trust

People’s level of trust, whether it is in a generalized form, in each other, in a
particular group, or in a government, has been an important dimension of the social
capital construct. Fukuyama (1995), for example, puts trust as central to his definition of
social capital: “social capital is a capability that arises from the prevalence of trust in a
society or certain parts of it” (p. 26), as does Molinas ( 1998): “social capital is defined
here as the level of trust and community networking” (p. 413). Research at the micro-
level has also found trust to be a central issue in how people create and maintain their
levels of social capital. Specifically, the norms and values of individuals or, “those
cultural values and attitudes that predispose citizens to cooperate, trust, understand, and
empathize with each other” (Newton, 1997, p. 576) have been studied in relation to social
capital by many researchers. Coleman (1988) alludes to these as ‘social norrns’; for
example, the “norm that one should forgo self-interest and act in the interests of the
collectivity” (p. S104). Foley and Edwards (1997) mention the “attributes of individuals

which favor their civic engagement” (p. 551). Woolcock (1998) refers to “information,

13

trust, and norms of reciprocity” (p. 153) that inhere within relationships. Portes (1998)
describes the internalized norms of trust and reciprocity that are necessary in the
formation of social capital; similarly, Schrnid (1999) mentions “an internalized sense of
obligation and ethical norm” (p. 3).

Despite the wealth of literature on this subject, past measurement of this
dimension is lacking for two reasons. For example, some researchers have used the
General Social Survey’s questions on generalized trust as proxies for social capital, when
these questions are not context-specific. For example, these questions ask respondents,
“Do you think most people would try to take advantage of you if they got a chance, or
would they try to be fair?” and “Generally speaking, would you say that most people can
be trusted or that you can’t be too careful in dealing with people?” It would be more
fruitful to ask whether a person trusts a specific person, place, or thing. Second, the
social capital construct is more than “just trust.” Improved measurement of this construct
would include many dimensions of social capital.

Cooperative Exchanges

Social capital researchers ofien refer to “norms of reciprocity,” which when
present in social relationships increase the potential of those relationships to be a
resource. The logic underlying this dimension of social capital is that this type of norm
makes people give back in exchange for taking. After an exchange occurs (whether it is
money, material goods, information, or emotional aid like support or advice), it is
understood by both parties that the exchange will be paid back at a later date. This is a

form of trust in itself; trust in the belief that cooperation is beneficial and that exchanges

14

will be reciprocated.

Past researchers have tapped into “norms of reciprocity,” or what I term
cooperative exchanges, by looking at patterns of giving and receiving in a community
(Hofferth & Iceland, 1998), or analyzing actions one person in a relationship took that
helped the other person maintain or acquire certain resources (Frank & Yasumoto, 1998).
Others have tapped into this dimension by asking respondents questions such as, “How
often do you and people in your neighborhood do favors for each other?” and “When a
neighbor is not at home how often do you and other neighbors watch over their
property?” (Sampson, Morenoff, & Earls, 1999). The cooperative exchanges dimension
could also be measured with questions tapping into how often people share particular
goods, or the level of cooperation within a particular group, such as a family,
neighborhood, work group, or community agency.

Group Cohesion

Because social capital research is often done at the community-level, researchers
have been interested in what makes groups cohesive. It is assumed that cohesive groups,
or groups that have members who are supportive or trustworthy of each other, who share
norms, and/or have similar beliefs, will have more social capital. Measurement of this
dimension can be as basic as the proportion of residents in a particular neighborhood that
are friends or acquaintances, the frequency that a group engages in social activities, or the
amount of people in a group that simply like each other (Sampson, 1991; Sampson,
Morenoff, & Earls, 1999; Bursick, 1999). Social ties that have emotional density, for

example, with a high level of mutual confiding and intimacy, also increase social capital

15

(see Granovetter, 1973). Norms about particular behavior also influence the performance
of that behavior; for example, Coleman (1988) found that whether mothers expected their
children to attend college affected whether they actually did. Cohesiveness has also been
measured by questions assessing similarity among group members. Bursick (1999), for
example, asked people whether they agree with the statements “I have a lot in common
with people in my neighborhood,” and “The people in my neighborhood are a lot like
me.” The underlying assumption of this dimension is that groups that “get along” and
share similar beliefs and characteristics will have more social capital than those that are
antagonistic or whose members share very different beliefs or values.
Social Support

This dimension of social capital has been closely tied to the actions of people in a
social relationship that help one member accomplish a particular goal. For example,
F urstenberg and Hughes (1995) examined the support given and received in a mother-
child dyad, and found it related to the child’s successful school outcomes. Other
researchers have investigated different types of social support in relationships, such as
financial, emotional, and providing services, and found that the type of support is often a
function of the type of relationship (e. g., whether the relationship is between friends,
family members, neighbors, etc.) (Wellrnan & Wortley, 1990). In short, this dimension is
usually measured in a particular context, such as the family, workplace, or community.
When social support is high, positive outcomes are more likely, and when it is low, these

outcomes are more difficult for the actors to obtain.

16

 

16

M.

Upo

SUP}

Police Social Capital

In the present study, I am specifically concerned with police relationships with
their peers and supervisors, and how these impact officers’ performance of community
policing activities. Patrol work is considered to occur “in the context of territorially
based work groups” (Klinger, 1997, p. 283). This territoriality affects officers’ attitudes
and actions, depending on which work group they occupy in the department (Reuss-Iarmi,
1983) (e. g., community policing, patrol, SWAT, or special gang-suppression or street-
crimes units). Assigning officers to work together in a geographically defined area
creates the opportunity for informal group norms to arise; they are considered to be far
more important in governing officer behavior “on the streets” than are departmental
regulations or police management (Klinger, 1997; Reuss-Ianni, 1983). Examples of these
group norms include “watching out” for one’s partners and the rest of the officers
working on the same shift and beat, not “sucking up” to supervisors, and other variations
of solidarity and loyalty (Crank, 1998; Reuss-Ianni, 1983). Group norms about
appropriate levels of activity are also important modifiers of police behavior because they
tend to “discourage innovation while they encourage the status quo” (Manning & Van
Maanen, 1978, p. 267; also see Rubinstein, 1973). Past research has substantiated the
importance of work groups in policing. An officer who does not conform to informal
norms about what constitutes “real” police work and how to accomplish these tasks, for
example, may not be fully included in his or her work group. Officers who cannot draw
upon relationships that are rich in terms of trust, cooperative exchanges, and social

support and/or who are not members of cohesive groups (i.e., they are excluded or

17

colic;
1116 V;

peer 1

relati
pmg

18%]

Mg;

marginalized from these groups), will have additional obstacles to overcome than officers
who benefit from work relationships that are rich in social capital.

Miller’s (1999) in-depth study of Neighborhood Policing Officers (NPOs)
provides some insight into the importance of police-peer relationships in the community
policing era. Specifically, she found that NPOs who assertively established relationships
with beat officers experienced “greater understanding and cooperation from their
colleagues” (p. 109). Although the community policing movement has drawn attention to
the value of police relationships, we have not specifically examined the role that police
peer relationships play in performance, and if and how it varies according to what
policing tasks are being performed.

Quality relationships with supervisors also occupy an important place in police
work. Social capital theory identifies people with decision-making authority, such as
supervisors, as “targets” who may be especially important contributors to one’s stock of
social capital (Wood, 1997, p. 599). Officers rely on supervisors for information,
support, and evaluations of their performance (Van Maanen, 1983). Positive
relationships between officers and supervisors are so vital to efficient police work that
programs specifically designed to increase positive interaction between the ranks have
recently been suggested (Beck & Wilson, 1997). It is also important to remember that
supervisor support is considered vital to the success of innovative community-oriented
police activities (Geller & Swanger, 1995; Goldstein, 1990; Skogan & Hartnett, 1997).
Without supervisor support the implementation, as well as instrumental success, of these

programs is considered unlikely.

18

Miller (1999) documents how supervisor support allowed NPOs to overcome
much of the stigma associated with performing community policing tasks (considered by
many officers to be “social work” or “women’s work”; i.e., not real police work).
Indeed, in the department studied by Miller (1999), many upper-level management
positions were held by former NPOs; this had a legitirnizing effect on the entire
community policing program. Police social capital, then, may be more important to
officers who are deemed to occupy marginalized roles within the police organization.
Specifying these relationships becomes especially salient given the implications for
performing community-oriented policing tasks.

The Impact of Social Capital

Research usually tends to link social capital to positive outcomes; however, the
term is much more encompassing and flexible than this value judgment would imply.
Accordingly, recent research has been critical of inherently benevolent views of social
capital (e.g., that originally formulated by Bourdieu). As Foley and Edwards (1997)
noted, “its uses may range from asocial to antisocial to broadly prosocial” (p. 552). This
section summarizes the findings from research framing social capital in a positive light,
followed by research that looks at its “dark side.”

Epsjtive Outcomes

Much of the social capital literature refers to inherently positive yet ambiguous
outcomes (i.e., civic virtue, quality of life, etc.), but empirical support for these benefits
boils down to two variables: education and crime. Specifically, social capital has been

shown to increase positive educational outcomes (Coleman, 1988; Teachman, Paasch, &

19

 

 

 

 

 

 

Carter. 1
ample
flow fro:
than
educatic
disadtar
p035°55é
it is irrel
amount.
concept
you’ll. '
enhancs
00mm

Socioec.

has ten
Kfil'acl
Sampsc
3008] c
COmmu
Char 3C1:

C(”men

Carver, 1997; F urstenberg & Hughes, 1995). Teachman, Paasch, and Carver (1997), for
example, viewed social capital as a “filter through which human and financial capital
flow from the parents and the community to the child” to produce improved educational
achievement. Similar to Bourdieu, Coleman (1988) used social capital to explain the
educational achievement of children growing up in economically and socially
disadvantaged communities. As he stated, “if the human capital [skills or knowledge]
possessed by parents is not complemented by social capital embodied in family relations,
it is irrelevant to the child’s educational growth that the parent has a great deal, or small
amount, of human capital” (p. 8110). Furstenberg and Hughes (1995) also used the
concept of social capital to explain the successful outcomes of a sample of disadvantaged
youth. They defined social capital as a “resource upon which individuals may draw to
enhance their opportunities” (p. 581). Their research suggested that family-based and
community-based social capital played an important role in helping youth overcome
socioeconomic disadvantage.

Most of the research employing a social capital framework to the study of crime
has revealed a negative relationship: as one increases the other decreases (Bursick, 1999;
Kawachi, Kennedy, & Wilkinson, 1999; Kennedy et al., 1998; Sampson, 1995, 1997).
Sarnpson’s (1997) work on juvenile delinquency, for example, led him to conceptualize
social capital as a “buffer” against the negative effects of high levels of delinquency in the
community. As he stated, “in a system involving parents and children, communities
characterized by an extensive set of obligations, expectations, and social networks

connecting the adults are better able to facilitate the control and supervision of children”

20

(1997, p. 52). Positive relationships among community members are key, and can be
partly facilitated by: (1) organizing supervised leisure-time for youths, (2) observing and
reducing street-comer congregation, and (3) establishing mentor relationships between
adults and youth, and also building adult acquaintances within communities (Sampson,
1995, p. 210). Thus, to build social capital in neighborhoods and communities is to build
barriers against crime and violence.

Other researchers have also posited a beneficial relationship between social
capital and crime rates. Using national state-level data, Kennedy et a1. (1998) found
support for their hypothesis that low levels of social capital were related to firearm
homicide and violent crime. They measured social capital using indicators from the
General Social Survey (GSS) such as levels of trust and civic engagement among
community members. Their results supported a path model whereby income inequality
(relative deprivation) decreased stocks of social capital, which in turn increased crime
rates. Later work conducted by the same authors (Kawachi, Kennedy, & Wilkinson,
1999) tested similar conceptual fi'amework using state-level ecological data. In addition
to GSS trust variables, they also included a single-mother household variable as an
indicator of social capital. This measurement of social capital was negatively related to
state levels of violent crimes as well as property crimes. The authors took this as
evidence supporting Sampson’s work on social disorganization theory; specifically,
“crime is... a mirror of the quality of the social environment” (Kawachi, Kennedy, &

Wilkinson, 1999, p. 719).

21

Negative Outcomes

The majority of the literature either explicitly or implicitly refers to social capital
as a positive feature of social life, but other research offers several examples of the
negative consequences of social capital (Portes, 1998; Portes & Landolt, 1996). First, the
same strong social relationships that are necessary for the formation of social capital
within a group may also serve to exclude new members. When new members are
excluded, so are new sources of social capital, and the resulting isolation may result in the
group’s downfall. In his analysis of neighborhood security, Hope (1998) pointed out the
positive (e. g., membership, natural surveillance, etc.) as well as negative (e. g., exclusion,
stagnation, etc.) effects of having a closed community structure. Waldinger’s (1995)
analysis of social capital in the New York City construction business, for example, led
him to conclude that, “social structures such as closure or network multiplexity may
generate social capital for insiders while also excluding outsiders from the resources that
social capital generates” (p. 560). Referring to Granovetter’s (1973) seminal work, these
communities lack the benefits which accrue from weak ties.

Second, a highly cohesive group (i.e., one with a dense social network) may
prevent the success of its members. This may result from suict demands for conformity
which restrict group members’ personal freedoms and subsequently encourage them to
leave the group. For example, tight-knit communities such as the Amish, while
benefitting from the social capital that rich networks of relationships produce, also suffer
from the exodus of young people from their community. Wilson’s (1987) work on the

plight of America’s inner-cities also provides evidence of the detrimental impact that may

22

occur to communities that experience a departure of people, skills, and resources.

Third, group solidarity may foster an “us versus them” mentality which in the
long-term discourages successful initiatives and dissolves group cohesion. Referred to as
“downward leveling norms,” these norms may take the place of “mainstream” norms,
especially in communities traditionally marginalized, stigmatized, or victimized by
society at large (Portes, 1998, p. 17). For example, “‘wannabes’ -- the latest lexical
contribution of inner-city youth to mainstream culture -- are those who imitate the ways
and lifestyles of the majority in search of success. Often, these efforts only meet scorn
from fellow members of their community, who see them as a threat to solidarity and their
own sense of self-respect” (Portes & Landolt, 1996, p. 21).

Few researchers have provided empirical evidence revealing the “dark side” of
social capital. One exception is the research conducted by McCarthy and Hagan (1995),
who incorporated notions of embeddedness (Granovetter, 1973), social capital (Coleman,
1988), and differential association (Sutherland, 1942) to explain onset of criminal
activity. Specifically, they proposed that “embeddedness in networks of deviant
associations provides access to tutelage relationships that facilitate the acquisition of
criminal skills and attitudes, assets we call ‘criminal capital’” (p. 63). Analyzing rates of
drug-selling, theft, and prostitution among a sample of homeless youth, they found
evidence supporting their hypothesis; the acquisition of criminal capital led to detrimental
consequences because the number and length of relationships increased criminal activity.
Sirrrilar to the relationship between social capital and entrance into the mainstream world

of work, criminal capital was related to entry into the underground world of criminal

23

work.

Although limited research exists on the negative consequences of social capital, it
is still apparent that to equate the concept with only positive outcomes would be
inaccurate as well as unnecessarily limiting. Conversely, taking a value-neutral stance
with regard to social capital allows the researcher to investigate all the potential outcomes
with which it may be associated. As it is doubtful that the outcomes discussed above are
the only ones related to levels of social capital, it would seem constructive to broaden the
research agenda. Policing as a public good can be considered an important outcome of
social capital that requires investigation. How levels of police social capital affect the
distribution of these “goods,” especially in a community policing context, is the focus of
the current research. The next chapter examines the literature on community policing, the
activities encouraged by this policing philosophy, and its place within the police

subculture.

24

CHAPTER 2

THE PHILOSOPHY AND PRACTICE OF COMMUNITY POLICIN G

Community Policing as a New Police Mandate

Common wisdom points to three events which precipitated the policing reform
movement known as “community policing.” First, civil unrest during the 19605
challenged police legitimacy and brought questionable police practices into the national
spotlight. Second, recognition of the isolation of the police from the public led to interest
in citizens being “co-producers” of police services. That is, police and community
members should share responsibility for crime reduction and work together toward
meaningful, long—lasting change. Third, the community policing movement arose out of
the ashes of research findings that constituted a “systematic demolition” of the
assumptions underlying the professionalism movement (Walker, 1984). As research
indicated that “nothing works,” reformists attempted to identify and adopt policing
strategies that might make a difference (e. g., foot patrol, permanent beat assignments,
mini-stations, etc).

The philosophy of community oriented policing is currently widespread and
embraced by many citizens, police administrators, scholars, and local and federal
politicians. For example, the 1994 Crime Act authorized $8.8 billion for community
policing programs, with the result that almost 90% of Americans have community

policing officers working in their communities (U .S. Department of Justice, 1999). The

25

under 15102 ll

 

noxidnals.
atomplist.
variation in
community
cities. rese.
programs 1
programs 1'
attempted
elements c
35 com u

Ht
(20111311111;

POllClllg 1

Policing l

 

 

 

underlying premise of the community policing philosophy is that the police and various
individuals, agencies, organizations, or community groups should work together to
accomplish mutual goals. At the implementation level, however, there can be substantial
variation in the tactics and strategies deployed by police departments as part of a
community policing program. For example, in a study of community policing in eight
cities, researchers concluded that there was more difference than similarity among the
programs (Grinc, 1994; Sadd & Grinc, 2000). Some of these community policing
programs focused on aggressive street enforcement and drug crackdowns while others
attempted community organizing and interagency cooperation. There are several
elements or categories of activities that are commonly recognized, and widely accepted,
as community policing that are discussed in the next section.

However, given the wide variety of police activities subsumed under the
community policing “umbrella,” it may be helpful to first document what community
policing is not (see Trojanowicz, Kappeler, & Gaines, 2001, p. 18-27). Community
policing is not a technique, it is a philosophy. Community policing is not public
relations, it is a substantive change in the police-public relationship. Community policing
is not soft on crime, it is “smart” on crime. Community policing is not flamboyant, it
achieves results through steady, long-term efforts. Community policing is not
paternalistic, it must empower officers and citizens in order to achieve results.
Community policing is not an independent entity within the department, it is a philosophy
that must inundate the entire department. Community policing is not cosmetic, it requires

that the department make substantial changes in how it deals with the community.

26

Community policing is not just another name for social work, it recognizes the fact that
the majority of police work involves non-crime related duties. Community policing is not
elitist, and special efforts need to be made to counteract hostility from general patrol
officers who might hold this belief. Community policing is not designed to favor the rich
and powerful, it is an egalitarian view that promotes providing assistance and support to
citizens of all jurisdictions. Community policing is not a panacea, it will not fix all
problems but will be more effective at addressing problems than traditional policing.
Community policing is not “safe,” but officers have to be trusted enough to take risks and
make mistakes. The next section describes police activities that are generally considered
to reflect a community policing philosophy, and the categories of activities
operationalized as community policing in the current research.
Community Policing Activities

Reflecting the diversity and reality of police work, Trojanowicz and Bucqueroux
(1992) identified 18 duties inherent in the role of officers engaged in community policing:
law enforcement, directed patrol, community involvement, identifying and prioritizing
problems, reporting, problem-solving, organizing, communicating, conflict resolution,
referrals, visiting, recruiting and supervising volunteers, proactive projects, targeting
special groups, targeting disorder, networking with the private sector, networking with
non-profit agencies, and administrative/professional duties. The multitude of activities
considered a reflection of the community policing philosophy can be grouped into three
general categories: (1) police engagement of the community in the production of order,

(2) a proactive response by police to community problems, for example using a problem-

27

solving strategy, and (3) use of the broadened police role to more frequently provide
general assistance to citizens. These community policing activities are discussed in the
sub-sections below, although it must be recognized that in practice these categories would
not necessarily be mutually exclusive.

One example from California demonstrates the overlap between different
elements of community policing: problem-solving and providing assistance to citizens. A
police sergeant established a new domestic violence protocol for his department after
analyzing crime data that revealed that the domestic incidents tended to escalate over
time, as well as comprising a significant proportion of calls for service. The new protocol
mandated that officers make personal contact with the victim within one week of the
incident, and again after one month. The purpose of these follow-up visits was to provide
victims with general assistance and information, and to provide referrals to appropriate
agencies in the community. To assess whether the problem was being solved, additional
crime data were analyzed. Results showed that calls for service decreased 57% from
1996 to 1997 in the “hot spot” domestic viOlence locations identified by the department
(Sampson & Scott, 1999).

Commqu Engagement

This theme of the community policing philosophy emphasizes an expanded police
presence in communities in order to facilitate community capacity to exercise social
control. As Rosenbaum (1998) stated, “the challenge for police today and into the 21St
century is to find creative ways to help communities help themselves” (p. 14). In other

words, police are no longer simply expected to enforce the law but to provide a broad

28

 

array of services aimed at increasing safety and order within communities. The
underlying premise guiding this expansion of the police role is that the police cannot
solve community problems without the help of citizens and corrrrnunity agencies.
Community policing advocates propose that the police and the public ought to become
“co-producers” of public safety, each contributing to the maintenance of law and order,
because “together, police and public are more effective and more humane co-producers of
safety and public order than are the police alone” (Skolnick & Bayley, 1988, p. 1). The
police must, therefore, engage the community in order to build a productive, meaningful,
working partnership. For example, community policing officers could attend meetings
with various community groups and associations to open channels of dialogue, ideally
leading to the identification of community problems and the creation of strategies for
their solution.

In Chicago, “building bridges between police and community members” was vital
for the success of community policing (Skogan & Hartnett, 1997, p. 110). Beat meetings
were how the department was able to convey to the community that the new policing
philosophy was a long-term strategy intended to stay. Unlike some community policing
initiatives in other departments, in Chicago the beat meetings were held regularly, at
various locations, and were attended by the officers patrolled the beats. These meetings
were an opportunity for community members to raise concerns to police officers, for the
participants to identify and work together to solve problems, and to exchange
information. As the authors conclude, “people can participate only where there are

opportunities to do so” (Skogan & Hartnett, 1997, p. 160). Community engagement is

29

thus a particularly important aspect of any community policing project.
Problem-Solving

One part of the community policing philosophy that concerns improved policing
is that police should not only respond in a reactive mode to crime and disorder, but
should also work in a proactive way to address these issues. Problem solving has been
recognized as a central characteristic of community policing departments because it uses
community input to identify crime problems and determine the appropriate strategies to
address them. To put it bluntly, “community policing without problem solving is not
community policing” (J olin & Moose, 1997, p. 291). As opposed to the traditional
strategy of random or preventive patrol, whereby police hope to decrease crime and
disorder by their mere presence, problem-solving is a strategy police use to fight specific
crimes with specific plans (Goldstein, 1990).

Eck and Spehnan (1987) developed the widely accepted and used SARA model of
problem—solving, which identifies four stages of the problem-solving process: (1)
scanning to collecting information to identify a crime problem, (2) analysis to determine
the nature and extent of the problem, (3) response through the creation of a specific
strategy to address the problem, and (4) assessment to determine whether the response
alleviated the problem. For example, officers engaged in problem-solving would attempt
to prevent the occurrence or recurrence of particular problems, and develop plans or
projects that go beyond merely responding to a particular call in order to address the
underlying cause of the problem. In contrast to traditional, reactive policing, the focus of

police effort within a problem-solving model is on the underlying condition — when that

30

is addressed then it is likely that calls for service will decrease to a significant extent.
One frequently cited study documenting the success of a problem-oriented
approach took place in Newport News, Virginia (Eck & Spelrnan, 1987). Police were
inundated with burglary incidents originating from a particular apartment complex.
Officers surveyed residents about this particular crime problem but also learned that the
physical state of the complex was of major concern to residents. Information gleaned
from other city agencies (e. g., the fire department, public works, etc.) confirmed that
physical deterioration was a serious issue and directly contributed to the burglary problem
(i.e., aged window and door frames that were rotting made break-ins easy to commit).
Having a clearer picture of the underlying condition helped police create a long-term
strategy for decreasing burglary incidents. Officers worked with the apartment manager
and city agencies to improve the physical state of the buildings. A neighborhood
association was formed with help from police that was able to successfully lobby for

continued upkeep of the complex. Due to these efforts, the burglary rate decreased by
35%.

Providing Assistance to Citizens

The community policing reform emphasizes a broad, social role for the police,
with the goal of police becoming more responsive to citizen concerns. Also referred to as
“personal service,” and following the trend in the private sector of putting “customers
first” or “listening to customers” (Skogan, 1998, p. 162), this philosophical dimension
aims to build trust and positive interactions between the police and the community they

serve (Cordner, 1998). No longer are police to be viewed solely as gatekeepers to the

31

criminal justice system; they are being called upon to monitor the turnstile to social
service and government agencies as well. Some goals of providing citizens with
assistance, information, and support include: alleviating citizen fear about particular
problems in the community; garnering citizen support for police initiatives to solve
problems; educating citizens about their vulnerability to crime; and helping citizens solve
problems for themselves (Goldstein, 1990).

Guided by a community policing philosophy, the police serve as instigators and
motivators for cooperation between agencies with the goal of creating networks of
services that benefit citizens. Community policing officers, therefore, are encouraged to
provide citizens with needed assistance and information, including referrals to other
community agencies that might be better suited to handle the citizen’s problem. Some
departments have implemented “swaps” where agency workers and police officers spend
time in each other’s work environment to better learn how to assist citizens (Goldstein,
1990). In addition to the usual gun and radio, community policing officers might also be
dispatched to calls armed with lists, contact information, and descriptions of services
provided by local community agencies.

Miller’s (1999) ethnography on community policing officers in one police
department suggests that they are often able to provide citizens, and abused women in
particular, access to more types of assistance compared to when they solely respond to
calls for immediate help. Compared to their traditional 911-driven counterparts, “[t]he
nature of the neighborhood position encouraged [community policing] officers to become

actively involved with the community they served: in prevention programs, case follow-

32

ups, working on continuing problems, and acting as liaisons with residents, businesses,
city services, and the criminal justice system” (p. 183). Consequently, community
policing officers often had more information about the citizens they served, which they
could then use to provide referrals, informally monitor problem citizens, and provide
reassurance, advice and support to citizens, and victims especially.
Marginalization of Community Policing
Within the Police Subculture

While the community policing philosophy has been warmly embraced by the
public, many police scholars and administrators, and the political establishment, it often
has encountered resistance from the rank-and-file officers who must ultimately translate
this complex and multifaceted concept into concrete policing strategies and tactics.
Several reasons exist for this less-than-enthusiastic acceptance by officers are often cited
in the community policing literature. First, it-has often been unclear to officers what is
meant by the term “community policing.” Officers may be able to understand and even
admire the concept in the abstract, but difficulty arises when they have to actually put this
philosophy into practice (Goldstein, 1990; Kelling & Moore, 1988; Sadd & Grinc, 2000).
A corrrmon refrain has been “what exactly do you want us to do?” and until recently,
police supervisors, scholars, and administrators were unable to adequately answer this
question.

Second, officers often have been expected to “do” community policing in addition
to other traditional police duties such as patrolling and responding to calls for service. A

lack of resources (particularly training) has often made the burden of this new reform

33

movement fall primarily on the shoulders of those least-equipped to handle it (Williams
& Sloan, 1990; Zhao, Thurman, & Lovrich, 2000). Lurigio and Skogan (2000), for
example, found that patrol officers were significantly less confident with their ability to
engage in community-oriented policing than their higher-ranking counterparts, although
patrol officers are the police employees expected to actually engage in community
policing. The above two issues facilitate patrol officer resistance to community policing
initiatives, which in turn facilitates the marginalization of community policing within the
police subculture.

Even when resources and training are devoted to community policing initiatives,
resistance from officers is still encountered. For example, in a research design involving
officer surveys before and after the implementation of a department-wide community
policing strategy in Chicago, the conclusion reached was that “the bulk of the officers in
the field had not yet ‘got the message’ or committed themselves to the program in a
significant way” (Skogan & Hartnett, 1997, p. 105), despite a significant level of
department-wide training. The first round of training involved an initial orientation and a
3-day skill building session of patrol officers. Supervisors were then given a 4-day
training curriculum comprising nine community policing modules, and were responsible
for conducting the second round of training during roll—call training sessions which
covered the nine community policing topics. This training made it clear that community
policing was to be a “real” program rather than a “paper” program and was in the
department to stay.

A subculture is commonly defined as the attitudes, norms, and beliefs systems

34

adopted by employees to make sense of their work environment. The police subculture
has been identified as the single largest barrier facing those who want to implement
community policing (Sparrow, Moore, & Kennedy, 1990). To be successful, researchers
suggest that community policing initiatives “be compatible with the existing culture and
organizational climate in a department and with the basic concerns and needs of police
personnel” (Lurigio & Skogan, 2000, p. 255). Unfortunately, the components of the
traditional police subculture and the community policing philosophy are often at odds
with each other. The police subculture rests on themes of uncertainty, danger, violence,
suspicion, and coercive authority — often leading to an “us versus them” mentality
regarding police relationships with citizens, and increasing the likelihood that officers
adopt work group norms of loyalty and solidarity (see Bittner, 1970; Crank, 1998;
Manning, 1997; Skolnick, 1997; Westley, 1970). These norms often put them at odds
with management and other “outsiders” and make officers likely to resist change and
protect the status quo.

Officers derive honor and status (and reduce uncertainty and degradation) from
their official mandate which is to enforce the law. Departmental selection, training,
reward, and promotion systems also reinforce the supremacy of law enforcement within
the police subculture. The end result is that law enforcement is viewed as the only “real”
police work. The police subculture also reinforces traditional notions of masculinity,
with “real” police work being “men’s wor ”; all other policing activities are viewed as
not “real” police work, and therefore the responsibility of female officers or social

workers.

35

 

scppor
police

produ

e cre
doing
office

C0111:

The community policing philosophy directly challenges many of the norms and
values underlying the traditional police subculture. For example, community policing
supports cooperation and trust between police and citizens, in contrast to the traditional
police view of the citizenry as a hostile enemy to be distrusted instead of engaged in a
productive partnership. Within a community policing context, officers are encouraged to
be creative and solve problems rather than to just “lay low” and “cover their ass” by
doing the bare minimum to avoid potentially negative attention from the public, other
officers, and their supervisors (see Van Maanen, 1978). Others have noted that the
community policing philosophy represents the “feminization” of police work, by valuing
stereotypical female qualities such as communication, cooperation, and supportive
interpersonal relationships (Miller, 1998). The community policing reform movement,
therefore, poses many threats to the officers whose cultural values have rested on themes
of masculinity, danger, suspicion, and violence.

To handle this new threat to the traditional police identity, there is evidence to
suggest that officers have marginalized community policing (and the officers who
practice it) within the police subculture. Wesiburd and McElroy (1988) found that when
given the choice, community policing officers in New York continued to choose policing
strategies that had a traditionally high status within the police subculture, such as
aggressive law enforcement. The officers who practice community policing are often
derided by general patrol officers as not doing “real” police work (Pate & Shtull, 1994);
the traditional police tactics of patrol, surveillance and arrest. Moore (1992) epitomized

the view of community policing officers within the police subculture with his observation

36

that “they became known as ‘grin and wave’ squads and ‘rubber gun’ squads” by the
regular patrol officers (p. 135). Marginalization may also arise from resentment related to
practical issues such as staffing and resources. Regular patrol officers may perceive
community policing officers to be “wasted resources” insofar as their assignments
involve a lowered expectation to engage in traditional policing tasks; the “slack” from
CPOs falls on the shoulders of patrol officers. One patrol officer reflected this sentiment
when he expressed that community policing officers should be “arrested for theft when
they pick up their paychecks” (Pino, 2001 , p. 209). The pronounced lack of trust and
respect noted in many departments has, not surprisingly, translated into negative work
experiences for community policing officers. Winfree and Newbold (1999), for example,
found that community policing officers perceived less supervisor support than regular
patrol officers.

In their national assessment of community policing implementation, Zhao,
Thurman, and Lovrich (2000) identified internal organizational impediments as the most
significant obstacle to successful implementation (more significant, for example, than
community impediments such as citizen resistance or a lack of local government support).
The items comprising the “organizational impediments” factor included: resistance from
middle-management, line-officer resistance, departmental confusion about what
community policing is, problems in line-level accountability, officer’s belief that
community policing is “soft” on crime, a lack of community policing training, and union
resistance. The two items that had the highest factor loadings were middle-management

and line-officer resistance. The authors concluded by suggesting that any long-lasting

37

organizational change toward community policing must correspond with a change in the
values and norms underlying the police subculture. While there is some evidence to
suggest that the traditional police subculture might be changing (see Haarr, 1997; Paoline,
Myers & Worden, 2000), it still unfortunately presents a major challenge to community
policing initiatives. The next chapter discusses three categories of factors hypothesized to
influence the likelihood that officers will engage in community policing activities: (1)

police social capital, (2) work environment, and (3) officer characteristics.

38

CHAPTER 3
FACTORS AFFECTING OFFICER

PERFORMANCE OF COMMUNITY POLICING

Police Social Capital

Past literature has revealed the important of police work groups on police
behavior, and there is no reason to expect these relationships to be less important in the
community policing era. Research has documented that officers marginalized or
excluded from their peer group (e. g., because they are of a minority race or are women)
have suffered a lack of acceptance, a denial of needed information, sponsorship, and
promotion opportunities (Buzawa, 1981; Ellison & Genz, 1983; Holdaway & Barron,
1997; Martin, 1980; Milutinovich, 1977). These issues can subsequently affect their
work experiences, performance, and advancement within the police organization.
Although not previously or explicitly stated as such, what makes certain officers
marginalized is their lack of social capital. Officers who lack social capital in their work
environments face higher hurdles and bigger barriers to getting the job done than their
counterparts who are embedded in productive, supportive, and trustworthy work
relationships. It is expected that officers who have relationships with peers and
supervisors that are rich in social capital will be more productive than officers without
this resource, who may not have the same level of opportunity or support to engage in

various community policing activities. Community policing officers, therefore, might

39

particularly need relationships that are strong in terms of trust, c00perative exchanges,
group cohesion, and social support to accomplish a type of policing not wholeheartedly
accepted within the police subculture.

While it is hypothesized in the current study that the social capital dimensions will
be positively related to the amount of time an officer spends engaged in community
policing, the social capital literature suggests that negative outcomes may also result. If
officers who have high levels of social capital are found to be significantly less likely to
spend time on community policing activities, this could be interpreted as an example of
the “dark side” of social capital. For example, officers rich in this resource might be
better able to circumvent departmental dictates supportive of community policing. In this
case, the support, cooperation, trust and group cohesion officers have in their work units
and/or with their supervisors could be used to cover up poor community-policing
performance or Shirk community-oriented activities, or to further other (possibly
negative) policing outcomes not included in this study. Despite this possibility, the
central hypothesis of the current study is that as levels of social capital increase, so will
the likelihood that officers engage in community policing activities.

Features of the Officer’s Work Environment
Department

The available evidence on the two departments included in this study suggests that
their work environments might differ in important respects relevant to community
policing, such as their interpretation of this policing philosophy. For example, one

department (Indianapolis) takes a “broken window” aggressive order maintenance

40

6“

approach, with the police chief emphasizing traditional’ law enforcement activity”
(Mastrofski et al., 2000, p. 317) while the other department (St. Petersburg) emphasizes
building positive police-citizen partnerships (Paoline, Myers, & Worden, 2000).
Furthermore, a greater proportion of officers in St. Petersburg are assigned as community
policing specialists (22% compared to 5% in Indianapolis) (Mastrofski et al., 2000). This
departmental difference could impact the frequency with which officers engage in
community policing activities, resulting in St. Petersburg officers performing more
community policing activities. Additional information about the two departments is
provided below.

Indianapolis, Indiana. The jurisdiction of the Indianapolis Police Department
(IPD) is referred to as the Police Services District, 3 portion of Indianapolis-Marion
County for which the department is responsible. At the time of the POPN study, the IPD
served a population of more than 377,000 people. The UCR Index Crime was 100 per
1,000 residents and 37 per officer (Parks et al., 1999). In the years 1996-1997, the
department employed about 1,000 full-time sworn officers, about half of which were
assigned to patrol. The sworn force was 83% male and 79% white (Parks et al., 1999).

The geographic responsibility for the Indianapolis Police Department was divided
among four patrol districts: North, West, East, and South. Within each district, officers
were assigned to one of five shifts: Day (5:00 am to 2:00 pm), Day Tact (9:00 am to 5:00
pm), Middle (1 :00 pm to 9:00 pm), Late Tact (7:00 pm to 3:00 am), and Late (10:00 pm
to 6 am). These shifts were staggered so that shifts overlapped when service needs were

high. Officers’ and supervisors’ work schedules were determined by their assignment to

41

one of three work schedules with rotating days off (referred to as A, B, or C “letter
days”). A work squad consisted of the officers and supervisors assigned to the same
district, shift, and letter day.

The department implemented community policing in 1992. The Deputy Chiefs of
each patrol district had considerable autonomy and latitude in determining the day-to-day
operation of their districts. This resulted in wide variation in the organization and
practice of community policing across districts. For example, in the West district the
community policing strategy took an aggressive order maintenance approach.
Conversely, the focus of community policing in the North district tended toward
“community building” (Mastrofski, Worden, & Snipes, 1995) than aggressive order
maintenance; for example, officers were encouraged to positively interact with
community members. Community policing in the East district was practiced by a special
unit and had a problem-solving focus. Finally, no community policing was practiced in
the South district.

St. Petersburg, Florida. Just over 240,000 residents inhabited St. Petersburg at the
time of the POPN study. While St. Petersburg has a smaller population, its UCR Index
Crime Rate is similar to that of Indianapolis, with 99 per 1,000 residents (Parks et al.,
1999). The violent crime rate in St. Petersburg was more than three times the national
average: about 2,250 violent crimes per 100,00 residents compared to 716 (Bureau of
Justice Statistics, 1995). The St. Petersburg Police Department (SPD) employed about
500 full-time sworn officers, and similar to the IPD, about half (n=283) were assigned to
patrol. The majority of officers were white (78%) and male (87%) (Parks et al., 1999).

Officers working in St. Petersburg are deployed in four shifts: Day (7 am to 3

42

pm), Evening (3 pm to 11 pm or 4 pm to 12 am), 4th Relief (7 pm to 3 am), and Midnight
(11 pm to 7 am or 12 am to 8 am). The department’s geographic responsibility is divided
into three districts (North, South, and West). The SPD implemented community policing
in 1990. Each of the three districts is responsible for a “zone” of the department’s 48
Community Policing Areas (CPAs). CPAs are analogous to the concept of a patrol beat;
it is the smallest unit of geographic responsibility. A zone consisted of three sectors, with
each sector representing a conglomeration of CPAs. At the time of this study, the SPD
had 63 community policing officers (CPOs), over twice that of the IPD even though the
department is half the size. In St. Petersburg, the permanent, geographic deployment of
officers who focused on community-building with neighborhood organizations (Parks et
al., 1999) resulted in a more uniform organization and practice of community policing
compared to Indianapolis. The available evidence regarding the organizational contexts
of the two departments suggests, therefore, that [PD officers will probably engage in less
community policing compared to SPD officers. It is also reasonable to suggest that
officers’ perceptions of how supportive their department is of community policing efforts
will also affect their proclivity to engage in community policing.
Beat Characteristics

Regarding the primacy of territorial knowledge, Rubinstein (1973) stated that an
officer “combines his knowledge of local behavior with his conceptions of how the public
streets are used to analyze and perform many of his routine obligations” (p. 151). An
officer’s assigned beat has been found to impact his or her level and type of activity
(Klinger, 1997; Smith, 1986). The conclusion by some scholars that community policing

tends to work the least where it is needed the most (i.e., in poor, crime ravaged, socially

43

disorganized and minority communities) also points to the profound impact that
community or beat characteristics may have on whether community policing goals are
accomplished (Skogan & Hartnett, 1997; Walker, 1999; Williams & Murphy, 1990).
Officers who work in beats that have a significant amount of major crime problems (such
as drug dealing, theft and burglary, or vandalism) might have less time to engage in
community policing activities than their counterparts working in less troubled areas. As
such, it is important to include officers’ perceptions of beat problems in a model
predicting community policing performance.
Shift and Assigment

Recent research has investigated performance differentials between community
policing officers and 911-responders. Although the study conducted by Mastrofski et a1.
(1995) did find a difference in arrest rates, only 1 of the 17 variables examined differed to
a statistically significant degree between the groups. Robinson and Chandek (2000a)
failed to find a significant difference between community policing and traditional units
when handling domestic violence calls. Recently, however, DeJong, Mastrofski, and
Parks (2001) found that community policing officers spent more time engaged in
problem-solving activities than did officers assigned to general patrol. Because the
dependent variable in this study is officer engagement in community policing activities, it
is important to include officer assignment (community policing versus general patrol
officer) as a control variable, since theoretically community policing officers might be
expected and given the resources to accomplish more community policing activities.

Similarly, officers working the day shift would be expected to have more Opportunity for

44

 

community policing activities because most citizens (and citizen groups) are awake and
functioning during this time.

Characteristics of the Officer

While most research finds very little difference in the performance of male and
female officers, performance differences might emerge when we start to measure non-
traditional policing activities, such as those guided by a community policing philosophy.
For example, DeJong (2000) found that female officers are more likely to provide
comfort to citizens than their male counterparts, and Hale and Wyland (1999) report that
female officers may communicate better and subsequently de-escalate potentially violent
situations. Although the evidence is limited, it is reasonable to believe that female
officers might more frequently engage in community policing activities.

Ric;

Research suggests that an officer’s race is not an important variable to consider
when measuring performance with traditional indicators such as making arrests or using
excessive or deadly force (see F yfe, 1981; Reiss, 1968). To conclude that minority
officers and white officers are identical, however, may be misleading. Mastrofski (1981)
found that black officers were more knowledgeable of local citizen organizations in black
neighborhoods. In Chicago, it was found that minority officers were significantly more
optimistic about community policing than their white counterparts (Lurigio & Skogan,
2000), and although we cannot assume that attitudes are always consistent with behavior,

it may be the case that racial differences emerge when we investigate non-traditional

45

 

 

police activities, such as corrrrnunity policing. It is therefore expected that nrinority
officers will engage in more community policing than their white counterparts.
Education

The relationship between levels of education and police performance is less
straightforward. While there is no evidence to suggest that college educated officers
behave differently on the street (Sherman, 1978), more recent research finds that
performance improves as education increases. For example, college educated officers
may receive fewer complaints compared to their less educated counterparts (Kappeler,
Carter, & Sapp, 1992). Researchers who followed a cohort of officers for ten years found
a positive relationship between college education and supervisor ratings of job lmowledge
(Truxillo, Bennett, & Collins, 1998). Kakar ( 1998) found that officers with some college
or a college degree reported performing better, and Palombo (1995) found that they were
more professional. The relationship between officer education and officer performance
warrants further investigation, but it is likely that as education increases so would the
skill and ability necessary for officers to engage in community policing activities.
Mute

Most research tends to find that as years of experience increase, the amount of
arrest activity decreases (Binner, 1990; Muir, 1977; Stalans & Firm, 1995). Roberg,
Crank and Kuykendall (2000) also report that younger officers tend to work harder and be
more productive than older officers. The effect of tenure on community-oriented
performance indicators has only recently been studied. DeJong (2000), for example,

found that tenure improved the likelihood that female officers would provide comfort to

46

citizens. Conversely, more experienced officers were found to spend less time on
problem-solving than their less experienced counterparts (DeJong, Mastrofski, & Parks,
2001). The available evidence, therefore, provides a conflicting account of the
relationship between tenure and community policing. However, the present study
assumes a relationship that has been supported by the majority of research, that tenure
will decrease activity, in this case community policing.

1%

Officers who have received more training on how to perform community policing
activities might be expected to spend more time engaged in these activities, due to an
increase in ability (and perhaps confidence) in how to perform community policing.
Although DeJong, Mastrofski, and Parks (2001) did not find community policing training
to significantly increase the amount of time an officer spends problem-solving, others
contend that training is the key to successful implementation of community policing
(Glensor & Peak, 2000; Zhao, Thurman, & Lovrich, 2000). It is therefore expected that
as the amount of training an officer has received on community policing increases, so will
the likelihood that he or she will perform community policing activities.

Table 1 presents a summary of the direct relationships tested in the current study.
The focus of this research is on the link between social capital and community policing.
It is expected that all four dimensions of social capital (trust, cooperation, group
cohesion, and social support) will significantly increase the likelihood that officers
engage in community policing. Additionally, because officers are constrained by features

of their work environment, it is expected that officers who are members of the IPD, who

47

are not assigned to community policing, who work at night, and who perceive a high level
of problems in their assigned beat will engage in less community policing. Officer
characteristics are hypothesized to play a small role in explaining community policing
performance, with the exception that community policing training is expected to be an
important predictor of community policing performance. Overall, police social capital
and characteristics of the officer’s work environment are expected to exert the strongest

effects, while officer characteristics will exert a relatively weak influence on community

policing.

48

Table 1

Summgy of Expected Direct Relationships.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Variable expected to impact officer Direction of expected Magnitude of expected
performance of community relationship. relationship.
policing.

Social Capital

Level of Trust positive strong
Cooperative Exchanges positive strong
Group Cohesion positive strong
Social Support positive strong
Work Environment

Characteristics

Department (Indianapolis) negative strong
Day Shift positive strong
Community Policing Assignment positive strong
Beat Problems negative strong
Department Support of CP positive strong
Officer Characteristics

Sex (female) positive weak
Race (minority) positive weak
Education positive weak
Tenure negative weak
Community Policing Training positive strong

 

49

 

Moderated Causal Relationships

Research has documented relationships between social capital and the
characteristics of individuals; however, we are still far from drawing blanket conclusions
about these relationships. Because the extant research implies a strong possibility that
stocks of social capital will vary according to different individual and organizational
factors, included in the analytic plan are tests of moderated causal relationships, also
known as interaction terms. The relationship between social capital and community
policing is expected to vary depending on certain officer characteristics and features of
their work environment (i.e., officer characteristics and work environment variables will
moderate the relationship between social capital and community policing) (see Figure 1).
Past studies that examine how social capital varies according to different individual and

organizational characteristics are reviewed below.

50

 

Figure l

Concpptual Model.

 

 

OFFICER
CHARACTERISTICS

 

 

 

SOCIAL
CAPITAL

 

 

 

 

 

 

COMMUNITY
POLICIN G

 

 

 

WORK
ENVIRONMENT

 

 

51

 

Sex and Social Capital

A person’s sex has been shown to covary with his or her level of social capital,
but the evidence suggests that information about the context in which these ties are
located is vital, as men may have an advantage in workplace networks, whereas women
may excel in familial or community networks (Hofferth & Iceland, 1998; Molinas, 1998;
Moore, 1990; Rountree & Warner, 1999; Wellrnan & Wortley, 1990). Specific to the
study at hand, Haarr’s (1997) research in a police patrol bureau found that officers tended
to interact most frequently with their same race-gender group. Martin (1980) found that
female officers avoided many interactions with peers and supervisors as a result of these
interactions being misconstrued as involving a sexual component. Because females are
already marginalized both numerically and within the traditional police subculture,
having fewer interactions with officers from other units or shifts compounds their
disadvantage. If female officers are not incorporated into workplace networks to the
same degree as their male colleagues, then their stocks of social capital would be lower
since they are removed from relationships that could provide them with support,
information, cooperation, and access to opportunities (Martin, 1980). Although women
might have relationships of a higher quality (e. g., involving more trust, support, etc.), due
to their small numbers in policing (especially in positions where they hold power, such as
supervisors), this may not be sufficient for overcoming the likelihood that they will have
lower stocks of social capital than their male peers. Since police departments, like all
human organizations, incorporate societal notions and expectations related to gender,

male and female officers probably have stocks of social capital that differ to a significant

52

extent.
Race and Social Capital

It is not unreasonable to believe that in a society where race is related to many
important variables, such as crime, poverty, and health, that it would not also be related to
social capital. Sociological research investigating this relationship has revealed its
complex nature. For example, while some researchers have not found a significant
relationship between race and levels of social capital (Antonucci et al., 1998), others have
found that compared to whites, minority persons and communities tend to have less social
capital (Bursick, 1999; Edwards & Foley, 1997; Portney & Berry, 1997; Sampson, 1997).
Brehm and Rahn (1997) discovered a significantly negative relationship between race and
generalized trust, considered to be a component of social capital because it affects civic
engagement. Waldinger’s (1995) ethnographic research into the construction trade found
that ethnic enclaves could produce both positive and negative manifestations of social
capital. On the one hand, close racial/ethnic work groups fostered trade and cooperation
among minorities, but on the other hand these same relationships were detrimental for
minorities trying to acquire the skills and connections necessary for success in
predominantly white fields. In other words, minority contractors created their own
networks apart fi'om the ‘old boys’ network,’ which both helped and hurt them. A
notable exception to this trend of social capital racial differences is found in research
conducted by Portrrey and Berry (1997). They did not find significant race effect with
respect to social capital measures such as respondents’ sense of community or levels of

participation in neighborhood associations.

53

In the context of police work, however, the relationship between race and social
capital might be more straightforward. For example, research shows that minority
officers experience more isolation (Buzawa, 1981) and receive less encouragement
(Milutinovich, 1977) than their white counterparts. Minority officers may also face
“exclusion from informal channels of support and information” (Ellison & Genz, 1983),
which may lead to negative consequences in terms of promotions. For example, Carter
(1986) found evidence that Hispanic officers in one department believed that the
administration discriminated in hiring promotions. Officers in Haarr’s (1997) study also
believed that the department made hiring decisions based on race: white officers thought
they were biased in favor of minority officers, while rrrinority officers thought they were
biased in favor of white officers. Black and Asian officers working in Great Britain
identified many ways in which they were omitted from full participation by their co-
workers (Holdaway & Barron, 1997).

The majority of the evidence (although much of it is dated) suggests that minority
officers would have less social capital than their white counterparts. This prediction
might be inaccurate, however, in departments that have significant minority
representation (and at all ranks), or that have a history of cooperative and supportive
relations between officers of different races. While this is not the case in the two
departments involved in the current study, minority officers might invest more in
relationships in order to counteract their marginalized status. The proposed research

could clarify these issues.

54

Education and Social Capital

Social capital researchers have long been interested in the link between education
and social capital. Not surprisingly, the available evidence indicates that these variables
share a positive and mutually enforcing relationship. Social capital is a resource which
facilitates successful school outcomes (Bourdieu, 1984; Coleman, 1988; Furstenberg &
Hughes, 1995; Teachman, Paasch, & Carver, 1997) and in turn, the size of people’s social
networks tends to increase as they become more educated (Antonucci et al., 1998;
Edwards & Foley, 1997; Moore, 1990; Stanton-Salazar & Sanford, 1995; Robinson &
Morash, 2000). Additionally, F urstenberg and Hughes (1995) found that completion of
high school and enrollment in college were related to many social capital measures in a
positive direction. Brehm and Rahn (1997) found that education and civic participation
shared a strong positive relationship; civic participation is ofien considered a component
of social capital (Pumam, 1995).

In the context of policing, however, the available evidence suggests that a college
education may decrease social capital. Stevenson (1988) found that more educated
officers experienced higher levels of burnout and social isolation that their less educated
counterparts. It is important to explore the relationship between education and social
capital in the unique organizational context of policing. It might be expected that more
educated officers do not have higher levels of social capital if they work in organizational
environments that neither support nor reward educational achievement.

Tenure and Social Capital

The relationship between officer tenure and attitudes has been more thoroughly

55

investigated, beginning with Niederhoffer’s (1967) finding that cynicism increases with
time on the job. The least cynical officers are those with less than two years of
experience (Wilt & Bannon, 1976). In general, research indicates that other negative
attitudes also become more predominant with age. For example, as tenure increases so do
negative attitudes toward domestic violence victims (Robinson & Chandek, 2000b) and
community policing (Lewis, Rosenberg, & Lewis, 1999; cf. Lurigio & Skogan, 1994).
Job satisfaction also decreases with tenure (Hoath, Schneider & Starr, 1998). Relevant to
social capital, officers with more eXperience tend to hold more negative views about their
work relationships, are more cynical about the flow of information between superiors and
subordinates (Lewis, Rosenberg, & Sigler, 1999), and perceptions of supervisor support
decrease (Winfree & Newbold, 1999). Most of the available evidence suggests a negative
relationship between officer tenure and any positive outcome related to their work, and
the present research could help us determine whether this is also the case with social
capital.
WorkIEnvironment and Social Capital

There is also research to suggest that social capital might vary according to the
work environment in which the work is embedded. For example, in a study designed to
assess the ability of the Michigan Victim Assistance Academy (MVAA) to increase the
social capital of victim assistance providers, qualitative and quantitative analyses revealed
that participants who reported substantial resources in their workplace (e.g., support from
supervisors and co-workers, adequate staffing levels) were better able to utilize the social

capital gained from attending the MVAA (Robinson & Morash, 2000). Specifically,

56

participants in supportive workplace environments were more likely to expand and
improve their networks of relationships relevant to improving assistance to crime victims
to a greater extent compared to participants who faced barriers in their workplace (e. g.,
negative attitudes from supervisors or co-workers, lack of time, money, or staffing, or
organizational problems). In short, it appeared that some workplace environments helped
rather than hindered the utilization of workers’ social capital.

In a recent study examining community policing using a social capital framework,
it was also apparent that the organizational context mattered a great deal (Pino, 2001).
The implementation of community policing in “Small City” Iowa faced its biggest
challenge from the police department itself. In particular, the department was always
understaffed and underfunded, creating a situation where patrol officers were forced to
work a lot of overtime to achieve adequate patrol levels. This contributed to patrol
officers’ animosity and lack of trust toward the few community policing officers who
were hired under a federal grant. This lack of trust among police also generated a lack of
trust between police and the neighborhood groups with whom they were supposed to
create partnerships. Pino (2001) summarized the organizational effect, “in a climate of
insufficient resources, an add-on COP program, and a lack of trust among officers, COP
was doomed to not live up to its potential” (p. 209). Despite the existence of any social
capital among officers, it would appear that the negative organizational climate was in
effect a workplace barrier that could not be overcome.

Given the possibility that the organization can have an overwhelming influence on

not only levels of police social capital, but the utilization of police social capital toward

57

community policing goals, the current study investigates whether the relationship
between social capital and community policing is moderated by the department in which
the officer works. In addition, officers’ perception of their departments’ support of
community policing might also moderate the relationship between police social capital
and community policing. It is an important indication of the officer’s work environment

and will also be included in the interaction models.

58

CHAPTER 4

DATA AND MEASUREMENT OF VARIABLES

The Project on Policing Neighborhoods

This study involves secondary data analysis from the Project on Policing
Neighborhoods (POPN), a large-scale study of police behavior funded by the National
Institute of Justice. Data for the study were collected from the Indianapolis, Indiana and
St. Petersburg, Florida Police Departments. This study was conducted during the
summers of 1996 and 1997, respectively, and involved two primary sources of data
relevant to the current study: Systematic Social Observation (SSO) and structured
interviews of police officers. Each method of data collection is described below.

Description of Data Collection

Systematic Socipl ObservLiop

The primary feature of the POPN is the systematic observation of police officers
at both research sites. It is ideal to have a comprehensive set of measures when
investigating police behavior. Official data, citizen and officer surveys, and observational
data are all useful in this regard. Observational data, however, may be particularly useful
for assessing officer performance during police-citizen interactions — Opening up for
examination the “black box” of police performance (Wycoff, 1982), or the “process” of
policing (Mastrofski, 1996, 1999; Reiner, 1998). In short, observational data allow for a

more accurate description of the craft of policing. Data are collected first-hand, rather

59

 

than relying on second hand sources.

Fieldwork in Indianapolis and St. Petersburg was conducted during the summer
months. Trained observers accompanied patrol officers during their normally scheduled
shifts. During ride-alongs, observers took notes on the behavior of patrol officers, as
well as other officers (peers and supervisors) and the citizens with whom they interacted.
At the conclusion of these observational sessions, observers used their notes to provide
detailed narrative accounts of the rides. This information was then converted into coded
data using observation instruments designed specifically for the project.

The observation instruments consisted of four forms: ride form, activity form,
encounter form, citizen form. The observational data therefore contain four levels of
analysis. One ride form was completed for every ride-along, and included information
on the site, district, rank, and shift of the officer. The activity form was used for events
that were not classified as “encounters” with other police or citizens (i.e., these behaviors
were typically performed alone). This form included the type of activity in which the
officer was engaged, the length of the activity, the type of problem at which the activity
was directed, and whether the activity was part of a long-term plan or project.

The encounter form was used to code information about situations in which police
engaged in some form of verbal or physical contact with a member of the public.
Encounters were classified into three categories. Brief encounters involved contact with
the public that lasted less than one rrrinute and involved police business, such as an
officer telling someone to “move along.” In these encounters fewer than three exchanges

(verbal or gestures) between the police and the public occurred. Casual encounters

60

involved contact with the public, but no police business, such as an officer running a
personal errand that involved talking with a clerk. Full encounters were police-public
contacts that lasted longer than one minute and also involved police business. During
these encounters words and/or gestures were exchanged more than three times.
Encounters that lasted less than one minute but involved the threat of violence by either
police or citizens were also coded as full encounters. The encounter form was used to
code information such as the length of the encounter, other participants in the encounter
(i.e., officers, citizens, or both), the type of problem at which the encounter was directed,
and the type of decisions that were made during the encounter. Lastly, the citizen form
captured information such as the age, race, sex, income, and demeanor of all citizens
involved in the encounter.
Structured Interviews of Officers

Structured interviews were conducted with patrol officers, sergeants and
lieutenants in both sites by trained interviewers during the officer’s regular work shift.
The interviews were designed to capture information on a variety of t0pics, such as the
officer’s beliefs about proper police roles, goals, and priorities; the officers’ perceptions
of their work group and supervisor; and their attitudes toward community policing.
Demographic information (e. g., race, sex) and background characteristics (e.g.,
education, tenure) were also obtained.

Sample
The current study uses both data sources for the measurement of independent and

dependent variables. Trained observers collected and coded observational data during

61

361 ride-alongs in Indianapolis and 368 ride-alongs in St. Petersburg (totaling 729 rides).
Ride-alongs lasted the duration of an officer’s regular shift (8 hours in St. Petersburg and
8.5 hours in Indianapolis), resulting in more than 5,700 hours of field observation (Parks
et al., 1999). Some officers were observed during more than one ride-along, some just
during one ride-along, and others were not observed at all. A majority of officers in each
site participated in the structured interview, resulting in a total of 728 surveyed officers.
In Indianapolis, 93% of the 426 patrol officers were interviewed; in St. Petersburg 98%
of the department’s 246 patrol officers completed the interview (Paoline, Myers, &
Worden, 2000).

Observational and survey data were merged at the officer level to obtain a sample
of officers that had responses to all measures necessary for a test of the conceptual model
proposed in the current study. Dependent measures were derived from the observational
data and independent measures were obtained from the officer surveys. The sample of
officers who both completed the interview and were observed by the POPN include 176
officers from Indianapolis and 142 officers from St. Petersburg. The total sample to be
analyzed in the present study consists of 318 officers.

The reason for the reduction in sample size (fi'om 728 to 318 officers) is that
while most officers were interviewed, not all officers were observed during ride-alongs.
Some officers were observed multiple times, instead of each officer being observed at
least once, because the sampling plan was designed according to rides rather than
officers. The sampling plan was created to ensure that rides were conducted for (1) every
work shift for all beats in both sites, (2) all units working in all beats (3) during days of
the week that varied in busyness (Parks et al., 1999). Consequently, the POPN captured

62

multiple observations for some officers, but no observations of others. The 318 officers
in this sample were observed a minimum of once and a maximum of 10 rides, with the
average officer observed for approximately two rides. Almost half (44%) were observed
for one ride, while 25% were observed for two rides, and 31% were observed for three or
more rides. The ride-based sampling strategy does have implications for this sample:
about half of the officers were observed only once and about half received multiple
observations. Whether the amount of observation varies according to characteristics of
the officers, their work environment, their social capital, and their community policing
performance is discussed in the next section.

The sampling strategy developed for the POPN is not ideal for the purposes of the
present study. A more suitable sampling plan for this research would be based around
officers rather than rides: the focus would be on observing every type of officer rather
than every type of ride. An officer-based sampling plan would have increased the sample
size of officers suitable for study and also avoided any biases resulting from comparing
officers who have been observed for various lengths of time. It should be noted that
these biases were addressed by standardizing the observational data to account for
officers being observed for different lengths of time (see the next section for a detailed
discussion of the measurement of the dependent variables). Furthermore, given the
objectives of the present study, it is imperative to conduct the analyses at the officer level
because social capital is an attribute of people, not rides. While the decision to conduct
the analyses at the officer level might not be methodologically intuitive (given the ride-
level sampling strategy), it reflects the theoretical framework of the current study.

Measurement of Dependent Variables

63

The present study is concerned with identifying the factors that significantly
impact officer performance of community policing. Based on previous research that has
identified three categories of activities guided by a community policing philosophy
(community engagement, problem-solving, and providing assistance to citizens), the
dependent variables are operationalized using six activities that reflect these‘community
policing dimensions (more information is provided in Table 2, discussed in the next
section). The six measures of community policing include: providing comfort to citizens;
providing information to citizens; providing referrals to citizens; attending community
meetings; problem-solving activity; and crime prevention activity.

Two dependent variables that provide different indicators of officer productivity
were created from the six measures: (1) the number of community policing acts
performed, per citizen encountered by the officer during the data collection period, and
(2) the number of minutes the officer engaged in community policing activities, per 8-
hour shift2 worked by the officer during the data collection period. Creating two
dependent measures avoided the problem of summing indicators that were measured at
different levels of analysis. Specifically, three community policing indicators are
measured at the activity level, and three are measured at the encounter level.
Consequently, some of the community policing indicators are collected at the level where
it makes intuitive sense to count the number of citizens receiving the act, while the other

community policing indicators lend themselves to a measurement of the number of

 

2

St. Petersburg officers worked 8-hour shifts while Indianapolis officers worked 8.5-hour shifts. The
decision was made to standardize the time measure by eight hours because most police departments use
shifts of this duration. Indianapolis is the exception rather than the rule.

64

minutes the officer was engaged in the activity.

Community policing acts (providing comfort, information, and referrals to
citizens) were coded from the observational data gathered during the ride-alongs. The
citizen form was used to quantify information about police-citizen encounters. This
allowed for a count of how many citizens were provided comfort, information, and/or
referrals from each officer during the observational period. The ‘comfort’ indicator was
derived from the question, “During the encounter, did the police comfort or reassure. the
citizen?” This was a yes/no question where comfort was only counted when it was
preformed by the primary officer under observation, or the primary officer along with his
or her partner or other police at the scene. Provided below are excerpts from the
narratives that provide examples of how police provide comfort to citizens. The primary
officer under observation is designated 01 (these are the officers included in my sample),
while his or her partner is designated 02 and other police at the scene are designated 03,

04, etc. Citizens involved in the encounter are designated C1, C2, C3, etc.

> At a park where the marchers are dispersing, 02 is lecturing the children on their
bad behavior during the march. 02 punishes the children by saying that they will
not be taken for a treat after the march. The children are very upset and one in
particular appears to be crying. 01 walks up to C1, who is a black male about 10
years old, and lower class based on dirty clothing. 01 comforts C1 and tells him
to get in the van to get a ride home. C1 is very upset and his head hangs low. Cl
acknowledges Ol's request and heads into the van. 01 leaves the scene.

v 01 started walking to her patrol car when she spotted a black female walking up
to the emergency room doors. Ol asked the lady if she is the mother of the
accident victim fi'om encounter 21. Cl, black female about 40 years of age,
stated that she is the mother. C1 is middle class based on her attire. She is
wearing pants and a short sleeved shirt. C1 is neat in appearance. Ol explained
to Cl what occurred and she explained that her son was taken to another hospital

65

for treatment. C1 seemed very concerned for her son’s well being. 01 reassured
her that he was going to be all right. The communication between the two was
very friendly. 01 said that she could follow her over to the hospital and C1 stated
that she knows where the hospital is and that she will meet her over there.

The measure ‘providing inforrnation’ was taken from the question, “Did the
police provide this citizen information on how to deal with a problem on their own
initiative (without the citizen’s request)?” This was a yes/no question. The following

examples demonstrate police providing information in practice:

v 01 walked to the front door of the house which was located in a residential
neighborhood. The door was opened by C1, a lS-year-old black female. 01
asked C1 if she had called the police and C1 stated that her mother had called the
police. C1 asked 01 to come inside and said she would go get her mother.
Before C1 left the room 01 asked her if the dog sitting in the living room would
bite. C1 told 01 that her dog wouldn’t bite and 01 began to pet the dog
hesitantly. C1 returned with her mother C2, a black female approximately 34
years of age. C2 explained that she had a restraining order against her husband
and that he had been at her window harassing her. C2 stated that he left when she
said she was going to call the police. C2 also told 01 that there is a warrant for
her husband’s arrest. At this point 01 asked C2 for a description of her husband
and C2 told 01 his name and gave a brief physical description of him. C1 told
01 where her step-father usually stays and the type of car he drives. C2 left the
room to get her husband’s social security number... At this point C2 returned
with her husband’s social security number. 01 told C2 that he was going to do a
report and would keep his eye out for him during the night. 01 told C2 that if he
returned she should call the police. 01 told C2 to call 911 and then leave the
phone off the hook. 01 told C2 that they would get an emergency 911 run and
would get to her a lot faster if she did this. C1 and C2 thanked 01 and 01 wished

them a nice night.

> C1 (a black female, 41 years old, middle class based on neat appearance and
driving'a newer model Toyota passenger car; upset but respectful) drove up and
told 01 that she was the complainant. She said that her husband had called her
from a store down the road and that he was out of breath and sounded really
worried that someone was after him, and she was very worried about his welfare.
03 then departed to the store. She told 01 that her husband was out of breath
when he called because he had run from the car wash to the store to escape the

66

robber. She said she did not know why her husband insisted on washing his car
that late at night. [When] 03 returned to the car [he] was with C2 (a black male,
45 years old, middle class based on dress and driving a newer pickup; upset but
respectfirl), who was Cl's husband. C2 said he had just started to wash his truck
when a black male wearing a black cap and green shirt stepped out of some
bushes, pulled out a handgun and started to cock it. When C2 saw that, he
sprayed the black male with a car wash hose and then took off running and called
his wife. He said he probably could not identify the man again if he saw him. 01
advised C2 that there had been many robberies in this area and that C2 should
wait until daylight to wash his vehicle. 01 provided C2 a pamphlet on victims’
rights, and C1 and C2 then thanked 01 and left the area. 02 and 03 also
departed.

The ‘referral’ indicator was supplied by the question, “Did the police ask/tell the
citizen to seek the help of other service agencies to solve the problem?” The police could
suggest, request, try persuasion, try negotiation, or command the citizen to seek the help

of an agency. Examples of police providing citizens with referrals are provided below.

> As 01 was patrolling a residential neighborhood a black male waved at the
officer and asked him to stop. C1, a black male of approximately 45 years of age
was standing at the side of the road and walked over to the patrol car. C1 told 01
that he wanted some advice on a problem he was having. C1 told 01 that he had
his car painted several weeks ago and that he was not pleased with the job the
person had done. He said that when he returned the car to the individual they
refused to fix the paint job. At this point 01 explained that this was a civil matter
and that he might have a case in civil court. 01 gave C1 a brochure about small
claims court and asked C1 some questions. 01 asked Cl if he had a receipt and
C1 said that he didn’t have a receipt. C1 said that he had several witnesses
though. 01 told C1 that he might not have a very strong case if he didn’t have a
receipt or contract for the work but that the cost of the small claims court was
only $40. C1 thanked 01 for his advice and the encounter ended. The encounter
was not wimessed by any bystanders. Both C1 and 01 interacted in a business-
like manner.

> C1, a middle class white female in her late 205, came to the door of the given
address. The house was located in a small neighborhood with very nice houses
and yards. C1 was dressed in clean shorts and a tank top. She had two small
children with her, a boy aged 5 and a girl aged 2. She told 01 that her concern

67

was with children getting hurt in the dump site. 01 shook his head and said he
could understand that. Cl said her husband was outside videotaping the area. We
followed Cl through her yard then walked through some very tall weeds. As we
got through the weeds, it began to smell very strongly of cow manure. C1 told

01 that trucks had been dumping all these materials in this vacant lot and there
were some questionable materials in it. She said every time it rained the
contaminated water ran into their yard and probably down to the septic tank. She
said it was probably getting into the water. When we got past the weeds, there
were several large hills of dirt and broken up concrete. C2, a middle class white
male in his late 205, was standing on one of the hills videotaping a large hole
filled with very dirty water and building material. C2 came down and asked 01 if
there was anything they could do to stop the company from dumping. C1 said she
had already called the Health Department. 01 said that they wouldn’t listen to
him any more than they listened to her. He recommended that she call the Zoning
Department, building inspectors, and/or the Environmental Protection Agency.

01 said that the EPA might test the water if they (C1 and C2) took a sample of it
to them. 01 also suggested that they call the local news station. C2 said that they
had called one of them and they were supposed to be coming out soon to do a
story on it. 01 told them to keep pursuing the news station because once they
made a big story out of it, the agencies involved would have to respond... 01
suggested that C1, C2, and C3 send fliers to their neighbors and inform them of
the conditions nearby. He said that if the neighborhood banded together, they
would probably get better results. He also said that if nothing else worked, they
could file a class action lawsuit... He told me [observer] this situation was a good
example of community policing: giving advice to citizens about resources in the
community that they can go to when there really isn’t anything the police can do.

The second dependent variable was derived from activity-level data. Activities

are distinguished from encounters in that the former do not necessarily involve an

interaction with a citizen, while the latter do. Narrative information was quantified and

allowed for identification of activities where officers engaged in problem-solving, crime

prevention, or attending community meetings. The variable for ‘community meetings’

was obtained ftom the question, “Did this activity involve a meeting with representatives

of a citizen organization?” Citizen organizations could include neighborhood or other

area-based groups, victim advocate groups, business groups, church or religious groups,

school groups or other unspecified community groups. Additionally, the coding

68

instructions required that representatives of the organization had to be acting as members
on behalf of that organization for the activity to count as a community meeting.

Examples of officers spending time at community meetings are provided below:

> En route to a meeting about weekly park activities at the Leisure Services
department. When it came time for the meeting, the head of the department
directed everyone to the conference room. 01 sat through the meeting which was
a weekly thing to discuss the activities at a local park every Sunday. Those
attending were some members of Vista, a police supervisor in the department,
another CPO in charge of some of the park, a representative for activities in the
park and another city representative. The meeting was supposed to be to discuss
how the activities in the park went the past Sunday. Vista members complained
about how their “Father’s day in the park” activity there the past Sunday had been
cut shorter than they would have liked due to a lack of funding. One Vista
member complained about the marijuana smoking that “was allowed” to go on in
the park and wanted to lorow why this and public drinking were not being
punished. The police supervisor from the department, a black male
approximately 50 years old, claimed that those activities were illegal and were not
being allowed to go on by the police who worked the park. Others issues were
brought up during the meeting as well. 01 voiced some citizens’ concerns over
public urination and the need to have some portable toilets placed in the park.

> The meeting was already in progress. Everyone attending the meeting was seated
around a long wooden table. 01 took a seat on a couch located near one end of
the table. There were 16 people attending this meeting. The meeting was held by
the K Business Association. It consisted of business owners located on K Street.
They were discussing a proposal for the 1st Annual K Baseball Festival (to
celebrate the new baseball team). They were also discussing plans to convert a
large portion of K Street into a commercial district. The goal was to offset the
negative image given to K Street by the riots. Toward the end of the meeting, the
group addressed 01 and asked if she had any suggestions based on their goals.
01 stated that the most important thing is that if a problem arises, don’t wait for
the problem to get out of hand before calling the police. Contact the police during
the initial stages of the project. 01 made her services available to everyone in the
group and she handed out a few business cards to the group. The meeting
officially concluded.

The ‘problem-solving’ indicator was derived from the question, “Was this activity

69

part of a long-term plan or project to deal with a problem?” Long-term was defined as

longer than the ride being observed. Furthermore, the officer must have planned this

activity prior to the ride. Plans that were deve10ped spontaneously or during the ride

were not considered long-term. Plans could focus on specific people or locations, this

kind of problem or crime in general, or unspecified long-term plans. Below are examples

of problem-solving activities from the narratives:

D

01 arrives at the scene where the drug march will begin. It is primarily a low-
class neighborhood. Many of the houses are boarded-up. The residences are very
run-down, with garbage in the yards. Windows do not have blinds or drapes, but
instead are covered with old dirty blankets. 01 is told by 02 to follow the back
of the march in the patrol car. Many of the citizens are wearing yellow T-shirts
that say “up with hope, down with dope,” the main chant of the march. The
citizens walk down the street, some with megaphones, shouting this slogan as
well as several others. Residents of the neighborhood come out of their homes to
see what is going on. Some are pleased with the march, but others shout
profanities at the marchers, telling them to leave. The marchers stop at one place
[supposedly a known drug selling spot] and continue their chanting. They chant
over and over differing slogans, while the neighborhood citizens stand in their
front yards and watch. 01 gets out of her car and stands with the marchers
sometimes clapping and shouting the slogans.

C3 described how the Center attracts vagrants who congregate, beg, sleep, drink,
and urinate about his property and other businesses along this street. C3
explained he is losing customers. The pe0ple who have done business with him
in the past complain about the vagrants bothering them. He has spoken to some
of the other business owners on this street. He intends to write a letter to the
legislator who would be responsible for this district. Further, C3 would like
police to help him. 01 told the owner that he had been working on the problem
for the past year. He mentioned other business owners who had contacted him
about the same issues. 01 asked C3 to inform him of any action C3 may take in
organizing the business owners in the area. 01 added he would like a copy of the
letter C3 plans to send to the legislator. Ol informed C3 of the legislator
responsible for this district. Further, 01 gave C3 the name and address of the
president of the neighborhood association who may also be able to offer
assistance to C3. 01 told C3 that signs that read “no trespassing” needed to be
posted on the property. [Later in the same shift...] The business owner with whom
01 spoke earlier had left the store to have lunch. His business partner and wife
talked to 01. Ol helped her complete the [trespassing] forms. 01 stated she and

70

her husband needed to post “no trespassing” signs in several places about the
property. During the encounter 01 was helpful and fiiendly. The business
partner’s wife calmly filled out the necessary paperwork. She seemed pleased
that police would take some action if they complained about the vagrants.

The final community policing activity, ‘crime prevention’ was derived from the
following question, “During this activity, were the police trying to prevent the occurrence
or recurrence of the problem?” This was a yes/no question where the activity was only
coded as crime prevention if the officer’s efforts were focused on a period beyond the
end of the shift. In other words, the action taken by the officer must be clearly future

oriented.

v 01 stopped in a large parking lot in fiont of the store (a home maintenance
outlet). He said that the store had suffered a string of shoplifting and other
problems, but that since he had started parking there regularly in visible areas,
they had only one small shoplifting reported. 01 said that his theory was that just
by being visible to the public, the police can stop a lot of illegal activity because
people see they are being watched and are likely to be more careful.

v 01 and 02 go to the ground level of the police station. They discussed getting
new dead bolts with keys. Ol explained that they were currently using an
apartment (located in a degraded, lower class apartment complex) as an office and
wanted to get new locks for the door. They were concerned that people would
attempt to break into the apartment once they found out that the police were using
it as a pseudo headquarters. 01 continued by stating that the local drug dealers
had already thrown rocks through the apartrnent’s windows after news of their
presence got around. 01 and 02 met with a black male in his 305, a maintenance
worker employed by the police department. 01 informed the maintenance worker
about getting the locks installed. 01 gave the worker the address of the apartment.
The worker said he would relay the request to his superiors. 01 gave the worker
her pager number and asked that his supervisor get in contact with her as soon as
possible.

71

These narrative examples provide an indication of what types of police activities
are conceptualized as community policing in the current study. The specific construction
of the two measures of community policing (time per 8-hour shift and acts per citizen) is
discussed in the next section, followed by a presentation of the descriptive statistics for
these variables and the six community policing indicators used to create them.

Providing Acts of Communig Policing to Citizens

This dependent variable (CP Acts) represents the number of citizens receiving
community policing acts by the officer, divided by the total number of citizens coming
into contact with the officer during the data collection period. Providing community
policing to citizens is measured by the number of times the officer (1) provided comfort
or reassurance to citizens, (2) provided referrals to citizens, and (3) provided information
to citizens. These three measures were summed to provide a variable representing the
total number of community policing acts the officer performed during the data collection
period. According to Carrnines and Zeller (1979), scales that produce reliability
coefficients greater than .70 are considered reliable. The reliability coefficient
(Cronbach’s Alpha=.60) indicates that these three activities fall below the conventional
standard; therefore findings related to this variable should be interpreted with caution.

Due to the method of data collection (i.e., field observation where some officers
were observed more than others), the total number of community policing acts was
divided by the total number of citizens with whom the officer came into contact during
the data collection period. The resulting variable (CP Acts) is therefore a standardized
measure of the number of community policing acts provided by officers per citizen

encountered during the data collection period. Despite this standardization process,

72

 

results could still be impacted by the ride-based sampling strategy and therefore
additional tests were performed on this variable. Analyses were conducted to determine
whether CP Acts varied significantly depending on the number of rides for which the
officer was observed. Results indicated that the mean CP Acts did vary according to
amount of observation: officers observed for one ride provided about one CP Act;
officers observed twice provided about two CP Acts; officers observed three or more
times provided almost 5 CP Acts F (2, N = 318) = 94.22, p <.001. Despite the
standardization of this variable, officers with multiple observations tended to provide
more CP Acts than officers observed only once. What is different about officers who
were observed multiple times? They were compared according to their demographic
characteristics, features of their work environment, and their levels of social capital.3
Only three (out of 14) of the independent variables varied significantly according to the
number of rides observed. Officers with more education were more likely to be observed
more than once F (2, N = 318) = 5.04, p <.01, as were officers working in SPPD x2 (2, N
= 318) = 8.54, p <.05. Conversely, officers with higher levels of community policing
training were more likely to be observed only once F (2, N = 318) = 3.47, p<.05. These
differences should be kept in mind as they indirectly affect the average number of CP
Acts provided to citizens.
Time Spent Engaged in Community Policing

This dependent variable (CP Time) represents the number of minutes an officer

spent on community policing activities, per 8-hour shift worked by the officer during the

 

3

Detailed descriptions of these independent variables are provided in the next section.

73

data collection period. Community policing is measured by the time spent by the officer:
(1) engaged in problem-solving activities, (2) engaged in crime prevention activities, and
(3) attending community meetings. These three measures were summed to provide a
variable representing the amount of time (in minutes) that the officer was engaged in
community policing during the data collection period. The reliability coefficient
(Alpha=.72) indicates that these three activities exceed the conventional standard of .70;
thus, this scale can be considered reliable.

Due to the method of data collection (i.e., field observation where some officers
were observed more than others), each officer’s community policing minutes were
divided by 480 minutes to standardize the measure for an 8-hour shift. The resulting
variable (CP Time) is therefore a standardized measure of the number of minutes per
shift the officer spent on community policing during the data collection period. Analyses
were conducted to determine whether CP Time varied significantly depending on the
number of rides for which the officer was observed. Results indicated that mean CP
Time did not vary according to the number of observations F (2, N = 318) =.64, p =.53.
Overall, this dependent variable appears to have less measurement error and sampling
bias compared to CP Acts.

Description of Dependent Variables
Descriptive statistics for the two dependent variables, and the variables used to

create them, are presented in Table 2.

74

Table 2

Measurement and Descriptive Statistics for Dependent Variables.

 

 

 

 

Variable Description m Mp; Mean §._D_,
# Citizens receiving comfort/reassurance 0.00 28.00 2.92 3.76
# Citizens receiving information 0.00 45.00 6.19 7.45
# Citizens receiving referrals 0.00 98.00 24.50 20.95
Total # citizens receiving CP acts 0.00 145.00 33.61 29.19
Total # of citizens encountered by officer. 1.00 54.00 17.17 9.25

CP Acts = TOTAL CP ACTS / TOTAL 0.00 14.50 2.42 2.41
CITIZENS ENCOUNTERED
“Acts of comforting, providing
information, and/or providing referrals per
citizen encountered.”
Alpha=.60
# Minutes attending community meetings 0.00 319.00 4.46 30.20
# Minutes problem-solving 0.00 640.00 9.37 53.53
# Minutes engaged in crime prevention 0.00 791.00 26.52 87.46
Total # Community Policing Minutes 0.00 1559.00 40.45 148.34
Total Shifts Observed (Total Minutes 0.15 10.13 2.30 1.84
Observed / 480 Minutes)

CP Time = TOTAL CP TIME / TOTAL SHIFTS 0.00 194.17 11.02 25.37

N=318

“Minutes spent attending corrrrnunity
meetings, problem-solving, or engaged in
crime prevention per 8-hour shift.”
Alpha=.72

Notes: Values provided for CP Acts reflect the change of one outlier from 34.0 to 14.5.
CP Acts and CP Time were transformed to integers for Negative Binomial regression requirements.

75

 

The three community policing activities comprising the CP Acts variable include
comforting or reassuring citizens, giving information to citizens, and/or providing
citizens with referrals. During the approximately 3-month long data collection period,
the average officer comforted approximately three citizens, provided information to six
citizens, and gave referrals to 24 citizens. Summing these three acts reveals that the
average officer provided a form of community policing to about 33 citizens during the
data collection period. The average officer came into contact with about 17 citizens
during the data collection period.

Descriptive statistics for the CP Acts variable indicate that the average officer
provided more than two acts of community policing per citizen encountered. The
variable ranges from a minimum of zero acts per citizen encountered, to a maximum of
15 acts per citizen encountered. The majority of officers (n=290; 91%) provided at least
one community policing act per citizen. Similar proportions of officers provided one act
per citizen (n=132; 42%) and from 2-5 acts per citizen (n=126; 40%). A small number of
officers provided (n=25; 8%) 6-10 acts, and six officers (2%) provided 11-15 acts of
community policing per citizen encountered. Only 28 officers (9%) did not provide a
single act of community policing during the data collection period.

The three community policing activities comprising the CP Time variable include
attending community meetings, problem-solving, and/or crime prevention. Each of these
indicators is measured in minutes. During the data collection period, the average officer
spent about four minutes attending community meetings, about nine rrrinutes problem-
solving, and about 26 minutes engaged in crime prevention. Summing these community

policing indicators reveals that the average officer spent about 40 minutes on these

76

 

 

community policing activities during the data collection period. The average officer was
observed for approximately two 8-hours shifts.

Descriptive statistics for the CP Acts variable indicate that the average officer
spent about 11 minutes per shift on community policing, or roughly 2% of each shift.
This variable ranges fiom a minimum of zero minutes per shift, to a maximum of 194
minutes per shift. Unlike the CP Acts variable, the majority of officers (n=200; 63%)
spent no time engaged in community policing as measured by problem-solving, attending
community meetings, or engaging in crime prevention activities. Per shift observed
during the data collection period, 48 officers (15%) spent 1-15 minutes on community
policing, 29 officers (9%) spent 16-29 minutes, 18 officers (6%) spent 30-45 minutes,
seven officers (2%) spent 46-60 minutes, 11 officers (4%) spent 61-120 minutes, and
four officers (1%) spent more than 120 minutes. Removing the officers who spent more
than 120 minutes from subsequent bivariate and multivariate analyses (presented in
Chapter 6) did not affect the results.

Measurement of Independent Variables

The independent variables included in the present study are grouped into three
categories: (1) officer characteristics, (2) characteristics of the officer’s work
environment, and (3) social capital dimensions. The respective measurement of and
descriptive statistics for these categories of variables are presented in Tables 3, 4, and 5.
Measurement of the variables comprising each category is first described, followed by a
discussion of the descriptive statistics for all the independent variables included in this

study.

77

Officer Chagrcteristics

Officer characteristics include dummy variables for officer race (0=white,
1=Afiican American, Hispanic, Asian or Other) and sex (0=male, l=female). The
variable for officer education has eight categories: 1=less than High School, 2=High
School/GED diploma, 3=Junior College, 4=Associate’s Degree, 5=two or more years of
college, 6=Bachelor’s Degree, 7=some graduate work, 8=Graduate Degree. Officer
tenure is an interval-level variable representing the number of years the officer has
worked at the department, created by subtracting the year the officer began working for
the department from 1996 (Indianapolis officers) and 1997 (St. Petersburg officers). A
community policing training scale was constructed from seven types of training that the
officer may have received: public speaking; computer/automated systems; community
policing principles; code enforcement/civil regulations; mediation; analyzing
neighborhood crime data; and organizing community groups. Items were coded so that
high values represented more training (l=none, 2=less than one day, 3=1- 2 days, 4=3-5
days, and 5=more than five days). The scale ranged from 7 to 32 and the reliability

coefficient for this scale is .81.

78

 

 

 

lz

 

Table 3

Measurement and Descriptive Statistics for Officer Characteristics.

Variable Description Values % / Mean SD.

Female Officer is female. 1 = yes 20.8 0.41
0 = no 79.2

Non-white Officer is non-white. 1 = yes 28.6 0.45
0 = no 71.4

Education Officer’s highest level 1 = Less than H.S. 0.7 1.63
of education. 2 = HS/GED 16.4
3 = Jr. College 26.6

4 = Assoc. Degree 6.8
5 = 2+ yrs. College 17.4
6 = Bach. Degree 27.3
7 = Some Grad. 4.4
8 = Grad. Degree 0.3

Tenure Years at department. 0-31 9. l 7.21

CP Training Scale of 7 types of Community 7-32 15.6 5.25
Policing training (Alpha = .81).

 

Public Speaking 1 = None 62.2 1.29

2 = < 1 day 15.6

3 = 1 to 2 days 9.5

4 = 3 to 5 days 3.4

5 = > 5 days 9.2
Computer/Automated 1 = None 2.4 0.98
Information Systems 2 = < 1 day 21.1

3= 1 t02days 41.5

4 = 3 to 5 days 23.8

5 = > 5 days 11.2

79

 

Table 3 (cont’d).

Concepts/Principles of CP

Code Enforcement/Civil
Regulations

Mediation

Using Crime Data to Solve
Problems

Organizing Community Groups

1=None
2=<1¢w
3=1t02days
4=3t05days
5=>5days

1=None
2=<1day
3=1t02days
4=3t05days
5=>5days

1=None
2=<1mw
3=1t02days
4=3t05days
5=>5days

1=None
2=<1day
3=1t02days
4=3t05days
5=>5days

1=None
2=<1day
3=1t02days
4=3t05days
5=>5days

9.6
16.1
34.6
26.7
13.0

33.0
28.9
19.9
8.2
10.0

54.8
18.5
16.4
6.5
3.8

50.2
29.4
13.7
4.8
2.0

76.4
13.4
6.8
1.4
2.1

1.28.

1.14

0.99

0.84

 

N=318

80

 

 

 

Work Environment

The second group of independent variables, officer’s work environment, consists
of five items. Three items are dummy variables: the officer works during a day shift
(1=yes, 0=other shift), the officer has a community policing assignment (0=general patrol
assignment, 1=community policing assignment) and the department where the officer
works (1=Indianapolis, 0=St. Petersburg). The next item in this category of variables is
an additive scale containing seven issues the officer perceives to be a major problem in
his or her beat: theft or burglary; litter and trash; vandalism of cars and property; drug
dealing; gangs; loitering; and abandoned buildings. All items in this scale were coded as
follows: 1=not a problem, 2=minor problem, 3=major problem. The scale ranged from 7
to 21 and the reliability coefficient for the scale is .72.

The fifth item is a scale that was created to reflect the officer’s perception of
whether his or her department is supportive of community policing. This scale includes
five items that were coded so that high values represent the officer’s belief that the
department does support community policing (l=poor, 2=fair, 3=good, 4=excellent).
Statements comprising this scale include “the department clarifies the role of officers in
community policing,” “the department fairly distributes the workload of community
policing and patrol officers,” “the department gives officers enough time for community
policing,” “the department gives officers information for community policing,” and “the
department rewards officers for community policing.” The scale ranged from 5 to 20 and

the reliability coefficient (Cronbach’s alpha) for this scale is .82.

81

Table 4

Measurement and Descriptive Statistics for Work Environment Variables.

Variable Description Values % / Mean S_.IL
Department Officer’s department 1 = Indianapolis 59.7 0.49
0 = St. Petersburg 40.3
Day Shift Officer works day shift 1 = yes 37.7 0.49
0 = no 62.3
CPA Community Policing Assignment 1 = yes 37.4 0.48
0 = no 62.6
Beat Problems Officers’ perceptions of 7 beat 7-21 15.7 2.96
Scale problems (Alpha = .72)
Theft 1 = Not a problem 1 4 0.53

2 = Minor problem 50.3
3 = Major problem 48.3

Litter 1 = Not a problem 20.0 0.73
2 = Minor problem 44.8
3 = Major problem 35.2

Vandalism 1 = Not a problem 15.9 0.64
2 = Minor problem 57.8
3 = Major problem 26.3

Drug Dealing 1 = Not a problem 7.3 0.63
2 = Minor problem 28.8
3 = Major problem 63.9

Gangs 1 = Not a problem 26.3 0.77

2 = Minor problem 40.1
3 = Major problem 33.6

82

Table 4 (cont’d).

 

Loitering 1 = Not a problem 17.2 0.74
2 = Minor problem 37.6
3 = Major problem 45.2
Abandoned Buildings 1 = Not a problem 25.6 0.78
2 = Minor problem 38.8
3 = Major problem 35.6
Dept. Pro CP Officer’s perceptions of dept. 5-20 9.9 3.34
Scale support of CP (Alpha = .82).
Dept. clarifies role of officers in 1 = poor 29.5 0.90
community policing. 2 = fair 34.2
3 = good 30.8
4 = excellent 5.5
Dept. fairly distributes workload 1 = poor 46.0 0.91
of community policing and patrol 2 = fair 30.4
officers. 3 = good 18.3
4 = excellent 5.2
Dept. gives officers enough time 1 = poor 46.4 0.94
for community policing. 2 = fair 28.2
3 = good 19.2
4 = excellent 6.2
Dept. gives officers info for 1 = poor 22.7 0.90
community policing. 2 = fair 40.5
3 = good 28.2
4 = excellent 8.6
Dept. rewards officers for 1 = poor 33.8 0.78
community policing. 2 = fair 45.6
3 = good 18.1
4 = excellent 2.4
N=318

83

Social Capital Dimepsions

Social capital dimensions include: (1) level of trust, (2) cooperative exchanges,

(3) group cohesion, and (4) social support. Table 5 contains the measurement and

descriptive statistics for each of the four social capital dimensions. Factor analyses were

conducted on three of the four social capital dimensions (cooperative exchanges, group

cohesion, social support), as they include multiple indicators. Results of those analyses

are presented in Table 6, located at the end of this section.

Table 5

Measurement and Descriptive Statistics for Social Capital Dimensions.

Variable Description

Trust Officer has complete faith in
supervisor.

Cooperation Officer gathers public safety
info from other officers.

Officer gathers public safety
info from supervisor.

Proportion of unit that officer
would share hard-to-get info.

Values

1 = disagree strongly

2 = disagree somewhat
3 = agree somewhat

4 = agree strongly

1 = never

2 = rarely

3 = sometimes
4 = often

1 = never

2 = rarely

3 = sometimes
4 = often

1 = none
2=afiw

3 = about half
4 = all or most

5.2
11.9
29.0
53.8

0.0

6.2
33.1
60.7

5.9
26.2
45.9
22.1

0.3
15.4
10.9
73.4

MEL

0.88

0.61

0.83

0.76

 

Table 5 (cont’d).

 

Group Cohesion Officer rating of work unit. 1 = not as good 3.1 0.55
2 = about the same 36.3
3 = better than most 60.6
Proportion of unit that officer 1 = none 2.1 0.87
considers to be friends. 2 = a few 19.9
3 = about half 21.0
4 = all or most 57.0
Officer enjoys working with 1 = disagree strongly 5.2 0.87
supervisor. 2 = disagree somewhat 9.3
3 = agree somewhat 11.4
4 = agree strongly 74.0
Support Supervisor supports officer 1 = disagree strongly 5.3 0.85
when he/she is right. , 2 = disagree somewhat 8.5
3 = agree somewhat 28.9
4 = agree strongly 57.4
Supervisor seldom criticizes 1 = disagree strongly 3.8 0.77
officer. 2 = disagree somewhat 5.9
3 = agree somewhat 19.4
4 = agree strongly 70.8
Supervisor looks out for 1 = disagree strongly 4.5 0.82
welfare of subordinates. 2 = disagree somewhat 7.3
3 = agree somewhat 21.6
4 = agree strongly 66.6
N=3 1 8

85

As T able 5 indicates, level 0 f II'ust is measured by the officer’s response to the
statement “I have complete faith in my Supervisor,“ and is coded -1=disagr€6 strongly,
2=<1isagree somewhat, 3=agree somewhat, and 4=agree strongly. High values therefore
represent a higher degree of trust Steffi-reported by the Officer about his or her supervisor.

The COOperative exchanges dimension is measured by three items. One item
measures the frequency that the officer gathers pubh'c Safety information from other
officers, and one item measures the frequency that the officer gathers public safety

information from his or her supervisor. Both items are coded as 1=never 2=rar61Y’

3=sometimes, and 4=often. The third item is the officer’s response t 0 the question, ‘

. . - . t
you obtained some hard-to— get information about the Identity of an off ndet causing a X0
6

Qt trouble 'm your district, With how many of the officers in Yew Uni ould you sham
t w

read?”

3‘10“?” 311d COded as 1=n0ne, 2=feW, 3=abOUt half {If—all moSt
: ~ or '

this inform
' ‘ lded com onent, o with an
Componentsfac‘OY Analers yle one P r factor, fro these it6 6
Eigenvalue of \. 3% . The Kaiser-Gunman rule suggests that factors With an E i gent/81113
greater than one should be retained (Kim & Mueller, 1978)- The factor expj ai 35 almost
40% of the variance in the included variables. Cronbach’s Alpha (.23) indicates that this
tor falls well below the conventional standard of reliability (.70). Overall’ there 1.5

fac

substantial measurement error in this factor and results should be interpreted With

caution.

Group COhesion is measured by three items. F irst, Officers were asked to give a

 

——

4

Webster‘s dictionary defines faith as “confidence or trust in a pf?
variable as a measure of trust, although it could also be measurlng

rson 01' thing.” I am the

re ing ‘
confidence or faith fore us " this

86

 

 

t' of their w ' '
ta mg ork unit. Their responses were coded as f0110ws: l‘not as good asmost
others, 2=abou
tthe same as most others, and 3=better than most other units. The second
item represents a proportion of officers in the reSpondent’s k unit that he or she
wor
considers friends, coded as 1=none, 2=a few 3‘abo thalf d 4 11 or most The thifd
a m u , an :3 ‘
item was the officer’s response to the state
ment “M . - f son I
’ Y SUpervrsor IS the type 0 per
enjoy working with ”5 and was coded 1 5disa
, gree Stron _. ' so what
gly, 2—drsagree me ,
3=agree somewhat, and 4=agree strongly. Principal Components Factor Analysis yielde

30/0 Of

one component from these items with an Eigenvalue o f 1 30 The fa t plains 4
- - c or ex
t

the variance in the included variables. The Alpha (.29) indicates th t thi f ctol’ (103‘5 no
a S 3
n the

meet the conventional level of reliability and this may have a
ental imp

results.
Three items are used to measure the third social Capital . 51$ support.
dlménSiOns 50 6
They represent the Officer" s responses to the statements “M ort m
’ y Sllpe ' 1 5‘39?
mi 501' W11
when I am right, ev en if it makes things difficult for him/her,» “Th of
e decisions

judgments I make are seldom criticized or modified by my SupewiSor ,, and “NY

supervisor looks out for the personal welfare of his/her SUbOl'dinates-” A11 items were
coded 1=disagree SlToneg, 2=disagTCe somewhat, 3=agree somewhat, and 4§egree
Stroneg, Principal Components Factor Analysis yielded one component from these
items with an Eigenvalue of1,91, The factor explains 64%. ofthe variance in the
included variables. The Alpha (.71) indicates that this factor meets the ConVentional

level of reliability and is therefore a reliable measure 0f the scam SUPPOIT Construct

‘—

s
This item originally read “My supervisor is NOT the type of person I enjoy Working With ., The item w
- 'as

\QN use oooed.

87

 

Table 6

Princi alCom onent Factor An
al 565 of So -
cral Ca ital Dimensions
Factor Item _ Factor
0.312 0.558

Cooperation Officer gathers public Safet .
from other officers. y lnformation

 

 

 

 

 

 

 

 

 

 

 

 

 

Officer gathers public Safe
t .
fi'om supervisor. y Information 0'3 83 0'619
Proportion of unit that offic
er we
share hard—to-get information wi tlil 1d 0488 0698
Alpha ..................................................................................................................
OE lgenvalue 1 .fi: ................
Group 0 fficer rating of work unit. 0 w
Cohesion 0 ‘572 0.
Proportion of unit that officer considers t of] 83
be friends, 0 0.62X
Officer enj oys Working with supervison 0.328
................................................... - - 010$
Alpha 29 - - ---.
glgenvalue 1.30
5 /° Variance Explained ‘ 43 .3 3
”FPO” .SuperVisor supports officer When he/she x J
18 right. 0'79 1 0.89
SUpervisor seldom criticizes Officer. 0 349
0.591
Supervisor looks out for welfare of
s b d' O. 77
u or mates. 0-8 7 7
Alpha- ~ “.7, i ..............................................................................
Eigenvalue 1,9 1 iiiiiiiiiii
% Variance Expw 6
N's?» 1 8 i 5 H

88

 

Description oflndependent Variables
The majority of officers are male (79%) and white (71%), More than one in {our
officers holds a bachelor’s degree (2 7%) , and more than half of the officers (56%) have
an associate’s degree or more education- The average officer worked in the police
depmem for nine years The mean re Sponse for the community policing training scale
was 15 -6- Most officers reported receiving no training on organizing community groups
(76%), public Speaking (62%), or mediation (55% ). Most officers received some training
in computer/automated information systems (98%), Co I] c ePtS/Principles of community
policing (90%), or code enforcement/civil regulations (67%)- About half of Officers
received some training on using crime data to solve Problems (50%) .
Thirty-eight percent of officers worked the day shjfi, 60% Worked mme
lmhanapolis police department during the data collection Period, and about one in three
{colours

officers was a community policing specialist. The mean reSDOnse 1: th beat ‘3
Q e

scale was l5 .1 ,indicat‘mg that most officers perceived seVeral issues t be problems in
0

their beats. The problem most frequently described by °ffi°ers as a “major problem” was
drug dealing (64%), followed by thefi (48%)? and loitering (450“) The issues most
frequently described by officers as “not a problem” were gangs and aband‘m'ed buildinos
D

(26% each), followed by litter (20%) The Department Support of COmmmli 1y Policing

S cale had a mean response of 99, Less than one in ten offi c ers described their

department as “excellent” on any of the five items comPfiS ing this scale. Officers most

frequently described their department as “poor” on two 0f the items: (1) “my department
fairly distributes the workload of community policing and patrol officers” (46%): and (2)

“we; gevmment gives officers enough time for corfll’nullity pOIiCing” (460/0),
89

 

 

 

Regarding the social capital dimensions, most officers scored high on level of
trust (83% agreed that they had “complete faith” in their supervisors). Items in the
cooperation dimension reveal that 61 % often gathered public safety information from
other officers, and 730/0 said they wou1d Share hard-to-get information with all or most of

the officers in their work group. Fewer cooperative exchanges occurred with
supervisors: less than one in four (22%) Often gathered public safety information from
their supervisors. The group cohesion dimension {eve a] s that about 6 out of 10 officers
consider their work units “better than most others,” (6 1 %) and consider all of most
officers in their unit to be friends (57%). A majority o f officers (78° /0) enjoy working
with their supervisor. The support dimension also shows a high degree of positiVe

sentiment among the Officers, More than eight Of ever-y ten melt
officer d max
8 agree

supervisor rarely criticizes them (900/0), looks out for the w 61 f {oifia‘es
are 0 1‘ their Subo

(88%), and supports the 0 fficer when he or she is right, even i f _ ’ngs
ay m

difficult (86%).

9O

 

CHAPTER 5

ANALYTIC MODELS AND METHODS

Analytic Models
Regression analyses were conducted on additive models (one per dependent
variable) to determine the direct effects of the independent variables on community
policing, and then a series of interaction models (seven per dependent variable) to
determine whether- the relationship between social Capital and Comfmmity policiflg is
moderated by various officer characteristics and features 0 f their Work environment. For
the additive models, 14 independent variables were inc 111de in the

- i
. allys‘s 0
regr‘e's’s‘mn w
6 resultin in roxirnatel l8 cases er v - no
262 officers, g app Y P anable. For the int 31730“

models, l8 independent V enables were included, resulting in a A cases per-
pp I‘(33-ti1natel3’
' , idin t B and Raudenbush (1992), a co
vanable A060 g o ryk mmon rllle of t b all for
multivariate analyses is a minimum of 10 cases per variable ineluded in the m ode]. The
sample to be analyzed in the present study therefore exceeds this minimum réquiremen t

Additive MOdeISI Assessin Direct Effects Amon Variables

Additive models were analyzed to determine the independent effects Ofotl‘icer

 

 

6

The statistical paCKage used in this study, LIMDEP, requires that the dataset be free from missing data
The liStWiSC deletion of cases Inissing scores on any ofthe variables resulted in the sample being eruC.
from 3 1 3 to 262 Officers. In order to test whether this changed the sample 0f officers in any meaningfufd
way, logistic regl‘i‘fission analyses were run on the full sample (N=313) ‘0 determine whether any OfIhc
independent variables significantly predicted the officer being excluded fi‘om the LIMDEP sample (”=2
The dependent variable in these analyses, missing (coded osinclnded meample, 1=missin from s 1 62).
was not predicted by any of the independent variables to a statistically Slgmficant eXtent (pi 05) amp e),

91

 

characteristics (0C), work environment (WE), and social capital (SC) on officer
performance of community policing (0}) Tirne and CP Acts). These models test the
hypothesis that social capital increases the likelihood that officers will spend time on
community policing and provide community policing to citizens. The general equations
for these mOdels are presented below. Recall that 0C, WE and SC represent categories
of variables.
CP Acts = 01+ lHOC) + I3(VVE) +3(SC) + 9
CP Time = a + WOC) + WWI?) + 5(SC) + 0.
Significant coefficients for the social capital Val-i ables indicate that $00.1 a1 capital
does matter,” and a positive coefficient means that so cial capital significan‘w increases
the ukefihood of an officer engaging in connnunity POIicing, A 81-ng fie an‘ positiv e h
coefficient would provide evidence that the central hYPOthesiS p ropose d by this teseafc
is supported ‘0‘} the data.
mdependent variatnes included in the additive models Were tested to determine if
multicollinearity Was a problem. A condition number was derived by dividing the largeSt
characteristic root from the correlation matrix by the smallest, then taking thé square root
ofthat numb er.7 According to Greene (2 000), a condition nmnber less than 20 indicates
that the variables are not muItieollinear. The condition number for the matri)K of
independent Variables included in the additive models was 4 - 32, indicating that COHCemS

regarding multicollinearity are unwarranted for the additive models. A correlation matrix

/ .—

 

-

I

Practically, this was accomplished by entering the correlation matrix :01; tile independent variables into a
database that was then read into LIMDEP. Commands were thfinfpe 1e ‘0 obtam the characteristic too
for the matrix- The Condition number was then derived “by han - ts

92

of all variables is provided in Appendix A

Interactive Models: Testin Moderated Cansal Relatio shi s
n

Moderated causal relationships occm- When the relationship between X (predictor
variable) and Y (outcome variable) Varies depending on the value of Z (moderator
variable) (Jaccard, Turris, & Wan, 1990). A mod erat or variable affects the direction

and/ or strength of the relationship between the Predictor and outcome, so moderators are

usefiil for establishing ‘Vvhen certain effects hold” (B at n & Kenny 1986) In Centrest to
0 9 '

mediated relationships (which attempt to account for the relationship b tween X and Y,
e

or suggest “why or how such effects occur”), moderated relationshj app ropfiate t0
ps are

test when there is greater interest in the predictor than the moderat M (1.13th
01-. C

relationshipS, on the other hand, suggest that the researcher is most st ed 3;; me
intete

mediator Variables rather than the predictor variables (Le sacs an

-, Officer Charactefi
red't‘lior of
driven P

interest in the current study is social capital, the methOdOIOgical lit upPorts the
erature 5

work environment rather than social capital). Since the theor e tic 1
a 1
y—

decision to examine moderated causal relationships rather than media ted relationships

(Baron & Kennedy, 1986; Hardy, I993; Jaccard, Tunisi, & Wan, 1990)- res ting

interactive models enables an evaluation of how the relationship between soQ fa] capit I
a

and community policing (if one exists) is moderated by officer characteristics an d/or

features of their work environment.

In Ofder to evaluate the presence of moderated causal relationShiPS, several
interaction mOdels were analyzed to determine the j oint effects 0f Pr ed1°Ctor (SC) and
moderator (0C, WE) variables on officer performance 0f community Policing. Recall

that SC, 0C, 311d WE represent categories of variables- The inter action models include

93

the independent variables from the additive model, with the addition of interaCti on terms
representing the social capital dizn eHSions (trust, cooperation, group cohesion, support)
moderated by officer characteristics (sex, race, education, tenure) and features of their
work envimnmem (department, community Policing assignment, department support of
community policing). Descriptive information for the Specific interaction terms to be
included in the interactive models is presented in Table 7. Each of the seven interaction
models was tested on the two dependent Variables. For ease of presentation, the
dependent variable in the equations is referred to as CP. In reality each equation is taste
on both CP Time and CP Acts. The general equations for these models are presented
below.
Social Capital x Officer Characteristics interaction models
C? = a + t3(SC)+ (KOCH BONE) + mm x Sex) + e
C? = e. +— @(SCV’ BQOCYF BONE) + BCSC x Race) + e
C? = u + MSC) + HOG) + BONE) + [3(SC x Education) 4, Q,
C? = a + [3(SC) + 5(oc) + [3(WE) + B(SC >< Tenme) _,_ 9
Social Capital x Work Environment interaction models
CP = a + [3(SC) + p(oc:)+ [5(WE) + NSC X Dept.) + e
CP = a + («3(3) + {3(OC) + 5(WE) + [3(SC >< CPA) + 9
CP = a + [3(SC) + {3(0C) + B(WE) + NSC x Dept SUPPOH ofCP) + 9
Results of the interaction analyses allow important information to be gained along

two fronts: Whether the coefficient is significant and Whether the direction of the

relationship is positive or negative. For example, if any 0f the 500131 Capital x officer

female coefficients are statistically significant, this would mdlcate that the relationship

94

between social capital and community POIiCing depends On the sex of the officerfile.,
female officers moderate or Change the relationship between social capital and
community policing in way that is Significantly different from male officers). If the
coefficient is positive, then the lik€1ih00d 0f community policing significantly increases
when females have high levels of social capital. If the coefficient is negative, then the
likelihood of community policing is significantly reduced when female officers have low
levels of social capital. This interaction term can tell us, therefore, Whether the
community policing performance of female officers is better understood as a function of
their 16V€15 0f social capital being either high or low.

In general, the interaction terms indicate when the relationship between social
capital and communi ty policing is moderated by officer character-1°Stics and features of
their work envirOMent. The interactiOn terms also enable an understanding of whether
high or low levels of social Capital are better able to account for community poliCing

performance, for Which Officers and under which circumstances.

95

T able "I

Measurement and Descri tive Statistics for Interaction Terms.

Interaction Term m M w _S_Ll_)_g
Trust Variable X Ofc. Female 0 4.00 0.45 1.18
Cooperation Factor x Ofc. Female -2.81 1.41 0.00 0-35
Group Cohesion Factor x Ofc. Female ~2.48 1.05 0.00 0.38
Support Factor X Ofc. Female -2.78 0.79 0.00 0.27
Trust Variable >< Ofc. Minority 0.00 4.00 0,72 1.42
Cooperation Factor >< Ofc. Minority -2.95 1.41 0.00 0.52
Group Cohesion Factor X Ofc. Minority -2.7 7 1.05 0.00 053
Support Factor X Of}; Minority -2.94 0.79 0.00 0.51
TI'USt Variable x OfC Education 2.00 32.00 14.00 655
Cooperation Factor >< Ofc. Education -17.73 0.99 0.00 4.48
Group Cohesion Factor X Ofc. Education -17.06 7.36 -026 4.74
Support Factor x Ofi. Education ’26'17 5‘5 1 ,020 4-73
Trust Variable X Ofc. Tenure 0-00 12400 29.74 25.92
Cooperation Factor >< Ofc. Tenure '65-13 39-51 - 1.20 1 1.90
Group Cohesion Factor X Ofc. Tenure ‘56-33 32-58 0.00 11.05
Support Factor X Ofc Tenure '78.17 24.42 "O.19 1215
Trust Variable X CP Officer 0'00 4'00 1 .22 1.69
Cooperationractor x CP Officer '2-33 1-41 0. 00 060
Group CohesiOn Factor x CP Officer '3'35 1'05 -0.01 0.66
SupportFactOT X C? Officer “2.94 0.79 0.00 057
Trust Variable x Dept. (IPD) 0.00 4.00 1.79 1 77
Cooperation Factor x Dept. (IPD) -295 1-41 0.00 0 '75
Group Cohesion Factor X Dept- (IPD) ‘335 1'05 0-00 0:7 6
Support Factor X DCpt. (IPD) -3.74 O79 0.00 0.79
Trust Variable x Dept. Support 5 80.00 33.57 15.94
Cooperation Factor x Dept. Support 37.22 28-22 0.74 9.99
Group Cohesion Factor x Dept. Support -2994 21.01 0.98 9.79
Support Factor X Dept" Support -40.64 15.76 0.76 9.67

 

 

 

 

 

 

/

5318

96

independent variables inclUded in the seven interactive models were tested to
determine if multicollinearity was a PTOblem. These sets of variables change for each
model, so a condition number W as derived for each model’s independent Variables. The
condition numbers obtained for the sex and race interactive models (1991 and 18.70,
respectively), fell just below the value of 20 specified by Greene (2000), indicating that a
problematic level of multicollinearity is (not quite) present in these models. The
condition numbers derived for the other interactive models, how ever, did exceed the
value of 20. Specifically, they were 29.02 for the education model, 2550 for the tenure
model, 21.53 for the CPA model, 24.93 for the department mode1, and 31.02 for the
department support of CP model.

Concerns regarding multicollinearity are therefore warranted for many of the
interactive models. Greene (2 000) Pomts out that one cause Of mu 1ticollinearity
problems is a shortage of infomation, the solution being to obtain more data. This is not
a feasible solution for the Present Study, but is one that future res‘eiiu'chers Should bear in

mind. The most practical SOlution suggested by Greene (2000) to deal With PTObIems of
muttieomnearity is to dmp the offending variables- This sugges‘ts that the researcher has
not been guided by theory during model specification. The models in the entrant
research, however, were specified according to theoretical considerations. Silnply
remoVing the offending variables (i.€-, the interaction terms) is not a practical SOIution
given the focus of the current research. Another method to reduce multico lhnearity is to
cwmbine variables that are highly correlated by creating factors or scales. Many of the

independent Variables in these models were conceptualized as factors and scales, and

31’ eady have been constructed as such.

97

 

Readers should be aware that multicollinearity is an issue that is presentin the
interactive models, but should also keep in mind that the most PIOHOunccd way that
multicollinearity affects results is by making important interaction terms more diffiCUIt to
detect statistically (Jaccard, TurriS, & Wan, 1990). The adverse effects are therefore
considered to be substantive rather than practical. Multicollinearity does not affect the

properties of regression estimates: they remain the best linear unbiased estimates (BLUE)

unless perfect coIIi nearity exists (J accard, Tunis, & Wan, 1990).
Analytic Methods
Dependent variables that involve counts of the number of times a particular act or
event occurred, such as those being tested in the current study, Can be found throughOUt
social science research. Examples include studies of the number 0 f articles published by
academics (Allison & Long. 1 990), the number Offimes that P601) 16 visit the doctor in a
certain period (Camaon & Tn'vedi, 1986; Beland, 1980), mortality rates (Hemstrom,
1999., Thouez, 1984; Vlahov et al., 2000), and suicide rates (Aaslamd, Ekeb erg, &
Schweder, 2001',Morre11 et al., 1999), among others. Examples specific to Cn'min 010gy
include studies of criminal careers (D’Unger Ct 31" 1998; Land, McCaH, & Nagin, 1 9 96),
victimizations in American cities (Nelson, 1980) and Britain (OSbOm & T381011]; 1998),
police killings (Jacobs & O’Brien, 1998), domestic violence incidents (Shel.man et al.,
1992), and homicide counts (Grogger, 1990; Sampson & Raudenbush, 1999). The
prevalence of count data in social science research, and criminological research

specifically, has led to the widespread application of regression models designed for

these data.

98

 

Regession Models for Count Data

There exists aburgeoning literature on regression models for count data that was
consulted for this study (D’Unger Ct al., 1998; Greene, 2000; Land, McCall, & Nagin,
1996; Long, 1997; Nelson, 1980; Zorn, 1998). Count outcomes have distributions that are
comprised only of nonnegative integerS, and are often skewed toward zero. in other
words, a count Variable represents types of events that are generally not experienced by

most of the saanle being studied, and is characterized by a nonlinear distribution.
Applying the linear regression model (LRM) would therefore produce values that are
inefficient, inconsistent, and biased. A more basic concern is that applying the LRM to
count data could produce predictions that are 1655 than 261'0- In the context of the current
research, applying the LRM could result in values indicating a negative number of
citizens served, or a negative amount of time spent on community policing, both of
which are nonsensical.
Fortunately there €Xists a group of regression models that have been deveIOped
particularly for count outcomes. In contTaSt to applying the LRM, these regression
models make the estimation ofcount models more efficient, more consistent, and less
biased. The most basic of these models is the Porsson regreSSIOIl model (PRM), Which
has two defining assumPtions, The first assumption is that the events being Counted are
independent of each other. When they are not, the process is known as CO" tagion and the
assumption of independence has been violated. This can 060111“ in two WaYs. First, if the
probability of performing community policing is the same for all officers but depends
upon prior Community policing performance, then the events are depencleIlt upon each
other- Secorxd, if the probability of performing community Policing is Constant over

99

 

time,but is not the same for all OffiCCI'S, then contagion has occurred. T116 second
assumption for the PRM is that the conditional mean of the outcome is equal to the
conditional variance. Equidisper-Sion is another term for the equality 0f the mean and

variance assumption. In practice both of these assumptions are often violated, leading to

the development of additional regression models for count data.
W The negative binomial regression Il’IOde1

(NBRM) builds Upon the PRM design but provides added flexibility by relaxing the.
assumptionsof independence and equidispersion that can limit the applicability of the
PRM. Contagion v i olates the assumption of independence required for PRM and can
result in a failure of the model to fit the data. Recall that one Way contagion can occur is
via heterogeneity in the rate among the individuals being studied; failure to account for
this heterogeneity can contribute to overdispersion in the marginal distribution (see Long,
1997, p. 221). The PRM includes Observed heterogeneity to accol—lnt for a different

likelihOOd of me count Outcome across individuals, but the NBRM also in 0111 des
unobserved hetefogeneity in an attempt to more completely address the prob Item of
contagion. With cross-sectional data, however, it is impossible to state definitively

whether the observed counts are produCtS 0f“apparent” contagion “'6" addressed by

including unobserved heterogeneity), or “true” contagion. It is a limitation ofthe

NBRM, but it is less of a concem compared to the PRM.
The primary benefit of the NBRM is that it was designed to address es the issue of

overdispersion that often exists in count data as a result of the variance exceeding the
mean- In “real life” it is rare that the variance equals the mean; the measures of

community policing in the current study also failed to meet this requirement. The

100

 

NBRM allows for the estimation Of overdispersed count Vaziables by adding a parameter
that allows the conditional variance t0 exceed the conditional mean, known as the
dispersion parameter, or alpha. When the alpha coefficient equals zero, the central
assumption of PRM has not been Violated and therefore PRM is suited to the data.
Another way of stating this is that the NBRM reduces to the PRM when the alpha
coefficient is equal to zero. Conversely, when the alpha coefficient is significantly
different fi'om zero, the NBRM is a more suitable model than the PRM because the data
are substantially over-dispersed. Erroneously applying the PRM to overdispersed data
will produce estimates that exaggerate the significance of the independent variables. The
NBRM avoids this problem by including the dispersion paramaer’ thereby achieVing
more accurate indications of the statistical importance of the Vatiab 165. Even among the
specially designed regression models for count outcomes there are substantial
repercussions for employing a model wrongly suited to the data.
When attempting to model count data, it is therefore imperative to determine
whether overdispersion exists in the data to a significant degree. The statistical Package
L IMDEP computes alpha, the dispeI'Sion parameter, for the NBRM. The PrOCess of
estimating the NBRM in LIMDEP is threefold: first the statistical software generates
prediction values using linear re gressiOn (0L5), then uses these values to eStiInate the
pRM, then uses the PRM values to estimate the NBRM. If the t-test 0f the alpha
parameter provided in the NBRM is statistically significant (p < .05) this is evidence 0 fa
Significant amount of overdispersion. A second way to test for overdispersion is to
Compare the log—likelihoods of the PRM and the NBRM. The Log Ratio (LR) test (see

Long, 1997, p- 237) compares the log-likelihood of the two models in order to determine

lOl

 

which is better (i -e., closer to zerO)- Comparing the Z-Scores for the coefficien ts between

the two models is a third method 0f testing for overdispersion When a significant
amount of overdispersion is present, the z-scores are generally smaller for the NBRM
than the PRM. Recall that when the PRM is wrongly applied the z-scores are inflated
and the Significance of the variables is exaggerated. The NBRM, on the other hand, more
precisely estimates the z-scores in overdispersed data, allowing the researcher to draw
more accurate conclusions. All three indicators of overdispersion are evident in the
models for both CP Acts and CP Time, indicating that PRM is an ill-suit ed regression
model for these data.

WW An additional issue raised
with count models is the prediction of zeros. A large number of Zero counts can facilitate
overdispersion in the data because when the dependent variable is skewed toward zero it
is not possible to achieve eclualit)’ between the mean and variance s Typically, the PRM as
well as the NBRM under-pre dict the amount of zeros in the dependent variable These
models would therefore be inefficient at predicting those officers that Spend zero minutes
per shift on community policing activities, or who provide zero acts of Conlmunily

policing per citizen encountered. Zero—inflated models for count outcomes address this
problem by modeling the predicted zeros specifically. In other words, zero.1-nfla ted
models assume that a different process occurs for officers who perform no Cominum'ty
policing compared to officers who perform some community policing. Similar to
evaluating the alpha parameter to determine whether the PRM or the NBRM is more
suited to the data, the tau parameter evaluates whether the data should be modeled With a
zero-inflated count model. prrior analyses demonstrate that the NBRM is more suited

102

 

to the data than me PRM, as was evident for both CP Time and CP Acts, then tau
indicates whether the data would be even better suited to a zero—inflated NBRM (known
as ZINE). LIMI)EP thus adds a fourth stage (ZINB) to the threefold process described
earlier (OLS, PRM, NBRM). A significant (p < .05) t-test of the tau coefficient indicates
that the ZINB is a more efficient model and better at predicting zeros in the outcome
measure compared to the standard NBRM.

In the present study, tau was significant for all the CP Time models, but was not
once significant for the CP Acts models. The ZINB is therefore a more appropriate
regression model for CP Time compared to the NBRM. This is n 0 t a surprising result
given that the majority of officers did not engage in any minutes of community policing
activity per shifi (i- e-, 165 of 262 officers had scores of zero). AS an example, results for
the CP Time additive model indicate that the ZINB predicted 162 Qf 165 actual zero
scores, whereas the NBRM predicted 158 of 165, and the PRM pre di ct ed 0 of 165. Each
successive model is better able to account for the large number of Zeros present in the

data. Similar improvements in the prediction of zero scores were Observed for all 0ftbe
C? Time models, and the tan parameter was statistically significant for all ofthe CP
Time models.
For CP Acts, on the other hand, tau was never significant and the prediCtion of
zeros never improved with the ZINE. When we consider that most officers did PI‘OVidc
at least one act of community policing per citizen encountered (i.e., only 23 of 252
officers had scores of zero), it makes intuitive sense that the NBRM is bett er suited to CF
Acts than is the 2%, In conclusion, results from LIIVEDEP indicate that NBRM is the

best suited lhodel for Cp Acts, whereas ZINB is the best suited model for CP Time.

103

 

CHAPTER 6

Table 8 presents the correlation analysis of the independent and dependent

variables, What is most notable from this table is the lack of significant findingsi only

dent
four 0f 28 relationships tested reached a level of statistical significance. E3011 depen

endent Variables.

variable is significantly (p <.01) associated with tvvo of the 14 indep
- On
Female officers engage in more community pelicing as measured by C? Tune.

. . . — Olvm a or
av erage they spent 23 minutes per shift engaged in crime prevention, 1)“)me S g
attending community meetings, While male officers spent less that] 1 0 minutes per Shili

on these activities. Indianapolis officers performed significantly less community
policing, as measured by bOlh CP Time and CP Acts, compared to St- Petersburg office
I S.
For example, IPD officers spent about eight minutes per Shin on Commum' .
13’ pollcin g
(compared to 18 minutes per shift for SPD officers) and provided two acts 0
. - 1° Community
pohcmg per citizen (compared to more than three aCtS per Citizen provided by S
PD
officers). Officer tenure was also negatively related to the number of CP Ac
tS per CltlZen
encountered. As tenure increases, the number 0f community p01icing acts per - ~
A CltlZen
decreases. Of the f ' 'ficant independent variables, then,
our srgm only one (Officer female)

increased the likelihood of community policing occum'ng — the Other thre 'gnif 1
e 51 103m y

reduced its likelihood.

104

Table 8

Bivariate Correlates of Communit Policin .

 

W per shift
’ p
Officer Characteristics
Offi cer Female 0-1 78 0.004
Officer Non-white -0.038 0.536
Officer Education -0.0l 3 0,839
Officer Tenure -0.054 0.388
CP Training Scale 0.008 0.895
Work En viron ment
Department app) -0.185 0.003
Day Shifi -0.01 0.899
CPA 0.101 0.102
Beat Problems Scale -0.03 0.626
Dept. Pro-CF Scale -0016 0.792
Socia\ Capit‘d
TrustVariab\e 0.057 0.360
CooperationFactor -0.01 0873
Group Cohesion Factor 0.024 0.697
Support Factor -0.014 0.827
N=262

 

cp Acts m
r P
0.074 - 0-234
-0052 0-4
0.075 0-228
-0.174 0.005
-0.084 0.177
0
-0218
-0104 0825
0.064 0 447
-0047 ' 67
-0069 0‘2
0.017 0.779
-0072 0.243
-0.042 0.502
0.053 0.396
\

 

Table 9 presents the correlations between the interaction terms and the tw
0

dependent measures. Trust x IPD was negatively associated with both Cp Tim
e and CP

Acts, indicating that officers who work in Indianapolis and have low levels oftru t
S tend

to engage in significantly less community policing per shift and provide significant]
Y

fewer community policing acts per citizen. Another way of interpreting this finding i
s

that the difference in community poliCing performance between IPD and SP D is a

function ofIPD officers «reporting less trust in their supervisors. For ex le almost
alhp , one

105

in four lPD officers disagreed that they trusted their sapervisor (2 1 , 5%), compared to
only 12.5% of SPD officers. This difference in trust between the departments, however,
did not reach the conventional level of statistical significance )8 (3, N = 262) = 6.04, p =
. 1 l 0. This is one indication of the iInportance of computing interaction terms: when trust
and deparunent are combined they Share a sigtnficam relationship with community

policing, even though trust levels did n0t differ to a significant degree between

departments.

. . . . ther
Trust x Tenure was negatively assoc1ated with CP Acts per Citizen- In 0

. . . ' their
words, officers vv1th a lot of time on the Job who reported low levels of “St in

of referrals to

ovided the

supervisors were less likely to engage in providing comfort, informafion’
dlfizens they encountered. Officers with less than two years on the job pr
highest mean C? Acts (2.98), While ofiicers with 1 8 year 5 01' more on the job PIOVided
the lowest mean C? Acts (1 .33). Mean CP Acts steadily decrease as Officer tenure
increases, and the difference in community policing based on t(inure categories was also
statistically significant, F (4, N = 262) = 3.212, p < .02. In short, tenure direCtIy impact
the amount of community policing performed, and this relationship is also Significanuy S
moderated by the amount of trust an officer has in his or her supervisor.
Trust x Female was positively related to GP Time, indicating that female offio
with high levels of trust tend to engage in more COIanlity POIiCing per shift. There :rs
also a significant difference between the amount 0f trust reported by male and female as
officers, x2 (3, N = 262) = 795, P < -05 - Four times as many male officers reported that

they did not trust their supervisors (201 % compared to 5.2%). About 950/ f fem l
o O a e

officers, on the other hand, reported that they trusted their supervisors Th high I
- e er evel

106

Of “390““ trust among female Officers is therefore one potential epraflatiOH for why
they engage in significantly more Community policing per shjfi than their male
counterparts.
The four significant interaction terms mirror the significant findings presented in
Table 8- We can infer from the bivariate results that trust is influencing the relationships
between Officer Female, Officer Tenure, IPD and community policing. One intereSting
feature revealed by the interaction term analyses is that Trust, rather than Cooperation,
Group Cohesion, or Support, is the social capital dimension that appears to make a

. _ , ble in
difference, albelt shght. Overall, the findings presented in Tables 8 and 9 are nota

‘ ° - - tion
the ”dc 0f Slgnlficant 1' elationsthS between the independent variables, the mterac
terms, and the dependent measures. Of the 84 biv an'ate relationships tested (42 per

dependent variable), six were statistically significant. In other Words, only one in every

14 relationships met the conventional level of significance.

107

 

 

Table 9

Bivariate Correlates of Com uni Policin — Interaction Term;

CP Time per shift CP Acts per citizen

 

 

 

 

r p r P
Trust Variable X Ofc. Female 0.190 0.002 001333 353:
Cooperation Factor X Ofc. Female -0.02 0.750 ‘0-103 0097
Group Cohesion Factor X Ofc. Female 0.00 0.904 ‘0‘011 0.863
Support Factor X Ofc. Female -0.08 0.220 0-

0 033 0.592
Trust Variable X Ofc. Minority '0.02 0780 i '055 0.376
cooperation Factor X Ofc. Minority . 0.030 0.628 0' 015 0.803
Group Cohesion Factor X Ofc. Minority -0-03 0-682 ’8‘019 0.756
Support Factor ><X Ofc. Minority 0.049 0.427 ' 0 288
66 ‘
Trust Variable >< Ofc- Education . 0.032 0.608 968090 0.1434]
Cooperation Factor X Ofc. Education . -0- ()2 0.693 $.05 0.3 9
Group Cohesion Factor X Ofc.. Education 0-033 3333 0.049 0-42
Support Factor X Ofc. Education 0.014 -
0 686 .0148 0.017
TrustVariable X Ofc. Tenure -0-03 0-701 0 046 0.461
CooperationFactor x Ofc. Tenure 0.024 0.496 ‘ . O 940
Group CohesiQnFactof X Ofc. Tenure 0.042 0‘795 0.005 '
Support Factor X Ofc. Tenure 0.016 ' 0.045 0.470
Trust Variable X CP Assignment 0.079 0203 0.094 0.129
Cooperation Factor X CP Assignment 0-053 0396 0.037 0 548
Group Cohesion Factor X CP Assignment (1032 0609 -029 0643
Support Factor X CP Assignment ‘0- 12 0051 0.102 0 '100
Trust Variable >< Dept. (IPD) '0-15 (“us 41199 0 00
Cooperation Factor X Dept (IPD) ‘0‘ 04 0'494 ‘0-044 0 '47 1
Group Cohesion Factor X Dept. (IPD) 0028 0‘656 0-008 0 89:
Support Factor X Dept. (IPD) 0005 0-936 0.039 0532
Trust Variable x Dept. Support 0016 “-791 ~0.03g 0.542
Cooperation Factor X Dept. Support Of CP ’0'002 8 '71 1 ‘0-052 0.401
Grp. Cohesion Factor X Dept. Support Of CP (3)03 0'338 0037 0550
support Factor X Dept. Support of CP ' - - 6 0.046 0.460
1 \__\
N=262
108

 

 

Multivariate Analyses

Additive Mo dels

 

Table 10 presents the findings 1mm the Additive Models. The Zero-Inflated
Negative Binomial regression model (ZINB) was used to regress officer characteristlcs,
work environment, and social capital Variables on (3}) Time per shift. The Negative

t of
Binomial regression model (NBRM) Was used to regress CP Acts on the same 56

. . - ' 11
independent variables. Once agam what 15 notable is the overall lack of statistica y

- . . . - blCS were
significant predictors of community policmg. None of the independent V3113

. lose at? 5
significant predictors of CP Time (although the departmait variable came C

. . , p Acts
.087), but the model itself is statistically significant, The regressron {of C

' - - - ' 5 receiving
produced three significant predictors that all reduce the llkellhOOd of citizen

acts of continuum}l Policing. FirSt, as officer tenure increases, the likelihood 0f CP Ads
decreases significantly (but not necessarily substantially ’ only by 3%),8 Second,
officers having a community policing assignment reduces the expeCted number of CP

0 ' ' b1 constant. Finally, bein an
Acts by 30 /o holding all other vana es g 1P D Officer decreases

the expected number of CP Acts by 47%, holding all other variables constant N
- one of
the social capital dimensions were significant predictors of either dependent var-lab]
e.

_‘_

8

‘ forming beta coefficients from Poiss -
Long (1997) prowdes a formula for tans . on or he . . .
regression into percentages for am: of interpretation (see page 228). The formula is (Elana/C 3121;321:111
The formula was computed for all sigmficant (p <.05) coefficrents. 0 [C P - D.

109

Table l()

Additive Models for Communi POIiCiDO Variables.

 

 

 

 

 

CP Time ZINB M 11518]
1—3- Si P 3. E 3
Constant 2.49 1.5 7 0_1—1 3 2.23 0.58 0000
Officer Characteristics 072
Officer Female 0- 56 0.34 0,105 0.27 0.15 0-334
Officer Non-white 0.05 0.42 0,901 -0.14 0.15 3-442
Officer Education 0.05 0.10 0.587 0.03 0'04 0'004
Otficer Tenure 0.01 0.03 0.721 -0.03 0'01 ' 8
CP Trarmng Scale 0.03 0.04 0.456 0.00 -
Work Environ ment 0 16 0.000
Department (IPD) -0.55 0.32 0.087 -0.63 0' 14 0989
Day Shin 0.15 0.35 0.675 0.00 0'18 0.038
C? Assignment —o.04 0.41 0.922 -o.36 0’02 0,480
BeatProblems Scale -o.01 0.06 0.816 -001 0' 02 0.581
Dept.Pro-CP Scale 0.01 0.05 0.826 -001 '
Socia\ Ca “3‘
Trust Variable 0.07 0.32 0.829 -012 0.13 0.388
CooperationFactor -0.08 0.18 - 0.642 ~0.07 0.05 0.196
Group Cohesion Factor 0.19 0.19 0.299 ~O. 02 0 07 0 826
_ 4 ' ‘
Support Factor 0.27 0.26 0.29 0.06 0.13 0.213
Alpha 1.09 0.22 0.000 0.27 O
Tau 0.13 0.04 0.002 -07 0.000
Model Fit
Log-L 693.33 108-L -503.52
Vuong 10-91 Chi-sq. 52.86
DP“ 2 Dt“ 1
Sig. Level 0000 Sig. Level 0.000
N=262

 

 

Notes: * Compared to the Poisson regression model.

Alpha compares Poisson to Negative Binomial (significance indicates a better fit ofth

compares the NB model with the Zero-Inflated NB model (significance indicates a be e NB model). Tau
model). ZINB regression models were tested on both dependent measures, Tau indi “er fit 0f the ZINB
always better than NB for CP Time, but not once improved the fit for CP ACtS. A V Gated that ZINB Was
1.96 favors the ZINB over the NB model. “Orig stat1snc less than

110

 

Interactive Models

 

Table 11 presents the results of the interactive models for CP Time and Table 12,

presents the results for CP Acts. Interaction terms (created with the social capital
dimensions and various officer Characteristic and Work environment variables) were
included along with the variables from the additive models in order to assess the extent to
WhiCh they might enhance our eXplanation of community policing performance. AS these

tables make readily apparent, the interaction terms do not contribute much explanatory
power to the originaI models. Only one of the 56 interaction coefficients (28 per
dependent meas ure) Was a significant predictor of community policing-

. . 1 Su Po“ Of
The Significant interaction term, Group Cohesion Factor x Bellman p

Comurutv Policing, reduces the likelihood of CP Time per shift by 1% Officers Who
Perceived they were part of a cohesive group, and who also Perceived a high degree Of
departmental support of community policing, spent significantly leSS time per Shlfi
engaged in community policing. Qne of the original variables Was 3130 Significant in this
model: Group Cohesion. Officers who scored high on the group cohesion factor were
more likely to spend time engaged in community policing; they increased t1) e likelih
of community policing by 239%, holding all other variables conStant. Together the 00d
variables indicate that group cohesion will increase the likelihood of CP Time, b1“ n:
when a high degree of departmental support of community Policing is also present.
Perhaps group cohesion is only effeCtiV’e at increasing officer Productivity precisely When
departmental support is lacking. Under these circumstances, officers who want to
practice community policing may more heavily rely on support from their peers or
workgroup because they cannot rely on similar support from

th
e department. Overall, it
1 1 1

 

is important to place this finding in the context of a general pattem which suggests that
social capital, whether in interaCtion form or not, does not help us understand or predict

officer performance of community POIiCil'lg.

Additionally, the Likelihood Ratio test indicated that the interaction model With
the significant term did not PTOVide Significantly more information compared to the
. . . . ' t
additive model. One way to judge the lmPOI'tance of a significant interaction term 15 0
. . does
determine whether the model including the interaction term fits the data better than
. . . __, - 93 .33 +
the Original model without the interaction term. The Log Ratio Test [LR " 2 ( 5
ed the critical
589.44) = 7.88 3 indicated that the obtained chi-square (7.78) did not ex“
. . . . \) so moludmg
chi-square (9-49, Wlth 4 degrees of freedom at the -05 Significance leVe ’

' - . , - udin mem-
the interaction terms did not prOVide Significantly more mformatl on than excl g

in other WOYdS, the two models were statistically indistinguishab le, despite the presence

of one significant interaction term, This casts further doubt on the iInpo l cc of this
finding. In conclusion, the results from Tables 1 1 and 12

Should be ConSl'dered notable
for their uniformity of effect: nil.

112

 

 

Tame 11

Interaction Models - Zero—Inflated Negative Binomial Regression on Community

olicino Time-

_P_,..——-——--=-——-——

Sex

Tenure

Race

Education

Constant
Trust Variable X Ofc. Female
Cooperation Factor X Ofc. Female
Group Cohesion Factor X Ofc. Female
Support Factor X Ofc. Female
Alpha
Tau
Log-Likelihood/V uong/ Si g. Level

Constant

Trust Variable X Ofc. Minority
Cooperation Factor X Ofc. Minority
Group Cohesion Factor X Ofc. Minority
Support Factor X Ofc. Minority

Alpha

Tau

Lo g-Likelihood/V uong/ Si g. Level

Constant

Trust Variable X Ofc. Education
Cooperation Factor X Ofc. Education
Group Cohesion Factor X Ofc. Education
Support Factor X Ofc. Education

Alpha

Tau

Log-Likelihood/V uong/ Si g. Level

Constant

Trust Variable X Ofc. Tenure
Cooperation Factor X Ofc. Tenure
Group Cohesion Factor X Ofc. Tenure
Support Factor X Ofc. Tenure

Alpha

Tau

Log-Likelihood/V uong/ Si g. Level

113

CP Time per shift

 

B
2.86
0.40
0.28
0.17
-0.71
1.03
0.14

-591.48

2.82
0.15

0.16
-0.66
0.72

1.01

0.13

-589.79

-2.53
-0.37
0.05
0.00
0.24
1.02
0.14

-591.43

2.83
0.03
0.02
0.03
0.00
1.11
0.12
-592.6

_S_E
1.60
1.19
0.60
0.46
1.38
0.21
0.04
10.91

1.66
1.54
0.60
0.39
0.97
0.20
0.04
11.25

3.59
0.24
0.14
0.14
0.23
0.20
0.04
11.14

3.27
0.10
0.04
0.05
0.08
0.23
0.04
10.05

2
0.073
0.735
0.637
0.711
0.604
0.000
0.001
0.000

0.089
0.921
0.795
0.089
0.455
0.000
0.002
0.000

0.480
0.125
0.740
0.995
0.294
0.000
0.001
0.000

0.387
0.748
0.656
0.598
0.968
0.000
0.005
0.000

 

 

 

 

 

 

Table 11 (cont’d),

 

CP. Constant 2.51 1.15 0.249
ASSIgn ment Female“ 0.77 0.37 0.037
Trust Variable X CP Assignment -0.02 0.75 0.977
Cooperation Factor X CP Assignment 0.43 0.44 0.322
Group Cohesion Factor X CP Assignment -0.07 0.38 0.857
Support Factor X CP Assignment -0.67 0.63 0.289
Alpha 1 .04 0.22 0.000
Tau 0.12 0.04 0.006
Log-Likelihood/V uong/ Si g. Level -5 89.61 10.83 0.000
DeP al’tment Constant 1.82 2.12 0.392
Trust Variable X Dept. (IPD) -0.5 0.81 0.534
C00peration Factor X Dept. (IPD) -0.53 0.42 0.211
Group Cohesion Factor X Dept. (IPD) 0.40 0.42 0.340
Support Factor X Dept. (IPD) 0.37 0.64 0.566
Alpha 1.02 0.21 0.000
Tau 0.14 0.04 0.001
Log-Likelihood/V uong/ Sig. Level -S90.63 10.97 0.000
Dept. Support Constant -0.51 3.70 0.891
of CP Group Cohesion Factor 1.22 0.54 0.024
Trust Variable X Dept. Support -0.11 0.11 0.320
Cooperation Factor X Dept. Support 0.03 0.71 0.712
Group Cohesion Factor X Dept. Support -0.12 0.06 0.050
Support Factor X Dept. Support 0.14 0.11 0.197
Alpha 1 .00 0.20 0.000
Tau 0.13 0.04 0.002
Log-Likelihood/V uong/ Sig. Level -5 89.44 1 1.34 0.000
N=262

N otes: *Each interaction model was analyzed with the variables included in the additive model for CP
Tune, but only those variables that reached statistical significance are presented in the table.
** This Variable also came close to attaining statistical significance in the following models: Race (p =.06),
Department (p =07), Dept. Support of CP (p =.07).

114

Table 12

Interaction Models — Negative Binomial Regression on Community Policing Acts.*

Sex

Race

Education

Tenure

Constant

Trust Variable X Ofc. Female
Cooperation Factor X Ofc. Female
Group Cohesion Factor X Ofc. Female
Support Factor X Ofc. Female

Alpha
Log-Likelihood/Chi-Squared/Sig. Level

Constant

Trust Variable X Ofc. Minority
Cooperation Factor X Ofc. Minority
Group Cohesion Factor X Ofc. Minority
Support Factor X Ofc. Minority

Alpha
Log-Likelihood/Chi-Squared/Sig. Level

Constant

Trust Variable X Ofc. Education
Cooperation Factor X Ofc. Education
Group Cohesion Factor X Ofc. Education
Support Factor X Ofc. Education

Alpha

Log-Likelihood/Chi-Squared/Sig. Level

Constant

Trust Variable X Ofc. Tenure
Cooperation Factor X Ofc. Tenure
Group Cohesion Factor X Ofc. Tenure
Support Factor X Ofc. Tenure

Alpha
Log-Likelihood/Chi-Squared/Sig. Level

115

B
2.33
0.14
-0.04
-0.03
0.00
0.27

-503.24

2.68
0.51
0.19
-0.13
-0.4
0.26
-500.47

2.01
-0.02
-0.01
-0.02
0.53
0.27
-503.08

2.77
0.02
0.00
0.00
-0.01
0.27
—502.86

S_E
0.61
0.40
0.20
0.18
0.37
0.07

51.84

0.60
0.40
0.15
0.16
0.32
0.07
48.41

1.66

0.10

0.04

0.05

0.10

0.07
52.79

0.87
0.02
0.01

0.01

0.02
0.07
52.31

2
0.000
0.718
0.818
0.861
0.987
0.000
0.000

0.000
0.202
0.205
0.419
0.210
0.000
0.000

0.226
0.862
0.770
0.696
0.957
0.000
0.000

0.001
0.465
0.637
0.802
0.609
0.000 .
0.000

Table 12 (cont’d).

 

CP Constant 2.80 0.68 0.000
Assignment Trust Variable X CP Assignment 0.37 0.28 0.188
Cooperation Factor X CP Assignment 0.12 0.12 0.324
Group Cohesion Factor X CP Assignment 0.01 0.15 0.941
Support Factor X CP Assignment -0.15 0.26 0.558
Alpha 0.26 0.07 0.000
Log-Likelihood/Chi-Squared/Sig. Level -500.64 47.50 0.000
Dept. Constant 2.02 0.77 0.008
Trust Variable X Dept. (IPD) -0.13 0.27 0.636
Cooperation Factor X Dept. (IPD) -0.1 0.11 0.341
Group Cohesion Factor X Dept. (IPD) 0.06 0.14 0.666
Support Factor X Dept. (IPD) , 0.15 0.26 0.552
Alpha 0.27 0.07 0.000
Lo g-Likelihood/ Chi-Squared/ Si g. Level -502.77 51.25 0.000
Dept. Support Constant 2.98 1.55 0.055
of CP Trust Variable X Dept. Support 0.02 0.05 0.686
Cooperation Factor X Dept. Support 0.01 0.02 0.657
Group Cohesion Factor X Dept. Support 0.01 0.02 0.554
Support Factor X Dept. Support 0.01 0.05 0.870
Alpha 0.26 0.07 0.000
Log-Likelihood/Chi-Squared/ Sig. Level -501.87 51.00 0.000
N=262

Notes: * Each interaction model was analyzed with the variables included in the additive models.

** Significant coefficients from the additive model (tenure, CPA, IPD) also had the same
sign and significance in these models, with the exception of tenure in the tenure interaction
model, CPA in the CPA interaction model, and IPD in the department interaction model.

116

Summary of Findings
Table 13 presents a summary of the statistically significant coefficients produced
by the bivariate and multivariate analyses.

Table 13

Summm of Findings.

 

 

 

 

 

 

 

 

 

 

 

 

 

Variable Bivariate Multivariate

Female Officer + CP Time
Female Officer X Trust + CP Time
IPD - CP Time

- CP Acts - CP Acts
IPD X Trust ‘— CP Time

" CP Acts
Tenure - CP Acts - CP Acts
Tenure X Trust - CP Acts
CP Assignment - CP Acts
Group Cohesion Factor + CP Time
Group Cohesion Factor X - CP Time
Dept. Support of CP

 

 

What should be given primary consideration from the results presented is the
overall lack of significant findings produced by the bivariate analyses, the additive
models, the interactive models, and the analyses of district effects. Of the 352 regression
coefficients produced in these analyses, only about 5% reached the conventional level of
significance. The findings related to social capital and community policing are notable in
their consistent pattern of unimportance. Of the three categories of independent variables

(officer characteristics, work environment, and social capital), the social capital group by

117

far offered the least explanatory power. The most consistent significant findings
produced in the analyses were contributed by officer characteristics and features of their
work environments. Moreover, the variables that were significant tended to reduce,
rather than increase, the expected amount of community policing time per shifi or the
number of community policing acts per citizen performed by the officers.

The most consistent significant result (both statistically and substantively) was
the organizational environment in which the officer worked. This was originally
conceptualized at the department level. Officers working in Indianapolis were
consistently found to produce fewer community policing minutes per shift and acts per
citizen compared to officers working in St. Petersburg. The substantial difference in
community policing performance between departments provides a strong indication of
the importance that organizational factors play in the likelihood of community policing
being performed.

The results of these analyses also reveal that different conceptualizations of
community policing can lead to different results. The additive model for community
policing time did not produce any significant results, but female officers spent
significantly more time on community policing compared to their male counterparts in
one of the interaction models. Additionally, another interaction model revealed that as
officers’ perceptions of group cohesion increased so did the amount of time they engaged

in community policing activities.9 For the other dependent variable, the additive model

 

9

While there was one significant interaction term for a CP Time model (group cohesion X dept. support of
CP), recall that the LR test showed that the interaction model did not provide significantly more
information than did the original model. The significance of interaction effects should be judged in terms
of the entire research process; the general pattern of non-significance makes this finding especially dubious.

118

produced several significant findings while there were no significant findings for the
interaction models. Officer tenure and CPA consistently reduced the expected number of
CP Acts per citizen, holding other variables constant. A strong department effect also
emerged: officers in Indianapolis were significantly less likely to provide acts of comfort,
referrals, and information to citizens compared to officers in St. Petersburg. The next
chapter puts these findings in context of the existing literature on social capital,
community policing, and police organizations. Implications for promoting community

policing activities across disparate organizational environments are discussed.

119

CHAPTER 7

ISSUES RAISED BY THE RESEARCH FINDINGS

The goal of this chapter is to discuss the issues raised by the present study. These
are grouped into two sections. The first section of this chapter describes the substantive
issues raised by the findings from the statistical analyses. The second section provides an
overview of the methodological issues relevant to the current study as well as directions
for future research investigating officer performance of community policing within a
social capital framework.

Substantive Issues
Research Questions Revisited

Let us first examine how the results from the statistical analyses provide answers
to the research questions presented in the introduction. The first question, “Is social
capital related to how police officers perform their jobs?” and its follow-up question,
“Specifically, what is the relationship between levels of social capital and officers’
engagement in community policing activities?” can be answered by referring to the
bivariate results, as they provide an indication of whether a significant direct relationship
is present. The multivariate results are used to answer the remaining research questions,
as they deal with the social capital-community policing relationship in the context of
other moderating variables.

Bivariate results did not indicate a direct relationship between social capital and

community policing. However, trust was revealed as an important dimension of social

120

 

 

capital through the presence of several interaction terms that were significantly related to
the measures of community policing. Specifically, the relationship between trust and
community policing was moderated by gender, department, and tenure. Levels of trust
significantly increased community policing time per shift for officers who are female.
This means that trust in supervisors is more relevant to female officers’ performance of
community policing. Male officers engaged in community policing regardless of their
levels of trust.

Levels of trust were negatively related to both community policing time and
community policing acts for officers working in Indianapolis. In other words, trust
influenced officers’ performance of community policing in [PD but not SPD. The
direction of the relationship suggests that trust makes community policing in Indianapolis
less likely. This could be viewed as a negative outcome of social capital — perhaps
having trust in their supervisors allowed officers to “get away” with performing less
community policing. A similar relationship emerged for tenure and trust. Levels of trust
decreased acts of community policing per citizen for experienced officers. When officers
had a lot of time on the job, they tended to have higher levels of trust in their supervisors,
and they used this trust to more easily circumvent the community policing mandate. The
likelihood that rookies would engage in community policing, on the other hand, was
unaffected by the amount of trust they had in their supervisors.

These results demonstrate that of the four social capital dimensions, trust played
an important role at the bivariate level. It affected community policing performance in
the expected direction only for female officers: high levels of trust promoted rather than
hindered their engagement in community policing. For officers in Indianapolis and with

121

 

high tenure, however, high levels of trust decreased the likelihood they would engage in
community policing.
The second research question addressed by the current research, “What is the
relative contribution of police social capital in a model that also includes characteristics
of the individual officer and their work environment?” can be answered in a
straightforward manner: none. The quality of officers’ relationships with their peers and
supervisors did not influence whether officers spent time on community policing or
provided community policing acts to citizens, controlling for officer characteristics and
features of their work environments. Given the wealth of literature pointing to the
potential importance of social capital in understanding police behavior, how can the
current results be explained? Aside from any methodological limitations that may have
contributed to the null findings (discussed in the next section), why would levels of trust,
cooperation, group cohesion, and social support among police not matter to community
policing performance?
One explanation is that the relationships that are really important to officers
Wanting to engage in community policing are not police relationships, but rather citizen
relationships. This study did not provide information on the extent to which officers
Were networked into relationships in the community. The four social capital dimensions
OF trust, support, cooperation, and group cohesion could be viewed as especially
irrlDortant elements of relationships between officers and citizens. Given that the central
tel-1 %t of the community policing philosophy is that police and citizens should work
to g ether to reduce crime and increase safety in communities, describing the qualities of

the: Se relationships could provide an important explanation of community policing

122

performance. Future researchers should consider assessing community policing
performance in terms of networks of relationships within police organizations, within
other relevant agencies, and within the citizenry, as well as relationships that reach across
these different groups.

Another way to interpret the finding that “social capital does not matter” is in
positive terms. Some might argue that police performance should not be dependent upon
levels of police social capital. In other words, police should engage in community
policing regardless of whether they have relationships with their peers and supervisors
that are rich in trust, cooperation, support and/or group cohesion. Police officers should
do their jobs no matter what their level of resources. This interpretation relies on a model
of policing that is individualistic and self-determined; officers’ performance is viewed as
based solely on their own will, motivation, and determination. It removes consideration

of environmental characteristics which have been shown to play an important role in
behavioral outcomes among officers.

The third question, “Do officer characteristics, such as their sex, race, education,

of tenure moderate the relationship between social capital and community policing?” can
b e answered by looking at the impact of the OC X SC interaction terms on the regression
m QCiels of community policing time and community policing acts. No significant
intéraction terms were revealed by the analyses. So while officer characteristics might
Sigl'iificantly moderate the relationship between social capital and community policing at
the bivariate level, once other relevant factors are controlled these effects disappear.
Th§re was one interesting direct relationship among these variables: as officer tenure

Incl-eased, the amount of community policing acts per citizen decreased. This replicates

123

the finding of most other research indicating that as years of experience increase,
productivity decreases (Bittner, 1990; DeJong, Mastrofski, & Parks, 2001; Muir, 1977;
Roberg, Crank & Kuykendall, 2000; Stalans & Firm, 1995).

The fourth question, “Do features of officers’ work environment, such as their
department and their perceptions of the department’s support for community policing, .
moderate the relationship between social capital and community policing?” is answered
by referring to the Work Environment X Social Capital interaction terms in the regression
models for community policing time and community policing acts. Recall that one social
capital interaction term significantly decreased the amount of community policing time
per shift: Group Cohesion X Department Support of Community Policing. This was
interpreted as indicating that a high level of group cohesion decreased time spent in
community policing when a high degree of departmental support of community policing

was also present. The importance of this finding is doubtful given the general pattern of
non-importance of social capital, and the Log Ratio test that revealed that the interaction

model did not provide significantly more information than the additive model (i.e., the

interaction term did not sigrificantly increase, in a statistical sense, our understanding of

Community policing time).

Two other direct relationships are also revealed by this group of variables:

CO Immunity Policing Assignment and Department. First, officers assigned as community

po 1 icing specialists provided significantly fewer acts of community policing per citizen

th Ext-:1 did regular patrol officers. This finding is inconsistent with original expectations

that community policing officers would provide more community policing acts. It does,

hQ‘OVever, replicate another analysis of data from the POPN that found that community

124

policing officers in both sites spent less time in encounters with citizens than did patrol
generalists (Parks, et al. 1999). These officers had more discretion than general patrol
officers, and they used it to engage in less “face time” with the public, or to spend more
time with citizens of higher status. This study generates a similar conclusion about how
community policing and patrol officers tend to do their jobs. Community policing
officers provided significantly fewer acts of community policing to citizens, and
regarding community policing time per shift there was no difference between the two
groups of officers.

The fourth research question does draw our attention to the most consistent
finding produced by the current study: the importance of officers’ work environment.
Specifically, the department where the officer works exerts a substantial impact on the
expected amount of community policing time and community policing acts performed by

officers, controlling for the effects of the other independent variables. To state it bluntly,
officers in Indianapolis were far less likely to engage in community policing time or acts
Compared to their counterparts working in St. Petersburg. What organizational factors
relevant to community policing can account for such a pronounced difference between
the practice of community policing in these two departments? The community policing
1i tel-attire offers several explanations as to organizational variables that may affect officer
p erformance of community policing. These are discussed below.
W
“Providing leadership and vision is an important part of any organizational
Change strategy” (Skogan & Hartnett, 1997, p. 91). Top police administrators are
S‘Z‘IDposed to communicate the department’s philosophy, mission statement, goals,

125

policies, and strategies to officers. They can provide leadership as to what activities are
encouraged within the department, as well as the activities that are discouraged.
Leadership is considered by some scholars to be especially important in the community
policing era, as leaders must effectively convey what community policing is, how
officers should practice it, and how the organization will provide the necessary support to
accomplish it. Leaders can also convey values and beliefs that they feel will increase
efficiency and productivity within a community policing context. In one study that used
social capital as a framework for understanding community policing partnerships, the
failure of the community policing program was attributed in part to, “a lack of proper
leadership in the police department to promote and enforce norms of trust, reciprocity,
and co-production” (Pino, 2001, p. 213).
Chiefs in both Indianapolis and St. Petersburg were hired due to their support and
promotion of a community policing philosophy, but they varied in how they translated
this philosophy into practice (see DeJong, Mastrofski, & Parks, 2001; Parks, et al. 1999).
In other words, the “vision” of community policing is substantially different for the two
Clniefs. In Indianapolis, the chief encouraged officers to engage in community policing
Vi a an aggressive order maintenance response. In other words, the leadership he
p1‘<)vided facilitated an increase use of traditional police tactics (e.g., stops, arrests,
Seaqches and seizures) in an attempt to increase residents’ feelings of safety. The
up Mership” element of community policing was accomplished primarily at the district
13"” (:1, with staff members attending community meetings. Officer-level engagement of
th% community was not encouraged. The small proportion of officers with specific

c0Immunity policing assignments were known as “Crime Bill” officers; they were

126

supposed to work together on community policing projects. Collaboration with
community groups or patrol officers was not emphasized. In short, community policing
efforts in Indianapolis were “compartmentalized” as the responsibility of a few
organizational members.

Alternatively, the style of community policing exhorted by the chief in St.
Petersburg focused on problem-solving. In fact, he had gained an international
reputation for the geographic deployment of officers to enhance their ability to engage
the community. In contrast to Indianapolis, community partnerships were encouraged at
the officer-level rather than at the district-level. Community policing officers were
supposed to work with patrol officers as a team to problem-solve in their assigned areas.
The chief emphasized that community policing was a responsibility of all the officers in
the department, not just those with special community policing assignments. In short,
community policing efforts in St. Petersburg were integrated into the responsibilities of
all organizational members. In addition, the Chief changed the performance appraisals of
all officers to reflect the new emphasis on community policing.

It should be noted that there are limits to what leadership can accomplish. A

nation-wide survey of police administrators found that 98% agreed that community

p0 licing was a worthwhile reform effort, but 47% admitted that what community policing

ac: t‘ually meant in practical terms was not clear (Wycoff, 1994). Perhaps most troubling,

on ly 27% felt that implementing community policing would require extensive

OI‘ganizational change, for example to policies, goals or training. Under these

circumstances, Mastrofski (1998) cautions that “police agency leadership is not a driving

f0 rce” for accomplishing organizational change; rather, successfirl long-term change in

127

policing usually results from leaders recognizing and “riding the wave” of broader
demographic, economic, social, and technological forces (p. 183).
Empirical research using data fiom the POPN also reveals the limits of
leadership. DeJong, Matrofski, and Parks (2001) concluded that leadership does not play
an important role in implementing new programs because officers’ belief systems (i.e.,
their acceptance of the community policing philosophy) were not related to the amount of
time they spent on problem—solving activities. To increase the amount of time spent on
problem-solving activities, the authors recommend assigning officers to special units that
emphasize this type of activity and where there is time to engage in these activities.
Their study is consistent with much police research finding that situational or
organizational factors are much more relevant determinants of officer behavior than are
attitudes or beliefs. As Trojanowicz et al. (1998) note, “administrators may expect only a
limited amount of problem solving to occur by decree” (p. 188). Leadership must be
coupled with the structural changes needed to support officer engagement in community
policing. Some of these relevant changes are discussed below.
_Organizational Structure
Geogaphic Responsibilig. “For community policing to be successful there must
be some level of geographic permanence” (Troj anowicz, Kappeler, & Gaines, 2002,
p. I 3). Geographic permanence promotes ownership and responsibility among police for
wh at happens “on their beat.” The community policing philosophy dictates that officers
ShQ ‘uld be integrated into the community, and this is best achieved by having them
pe31‘1-nanently assigned to a particular area. In St. Petersburg, community policing officers
W'erked with the general patrol officers in their assigned beats. The combined strategy of

128

geographic permanence and having all officers work together might be one explanation
as to why community policing performance was more likely to occur in SPD.

On the other hand, in Indianapolis, community policing officers were supposed to
work together to accomplish community policing goals. In effect, this meant that their
geographic responsibility covered the entire city. Goldstein’s (1990) sums up the
limitation of such a strategy with his statement that “... so much of policing consists of
dealing with problems. And while some problems can be viewed as citywide and
relatively uniform wherever they occur, most have a local character to them or may even
be unique to a specific beat. It requires officers close to a community to identify them
and to deal with them” (p. 160). Coupled with the fact that officers in Indianapolis were
not encouraged to work with the general patrol officers assigned to particular areas, it is
not surprising that their levels of community policing performance were lower than in St.
Petersburg.

Decentralization. This strategy assumes that community policing will be best

accomplished when officers work in an organization that is not controlled centrally, but
rather decentralized to enable variation in policing styles and strategies based on the
Characteristics and needs of different neighborhoods within a department’s jurisdiction.
D epartments serious about community policing therefore push responsibility and
au tlrority down the organizational hierarchy rather than keeping it at headquarters. This
re 8 tructuring is expected to enhance officer effectiveness because they are freed from
rigi d, standardized operating procedures and given the flexibility to create custom plans
to ='=‘I.ddress specific problems in their assigned beats. Decentralization empowers officers
to use their discretion creatively without having their activities dictated to them by upper-

129

management.

Attempts to flatten the organizational structure to facilitate community policing
took different forms in the two departments. In Indianapolis, community policing tasks
were decentralized to the district level. This meant that district commanders were
responsible for setting community policing goals and tasks and overseeing community
policing projects occurring within their districts. To a certain extent then, officers still
had their activities and priorities set for them by a member of management. St.
Petersburg more fully realized decentralization because community policing was
decentralized to the officer level. This meant that individual officers would implement
and develop community policing projects with the citizens they encountered on a daily
basis. They were trusted to use their discretion appropriately to determine the types of
community policing activities in which to engage. In terms of designing an
organizational structure that facilitated officers engaging in community policing,
therefore, SPD was more successfirl than IPD.

Methodological Issues

Several methodological issues were raised by this research, including the

inelusion of interaction terms in the study of social capital, the revelation that different

regression equations are needed for modeling community policing time and community

po 1 icing acts, the problem of establishing causality, the sampling strategy, and the

“1% asurement of social capital and community policing. They are discussed in the

83% tions below.

t eraction Terms

\—

Interaction terms are an important area for future researchers to explore. While

130

the present study did not reveal any significant interaction effects, it should be noted that
this was more than likely due to the multicollinearity present in most of the interaction
models. Recall that multicollinearity does not bias the regression results, but makes
finding significant effects more difficult because the confidence intervals are increased
and the t-statistics tend to be small. Significant interaction effects in the data, therefore,
could have been masked by multicollinearity. Given that the interaction models tested in
the current study were structured according to theoretical considerations, future
researchers may want to test social capital interaction terms on larger sample sizes. This
may enable enough statistical power to reveal significant effects.

Friedrich (1982) reminds us that there are many beneficial reasons for testing
interaction effects. Most importantly, they are more accurate and detailed descriptions of
relationships that exist in social science data. In other words, including interaction
effects allows the researcher avoid falling into the trap of oversirnplifying what is by all
accounts an extremely complex social reality. Assuming additive effects is the most
common oversirnplification of the social world. Instances of an independent variable
always having the same effect on a dependent variable, regardless of the levels of the
other independent variables, are surely more rare than instances where these relationships
are conditional. As Friedrich (1982) notes, conditional relationships may be “less than
wholly satisfying” given our predisposition to favor simple, consistent relationships, but
they are “often an accurate reflection of social reality” (p. 832). Interactive models are
more likely to represent the effect of social capital on any outcome under investigation,

and should be included in future studies of both social capital and community policing.

131

Regession Equations
That the Zero-Inflated Negative Binomial model provided the best fit to the

community policing time models indicates that a different process is occurring for those
officers who spend no time on community policing compared to those officers that spend
some time on community policing. In contrast, no such distinction occurs for the
community policing acts models, as the fit of these models to the data did not improve
from zero alteration. How can this difference be explained?

Recall that community policing time was derived from activity-level data,
including problem-solving, crime prevention, and attending community meetings. These
activities are under the purview of individual officers; they could choose whether and
when to engage in them. In this way they can be considered proactive activities, and
evidently many officers used their discretion to avoid engaging in them. Community
policing acts (providing comfort, referrals, or information), however, require the
presence of a citizen. This implies that officers were responding to calls for assistance,
or engaging in some form of reactive policing. Community policing acts could be
considered less under officer control compared to community policing time. This is a
fundamental difference between the two dependent measures that could explain why the
different regression equations were necessary. Officers were ultimately able to choose
whether to spend time on community policing, whereas providing community policing to
citizens was at least in part a function of whether citizens were present and needing
assistance. Future research on community policing should take into account that
different equations are needed to model different conceptualizations of community
policing.

132

Causal Order

The major problem with cross-sectional research, such as the current study, is that
the researcher is unable to definitively specify the temporal order of variables, which is a
major barrier to drawing causal inferences. Many researchers use theory or intuition to
specify which variables are likely to come first, or predict the dependent variable in a
regression equation. But these specifications may be faulty, or at least open to debate if
they are tested using cross-sectional data. In the field of social capital, for example,
researchers have expressed a concern related to social capital and how to “separate what
it is from what it does” (Edwards & Foley, 1997, p. 669) because “equating social capital
with the resources acquired through it can easily lead to tautological statements” (Portes,
1998, p. 5). The model tested in this research conceptualized social capital as a predictor
of community policing, the idea being that officers with high levels of social capital
would be more productive than officers who did not have this resource to draw upon to
“get things done.” In light of the findings, however, it might be productive to reconsider
this conceptualization. The relationship between social capital and community policing,
if one exists, may be more complex. For example, a feedback-loop arrangement could
exist where social capital and community policing are mutually reinforcing. That is, an
increase in one leads to an increase in the other, and vice versa, and the cycle continues.
The literature on social capital and educational outcomes has demonstrated that this is a
distinct possibility.

Alternatively, we might conceptualize the relationship as having the reverse
causal order: community policing performance could be influencing levels of police

social capital. This model is feasible if we consider the possibility that productivity may

133

increase social capital. For example, it makes sense to think of officers who frequently
engage in problem-solving, crime prevention, or attending community meetings as being
able to increase their social capital because they are involved in projects that put them
into contact with people with whom they may eventually form quality relationships.
They may also be forced to share information and cooperate with other officers in order
to successfully complete many community policing projects; this could also increase
their social capital. Similarly, officers who frequently provide comfort, referrals, and
information to citizens may be expected to have higher levels of social capital compared
to officers who do not frequently engage the citizenry. These examples suggest that
future researchers may want to carefully consider the causal order of the social capital
and outcome constructs. However, it must be restated that the body of literature
consulted for this research suggested that it was entirely appropriate to conceptualize
social capital as a predictor, rather than an outcome, of community policing.
Sampling Strategy

One significant limitation of the present study was that it relied on secondary
analysis of data that were collected using a sampling strategy based on rides instead of
officers. Consequently officers included in the sample had varying levels of observation
— some officers were observed for many rides while others were only observed once.
Officers who were not observed at all (but who participated in the interviews) were
excluded from the present study. Future researchers would be advised to carefully
consider the implications of having a sampling fiamework that is not consistent with the
goal of the study or the unit of analysis. The present study may have yielded different
results had the observational data been collected according to an officer-level sampling

134

strategy, ideally with each officer being observed multiple times over an extended period
(say, six months or one year). This would be a much more effective way for gather
information about specific officers and the relative impact of social capital on their
community policing performance.

Measurement of Communig Policing

Community policing time per shift and acts per citizen represent less-than-perfect
measures of community policing performance. While the goal was to create
comprehensive measures that included all the activities relevant to community-oriented
policing, it cannot be assumed that every community-oriented activity is captured by the
data. Despite this limitation, the dependent variables in this study included a broad set of
activities, all of which are guided by the community policing philosophy, representing
the themes of community engagement, problem-solving, and providing assistance to
citizens. The use of two different indicators (one based on time, one based on acts per
citizen) also advances the study of community policing as it draws attention to the
different ways that the philosophy is translated into practice.

Future researchers may want to consider the effect of combining different
community policing indicators into global measures of community policing. It might be
more instructive to model the impact of social capital on each of the community policing
tasks separately to avoid masking different relationships across these tasks. For example,
it is reasonable to suppose that police social capital might matter more to officers
engaged in problem-solving efforts than to officers attending community meetings. The
former is often a group activity where personal relationships could make a difference

whereas the latter is usually conducted at the officer’s discretion. Similarly, police social

135

capital could impact the proclivity of officers to provide information or referrals to
citizens as they may draw upon their work relationships for information relevant to
citizen needs. Providing comfort, however, is more than likely the result of either
necessity or an individual officer’s belief that comforting citizens is important rather than
the quality officers’ relationships with their peers or supervisors. Modeling these
activities separately could shed light on the different ways that social capital might
impact officer performance.
Measurement of Social Capital

The current study makes several improvements in the measurement of the social
capital construct. While the social capital research has been plagued by inadequate
operationalization of this complex construct, the measures used in the present study are
both more specific and more comprehensive than those previously used in social capital
research. Four social capital dimensions are specified: level of trust, cooperative
exchanges, group cohesion, and social support. Previous social capital research typically
uses only one or two variables representing one part of what has come to be known as
“social capital,” whereas this study advances the notion that this construct is
multidimensional, and uses multiple items to measure these dimensions.

Despite the advances the current study makes over previous social capital
research, there are limitations to the measures of social capital. The one that most clearly
stands out is the variable used to measure trust. Ideally several measures would be used
to tap into thisidimension. In addition, it would be beneficial to have questions assessing
officers’ trust for each person in their work unit, as well as their supervisors. The same
can be said for the social support dimension. Although a reliable factor, it only reveals

136

the level of support an officer receives from his or her supervisor. The level of support
an officer perceives having from his or her peer group would also be an important
element of social capital that is missing fiom this measure. The other two social capital
dimensions (cooperative exchanges and group cohesion) have multiple measures that
assess these areas in terms of both the officers’ peers and supervisors, but neither scale
reached the conventional standard of reliability. Given the limitations of these key
theoretical variables, the lack of significant findings is not surprising.

Future research on police social capital should attempt to incorporate information
on both the quantity and quality of both peer and supervisor relationships. An in—depth
examination into one specific work group, including both qualitative and quantitative
data collection, would aid our understanding of how and why police social capital is
related to the performance of various policing activities. Officers could be asked specific
questions about their peer and supervisor relationships, how these relationships help or
hinder their performance, and their perceptions of the dimensions in these relationships
that constitute the most important source of social capital (trust, cooperation, cohesion, or
support). Additionally, observations of police-peer and police-supervisor interactions
would constitute a valuable source of data that would reveal the formation and utilization

of police social capital within this unique work environment.

137

APPENDIX A

Table 14

Correlation Matrix of All Variables.

 

CP Time CP Acts Female Non-White Educ. Tenure
CP Time 1 -0.04 0.07 -0.05 -0.07 -0.02
CP Acts -0.04 1 0.07 -0.05 0.07 —0.17***
Female 0.07 0.07 l 0.15** 0.03 0.02
Non-White -0.05 -0.05 0.15** 1 0.06 -0.07
Education -0.07 0.07 -0.03 0.06 1 -0.31***
Tenure -0.02 -0. l 7*** 0.02 -0.07 -0.31*** l
CPO 0.11* 0.06 -0.07 0.02 -0.05 0.04
CP Training 0.05 -0.08 0.11* 0.20*** -0.03 -0.28***
IPD -0.1 -0.22*** 0.09 0.03 0.11* -0.04
Day Shift 0.01 -0.10* 0.04 0.08 -0.12** 0.42***
Beat Prob. -0.09 -0.05 0.18*** 0.06 0.16** -0.21***
Dept. Pro-CP 0.00 —0.07 0.17** 0.04 -0.09 -0.05
Trust 0.05 0.02 0.06 0 -0.11* -0.05
Cooperation -0.01 -0.07 0.01 -0.07 -0.03 -0.14**
Cohesion 0.00 -0.04 -0.05 —0.1 1* -0. 14** -0.02
Support 0.00 0.05 0.04 -0.03 -0.12* -0.05

138

Table 14 (cont’d).

 

CPO CP IPD Day Beat
Training Shift Problems

CP Time 0.1 1* 0.05 -0.1 0.01 -0.09
CP Acts 0.06 -0.08 -0.22*** -0.1* -0.05
Female 007 0.11* 0.09 0.04 0.18***
N on-White 0.02 0.20*** 0.03 0.08 0.06
Education -0.05 -0.03 0.11* -0.12* 0.16**
Tenure 0.04 -0.28*** -0.04 0.42*** -0.21***
CPO 1 -0.40*** -0.72*** 0.32*** -0.21***
CP Training -0.40*** l 0.39*** -0.26*** 0.2***
IPD -0.72*** 0.39*** l -0.08 0.33***
Day Shift 0.32*** -0.26*** -0.08 1 -0.04
Beat Prob. -0.21*** 0.20*** 0.33*** -0.04 1
Dept. Pro-CP —0.40*** 0.43*** 0.31 *** —0.15** 0.05
Trust 0.05 0.1 -0.09 -0.02 -0.16**
Cooperation -0. 12* 0.24*** 0.08 -0.1 1* 0. 14*
Cohesion -0.23*** 0.25*** 0.07 —0.21*** -0.01
Support 0.04 0.06 -0.06 -0.04 -O.l3**

139

Table 14 (cont’d).

 

 

Dept. Trust Cooperation Cohesion Support
Pro-CP

CP Time 0 0.05 -0.01 0 0
CP Acts -0.07 0.02 -0.07 -0.04 0.05
Female 0.17** 0.06 0.01 -0.05 0.04
Non-White 0.04 0 -0.07 -0.1 1* -0.03
Education -0.09 -0.1 1* -0.03 -0.14** -0.l2*
Tenure -0.05 -0.05 -0.14** -0.02 -0.05
CPO -0.4*** 0.05 -0.12* -0.23*** 0.04
CP Training 0.43*** 0.1 0.24*** 0.25*** 0.06
IPD 0.31 *** -0.09 0.08 0.07 -0.06
Day Shift -0.15** -0.02 -0.11* -0.21*** -0.04
Beat Prob. 0.05 -O.16** 0.14** -0.01 -0.13**
Dept. Pro-CP 1 0.27*** 0.23*** 0.32*** 0.25***
Trust 0.27*** 1 0.25*** 0.35*** 0.83***
Cooperation 0.23*** 0.25*** 1 0.33*** 0.26***
Cohesion 0.32*** 0.35*** 0.33*** 1 0.35***
Support 0.25*** 0.83*** 0.26*** 0.35*** 1
N=262
Notes: * (p < .10) ** (p < .05) *** (p < .01)

140

REFERENCES

Aasland, O. G., Ekeberg, O., & Schweder, T. (2001). Suicide rates fiom 1960 to
1989 in Norwegian physicians compared with other educational groups. Social Science

and Medicine, 52(2), 259-265.

Allison, P. D., & Long, J. S. (1990). Departmental effects on scientific
productivity. American Soci01001ca1 Review 55 469-478.

 

Alpert, G. P., & Moore, M. H. (1998). Measuring police performance in the new
paradigm of policing. In G. P. Alpert & A. Piquero (Eds), Community Policing:
Contemporary Readings (pp. 215-232). Prospect Heights, IL: Waveland.

Antonucci, T. C. (1994). A life-span view of women’s social relations. In B. F.
Turner & L. W. Troll (Eds), Women Growing Older (pp. 239-269). Thousand Oaks, CA:
Sage.

Antonucci, T. C., Akiyarna, H., and Lansford, J. E. (1998). Negative effects of
close social relations. Family Relations, 47, 379-384.

Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable
distinction in social psychological research: Conceptual, strategic, and statistical

considerations. Journal of Personality and Social Psychology, 51(6 ), 1173-1182.

Bayley, D. H. (1986). The tactical choices of police patrol officers. Journal of
Criminal Justice 14 329-348.

 

 

Bayley, D. H. (1994). Police for the future. New York: Oxford University Press.

Bayley, D. H., & Bittner, E. (1993). Learning the skills of policing. In Dunham, R.
G., & Alpert, G. P. (Eds.) Critical Issues in Policing (pp. 106-129). Prospect Heights, IL:
Waveland.

Beck, K., & Wilson, C. (1997). Police officers’ views on cultivating
organizational commitment: Implications for police managers. Policing: An International
Journg of Police Strategies and Management, 20, 175-195.

Bittner, E. (1970). The functions of police in modern society. Rockville: NIMH.
Bittner, E. (1983). Legality and workmanship: Introduction to control in the police

organization. In M. Punch (Ed.), Control in the Police Organization (pp. 1-11).
Cambridge, MA: MIT Press.

141

Bittner, E. (1990). Aspects of police work. Boston: Northeastern University
Press.

Bloch, P., & Anderson, D. (1974). Policewomen on patrol: Final report.
Washington, DC: Urban Institute.

Bollen, K. A., & Long, J. S. (1993). Testing structural equation models. Newbury
Park, CA: Sage.

Bourdieu, P. (1984). Distinction: A social critique of the judgment of taste.
Cambridge, England: Harvard University Press.

Bourdieu, P. (1986). The forms of capital. In J. G. Richardson (Ed.), Handbook of
Theog and Research for the Sociology of Education (pp. 241-258). New York:

Greenwood Press.
Bourdieu, P. (199). The logic of practice. Cambridge, UK: Polity.

Brehm, J ., & Rahn, W. (1997). Individual-level evidence for the causes and
consequences of social capital. American Journal of Political Science. 41. 999-1023.

Bryk, A. S. & Raudenbush, S. W. (1992). Hierarchical linear models. Newbury
Park, CA: Sage.

Bursick, R. J ., Jr. (1999). The informal control of crime through neighborhood
networks. Sociological Focus, 32, 85-97.

Burt, R. S. (1997). The contingent value of social capital. Administrative Science
QuarterlYJZ. 339-365.

Buzawa, E. S. (1981). The role of race in predicting job attitudes of patrol
officers. Journal of Criminal Justice 9, 63-77.

 

Caldwell, E. ( 1978). Patrol observation: The patrol encounter, patrol narrative,
and general shift information forms. Police Services Study Methods Report Mr-2.
Bloomington, IN: Indiana University.

Cameron, A. C., & Trivedi, P. K. (1986). Econometric models based on count
data: Comparisons and applications of some estimators and tests. J oumal of Applied
Econometrics, L 29-53.

Carmines, E. G., & Zeller, R. A. (1979). Reliability and validitv assesament.
Beverly Hills, CA: Sage.

142

Carter, D. L. (1986). Hispanic perception of police performance: An empirical
assessment. Journal of Criminal Justice. 13. 487-500.

Coleman, J. S. (1988). Social capital in the creation of human capital. American
Journal of Sociology. 94. 895-8120.

Cooper, R. K. (1986). Occupational stress in police work (Doctoral dissertation,
University of Minnesota). Ann Arbor, MI: University Microfilrns International.

Cordner, G. W. (1998). Community policing: Elements and effects. In G. P.

Alpert & A. Piquero (Eds), Communig Policing: Contemporg Readings (pp. 45-62).
Prospect Heights, IL: Waveland.

Crank, J. P. (1998). Understanding police culture. Cincinnati, OH: Anderson.
Cutrona, C. E. (1996). Social support in couples. Thousand Oaks, CA: Sage.

DeJong, C. (2000). Gender differences in officer attitude and behavior: Providing
comfort to citizens. Manuscript submitted for publication, Michigan State University.

DeJong, C., Mastrofski, S. D., & Parks, R. B. (2001). Patrol officers and problem
solving: An application of expectancy theory. Justice Quarterly. 18. 31-62.

Duffee, D. E., Fluellen, R., & Roscoe, T. (1999). Constituency building and urban

commrmity policing. In R. H. Langworthy (Ed.), Measuring what matters: Proceedings

from the policing research institute meetings (pp. 91-119). Washington, DC: National
Institute of Justice.

D’Unger, A. V., Land, K. C., McCall, P. L., & Nagin, D. S. (1998). How many
latent classes of delinquent/criminal careers? Results from mixed Poisson regression
analyses. American Journal of Sociology, 103(6). 1593-1630.

 

Eck, J. E. & Spelman, W. (1987). Who ya gonna call? The police as problem-
busters. Crime and Delinquency, 33, 31-52.

Edwards, B., & Foley, M. W. (1997). Social capital and the political economy of
our discontent. American Behaviofiral Scientist, 40. 669-678.

Ellison,-K. W., & Genz, J. (1983). StreSflnd the police officer. Springfield, IL:
Thomas.

Feinman, C. (1994). Women in the criminal iustice mtem. Westport, CT:
Praeger.

143

Foley, M. W., & Edwards, B. (1997). Escape from politics? Social theory and the
social capital debate. American Behavioral Scientist. 40. 550-561.

Frank, K. A., & Yasumoto, J. Y. (1998). Linking action to social structure within
a system: Social capital within and between subgroups. American Journal of Sociology.
104, 642-86.

Friedrich, R. J. (1982). In defense of multiplicative terms in multiple regression
equations. American Journal of Political Science. 26(4). 797-833.

Furstenberg, F. F., & Hughes, M. E. (1995). Social capital and successful
development among at-risk youth. Journal of Marriage and the Family, 57, 580-592.

F urstenberg, F. F ., Jr., & Wellford, C. F. (1973). Calling the police: The
evaluation of police service. Law and SocietyReview. 7. 393-406.

Fyfe, J. J. (1981). Who shoots? A look at officer race and police shooting.
Journal of Police Science and Administration 9, 367-382.

 

F yfe, J. J. (1999). Police use of deadly force: Research and reform. In Gaines, L.
K., & Cordner, G. W. (Eds), Policing Perspectives: An Anthology (pp. 408-434). Los
Angeles: Roxbury.

Geller, W. A., & Swanger, G. (1995). Managing innovation in policing: The
untapped potential of the middle manager. Washington, DC: Police Executive Research

Forum.

Glensor, R. W., & Peak, K. J. (2000). Implementing change: Community-oriented
policing and problem-solving. In R. W. Glensor, M. E. Correia, & K. J. Peak (Eds),

Policing Communities: Understanding Crime and Solving Problems (pp. 239-245). Los
Angeles: Roxbury.

Goldstein, H. (1977). Policing a free society. Cambridge, MA: Ballinger.
Goldstein, H. (1990). Problem-oriented policing. New York: McGraw-Hill.

Goldstein, H. (1998). Toward community-oriented policing: Potential, basic
requirements, and threshold questions. In Alpert, G. P., & Piquero, A. (Eds), Community

Policing: Contemporary Readings (pp. 3-22). Prospect Heights, IL: Waveland.

Granovetter, M. S. (1973). The strength of weak ties. American Jouma_qf
Sociology, 78, 1360-1380.

144

Greene, J. R. (1998). Evaluating planned change strategies in modern law
enforcement: Implementing community-based policing. In J -P. Brodeur (Ed.), How to
recogpize good policing (pp. 141-160). Thousand Oaks, CA: Sage.

Greene, J. R., & Klockars, C. B. (1991). What police do. In Klockars, C. B., &
Mastroflcsi, S. D. (Eds), Thinking About Police: Contemporarv Readings (pp. 273-284).
Boston: McGraw-Hill.

 

Greene, W. H. (2000). Econometric analysis (4th Edition). Upper Saddle River,
NJ: Prentice-Hall.

Grogger, J. (1990). The deterrent effect of capital punishment: An analysis of
daily homicide counts. Journal of the American Statistical Association. 85. 295-303.

Gulati, R. (1995). Social structure and alliance formation patterns: A longitudinal
analysis. Administrative Science Quarterly. 40. 619-652.

Haarr, R. N. (1997). Patterns of interaction in a police patrol bureau: Race and
gender barriers to integration. Justice Quarterly, 14, 53-85.

Hagan, F. E. (1993). Research methods in criminal justice and criminology. New
York: Macmillan.

Hale, D. C., & Wyland, S. M. (1999). Dragons and dinosaurs: The plight of patrol
women. In Gaines, L. K., & Cordner, G. W. (Eds), Policing Perapectives: An Anthology
(pp. 450-458). Los Angeles: Roxbury.

Hemstrom, O. (1999). Does the work environment contribute to excess male
mortality? Social Science and Medicine. 49(7). 879-894.

Hoath, D. R., Schneider, F. W., & Starr, M. W. (1998). Police job satisfaction as a
function of career orientation and position tenure: Implications for selection and
community policing. Journal of Criminal Justice 26, 337-347.

 

Hofferth, S. L., & Iceland, J. (1998). Social capital in rural and urban
communities. Rural Sociology, 63, 574-598.

Holdaway, S., & Barron, A. M. (1997). Resigers? The experience of Black and
Asian police officers. London: Macmillan.

Hope, T. (1998). Privatopia on trial? Property guardianship in the suburbs. In K.

Painter & N. T illey (Eds), Surveillance (special issue, Crime Prevention Studies,
forthcoming). Morsey, NY: Criminal Justice Press.

145

J accard, J ., Tunisi, R., & Wan, C. K. (1990). Interaction effects in multiple
regession. Sage Series in Quantitative Applications in the Social Sciences, volume 72.

Jacobs, D., & O’Brien, R. M. (1998). The determinants of deadly force: A
structural analysis of police violence. American Journal of Sociology, 103(4), 837-862.

J 01in, A., & Moose, C. A. (1997). Evaluating a domestic violence program in a
community policing environment: Research implementation issues. Crime and
DeliquencyL43. 279-97.

Kakar, S. (1998). Self-evaluations of police performance: An analysis of the
relationship between police officers’ education level and job performance. Policing: An
International Journal of Police Straiegies and Management, 21, 632-47.

Kappeler, V. E., Carter, D., & Sapp, A. (1992). Police officer higher education,
citizen complaints, and departmental rule violations. American Journal of Police. 11. 37-
54.

Kawachi, 1., Kennedy, B. P., & Wilkinson, R G. (1999). Crime: social
disorganization and relative deprivation. Social Science and Medicine. 48. 719-731.

Kelling, G. (1999). Measuring what matters: A new way of thinking about crime
and public order. In R. H. Langworthy (Ed.), Measuring What Matters: Proceedings From
the Policing Research Institute Meetings (pp. 27-35). Washington, DC: National Institute
of Justice.

Kelling, G. L. & Moore, M. (1988). The Evolving Strategy of Policing.
Perspective on Policing (No. 4). Washington, DC: National Institute of Justice.

Kennedy, B. P., Kawachi, I., Prothrow-Stith, D., Lochner, K., & Gupta, V.
(1998). Social capital, income inequality, and firearm violent crime. Social Science an_d
Medicine. 47. 7-17.

Klinger, D. A. (1997). Negotiating order in patrol work: An ecological theory of
police response to deviance. Criminology, 35, 277-306.

Land, K. C., McCall, P. L., & Nagin, D. S. (1996). A comparison of Poisson,
negative binomial, and semipararnetric mixed Poisson regression models with empirical

applications to criminal careers data. Sociological Methods and Research, 24(4), 387-
442.

Lewis, S., Rosenberg, H., & Sigler, R. T. (1999). Acceptance of community
policing among police officers and police administrators. Policing: An International

146

J ournral of Police Strategies and Management, 22, 567-588.

Long, J. S. (1997). Regression models for categorical and limited dependent
variables. Thousand Oaks, CA: Sage.

Lurigio, A. J ., & Rosenbaum, D. P. (1994). The impact of community policing on
police personnel: A review of the literature. In Rosenbaum, D. P. (Ed.), The Challenge of

Communiiy Policing: Testing the Promises (pp. 147-63). Thousand Oaks, CA: Sage.

Lurigio, A. J ., & Skogan, W. G. (2000). Winning the hearts and minds of police
officers: An assessment of staff perceptions of community policing in Chicago. In R. W.
Glensor, M. E. Correia, & K. J. Peak (Eds), Policing Communities: Understanding
Crime and Solving Problems (pp. 246-256). Los Angeles: Roxbury.

Lyons, W. (1999). The politics of community policing: Rearranging the power to
punish. Ann Arbor, MI: The University of Michigan Press.

Mahar, C., Harker, R., & Wilkes, C. (1990). The basic theoretical position. In R.
Harker, C. Mahar, & C. Wilkes (Eds) An introduction to the work of Pierre Bourdieu:
Thmctice of theory (pp. 1-25). London: Macmillan.

Manning, P. K. (1994). The police: Symbolic capital, class, and control. In G. S.
Bridges & M. A. Myers (Eds), Inequalig, Crime, and Social Control (pp. 80-96).
Oxford: Westview.

Manning, P. K. (1997). Police work (2nd edition). Prospect Heights, IL:
Waveland.

Manning, P. K., & Van Maanen, J. (1978). Socialization for policing. In P. K.
Manning & J. Van Maanen (Eds). Policirg: A View from the Street (pp.267-272). Santa
Monica, CA: Goodyear.

Martin, S. E. (1980). Breaking and entering: Policewomen on patrol. Berkeley,
CA: The University of California Press.

Martin, S. E. (1993). Female officers on the move? A status report on women in
policing. In Dunham, R. G., & Alpert, G. P. (Eds), Critical Issues in Policing (pp. 327 -
347). Prospect Heights, IL: Waveland.

Martin, S. E. (1994). ‘Outsider within’ the station house: The impact of race and
gender on black women police. Social Problems 41, 383-400.

 

Martin, S. E., & Jurik, N. C. (1996). Doing justice, doing gender. Thousand Oaks,
CA: Sage.

147

Marx, G. T. (1978). Alternative measures of police performance. In R. C. Larson
(Ed), Police Accorflzfirility (pp. 15-32). Lexington, MA: Lexington Books.

Mastrofski, S. D. (1981). Surveying clients to assess police performance:
Focusing on the police-citizen encounter. Evaluation Review, 5, 397-408.

 

Mastrofski, S. D. (1982). Issues in police performance measurement. Urban
Interest 4, 3-6.

 

Mastrofski, S. D. (1996). Measuring police performance in public encounters. In
L. T. Hoover (Ed.), uanti 'n uali in Policin (pp. 207-241). Washington, DC:
Police Executive Research Forum.

Mastrofski, S. D. (1999). Policing for people. Washington, DC: Police
Foundation.

Mastrofski, S. D., Snipes, J. B., Parks, R. B., & Maxwell, C. D. (2000). The
helping hand of the law: Police control of citizens on request. Criminology, 38, 307-342.

Mastrofski, S. D., Worden, R. E., & Snipes, J. B. (1995). Law enforcement in a
time of community policing. Criminology, 33, 539-563.

McCarthy, B., & Hagan, J. (1995). Getting into street crime: The structure and
process of criminal embeddedness. Social Science Research. 24. 63-95.

Miller, S. L. (1998). Rocking the rank-and-file: gender issues and community
policing. Journal of Contemporaiy Criminal Justice, 14, 156-172.

Miller, S. L. (1999). Gender and communim policing: Walking the talk. Boston:

Northeastern University Press.

Milutinovich, J. S. (1977). Black-white differences in job satisfaction, group
cohesiveness, and leadership style. Human Relations 30, 1079-87.

 

Molinas, J. R. (1998). The impact of inequality, gender, external assistance and
social capital on local-level cooperation. World Development, 26, 413-431.

Moore, G. (1990). Structural determinants of men’s and women’s personal
networks. American Sociological Review. 55. 726-735.

Moore, M. H. (1992). Problem solving and community policing. In M. Tonry &
N. Morris (Eds), Modern Policing (pp. 99-158). Chicago: University of Chicago Press.

Morash, M., & Greene, J. R. (1986). Evaluating women on patrol: A critique of

148

contemporary wisdom. Evaluation Review. 10. 230-255.

 

Morash, M., & Haarr, R. N. (1995). Gender, workplace problems, and stress in

policing. Justice Quarterly, 12, 113-140.
Morrell, S., Taylor, R., & Slaytor, E. (1999). Urban and rural suicide differentials

in migrants and the Australian-bom, New South Wales, Australia 1985-1994. Social
Science and Medicine, 49( 1). 81-91.

Muir, W. K., Jr. (1977). Police: Streetcomer politicians. Chicago: The University
of Chicago Press.

Nagin, D. S., & Land, K. C. (1993). Age, criminal careers, and population
heterogeneity: Specification and estimation of a nonparametric, mixed Poisson model.

Criminology, 31(3 ), 327-362.

Nelson, J. F. (1980). Multiple victimization in American cities: A statistical
analysis of rare events. American Journal of Sociology. 85(4). 870-891.

Newton, K. (1997). Social capital and democracy. American Behavioral Scientist
@, 575-586.

 

Niederhoffer, A. (1967). Behind the shield: The police in urban sociegi. Garden
City, NY: Doubleday.

Osborne, D. R., & Tseloni, A. (1998). The distribution of household property
crimes. Journal of Quantitative Criminology, 14(3), 307-330.

Palombo, B. J. (1995). Academic professionalism in law enforcement. New York:
Garland.

Paoline, E. A., Myers, S. M., & Worden, R. E. (2000). Police culture,
individualism, and community policing: Evidence from two police departments. Justice

Quarterly, 17(3 ), 575-605.

Parks, R. B., Mastrofski, S. D., DeJong, C., and Gray, M. K. (1999). How officers
spend their time with the community. Justice Quarterly, 16(3), 483-518.

 

Pate, A. M., & Shtull, P. (1994). Community policing grows in Brooklyn: An
inside view of the New York City police department’s model precinct. Crime and

Delinquency, 40, 384-410.

Peillon, M. (1998). Bourdieu’s field and the sociology of welfare. Journal of
Social Policv. 27. 213-29.

Pino, N. W. (2001). Community policing and social capital. Policing: An

149

Interaational J ourng of Police Strategieaand Management, 24(2 ), 200-215.

Portes, A. (1998). Social capital: Its origins and applications in modern sociology.
Annual Review of Sociology. 24. 1-24.

Portes, A., & Landolt, P. (1996). The downside of social capital. The American
Prospect. 26. 18-21, 94.

Portney, K. E., & Berry, J. M. (1997). Mobilizing minority communities: Social
capital and participation in urban neighborhoods. American Beliayioral ScientistAO.
632-344.

Reiner, R. (1998). Process or product? Problems of assessing individual police
performance. In J -P. Brodeur (Ed.), How to recogpize good policing (pp. 55-72).
Thousand Oaks, CA: Sage.

Reiss, A. J ., Jr. (1968). Police brutality — annswers to key questions. Transaction,
5, 10-19.

Reuss-Ianni, E. (1983). Two cultures of policing: Street cops and management
cops. New Brunswick, NJ: Transaction.

Roberg, R., Crank, J ., & Kuykendall, J. (2000). Police and society. Los Angeles:
Roxbury.

Robinson, A. L., & Chandek, M. S. (20003). Philosophy into practice?
Community policing units and domestic violence victim participation. Policing: An

International Journal of Police Strategies and Management, 23(3 ), 280-302.

Robinson, A. L., & Chandek, M. S. (2000b). The Domestic Violence Arrest
Decision: Examining Demographic, Attitudinal, and Situational Variables. Crime and

Delinquency, 46(1), 18-37.

Robinson, A. L., & Morash, M. (2000). Use of the social capital fi'amework for
assessment of a training and education program for people working with crime victims.
The Justice Professional 13(4), 307-341.

Rose, D. R., & Clear, T. R. (1999). Incarceration, social capital, and crime:
Implications for social disorganization theory. Criminology, 36, 441-479.

Rosenbaum, D. P. (1998). The changing role of the police: Assessing the current
transition to community policing. In J-P. Brodeur (Ed.), How to recogize good policing
(pp. 3-29). Thousand Oaks, CA: Sage.

Rosenbaum, D. P., & Lurigio, A. J. (1994). An inside look at community policing

150

reform: Definitions, organizational changes, and evaluation findings. Crime and
Delinquency, 40, 299-314.

Rountree, P. W., & Warner, B. D. (1999). Social ties and crime: Is the
relationship gendered? Criminology, 37, 789-814.

Rubinstein, J. (1973). Cig police. New York: Farrar, Straus, and Giroux.

Sacco, V. F. (1998). Evaluating satisfaction. In J -P. Brodeur (Ed), How to
recogr_1ize good policing (pp. 123-137). Thousand Oaks, CA: Sage.

Sadd, S., & Grinc, R. M. (2000). Implementation challenges in community
policing: Innovative neighborhood-oriented policing in eight cities. In R. W. Glensor, M.

E. Correia, & K. J. Peak (Eds), Policing Communities: Understanding Crime and
Solving Problems (pp. 97-118). Los Angeles: Roxbury.

Sampson, R. J. (1991). Linking the micro- and macro-level dimensions of
community social organization. Social Forces. 70. 43-64.

Sampson, R. J. (1995). The cormnunity. In J. Q. Wilson & J. Petersilia (Eds)
Crime (pp. 193-216). San Francisco, CA: ICS Press.

Sampson, R. J. (1997). The embeddedness of child and adolescent development:
A community-level perspective on urban violence. In J. McCord (Ed.), Violence and
Childhood in the Inner Ciiy (pp. 31-77). Cambridge, UK: Cambridge University Press.

Sampson, R. J ., Morenoff, J. D., & Earls, F. (1999). Beyond social capital: Spatial
dynamics of collective efficacy for children. American Sociological Review. 64. 633-
660.

Sampson, R. J ., Raudenbush, S. W., & Earls, F. (1997). Neighborhoods and
violent crime: A multilevel study of collective efficacy. Science 277 918-24.

 

Sampson, R. J ., & Raudenbush, S. W. (1999). Systematic social observation of
public spaces: A new look at disorder in urban neighborhoods. American Journal of
Sociology. 105(3). 603-651.

Sampson, R., J. & Scott, M. S. (1999). Tackling crime and other public-safeiy
problems: Case studies in problem-solving. Washington, DC: Department of Justice,

Office of Community Oriented Policing Services.

Schumacker, R. E., & Lomax, R. G. (1996). A beginner’s gpide to structural
equation modeling. Mahwah, NJ: Lawrence Erlbaum.

151

Sherman, L. W. (1978). The quality of police education. (Ch. 7). San Francisco:
Jossey-Bass.

Sherman, L. W., Smith, D. A., Schmidt, J. D., & Rogan, D. P. (1992). Crime,
punishment, and stake in conformity: Legal and informal control of domestic violence.
American Sociological Review. 57. 680-690.

Sichel, J. L., Friedman, L. N., Quint, J. C., & Smith, M. E. (1078). Women on
patrol: A pilot study of police performance in New YoraCity. Washington, DC: National
Institute of Law Enforcement and Criminal Justice.

Skogan, W. G. (1989). Communities, crime, and neighborhood organization.
Crime and Delinquency, 35, 437-457.

Skogan, W. G. (1998). Community policing in Chicago. In G. P. Alpert & A.

Piquero (Eds), Commqu Policing: Contemporgy Readings (pp. 159-174). Prospect
Heights, IL: Waveland.

Skogan, W. G. (1999). Measuring what matters: Crime, disorder, and fear. In R.

H. Langworthy (Ed.), Measuring What Matters: Proceedings From the Policing Research
Institute Meetings (pp. 37-53). Washington, DC: National Institute of Justice.

 

Skogan, W. G., & Hartnett, S. M. (1997). Communig policing, Chicago style.
New York: Oxford University Press.

Skolnick, J. H. (1994). Justice without trial (3rd edition). New York: Macmillan.

Skolnick, J. H., & Bayley, D. (1988). Communigi policing: Issues and practices

around the world. Washington, DC: National Institute of Justice.

Smith, D. A. (1986). The neighborhood context of police behavior. In Reiss, A. J .,
Jr., & Tonry, M. (Eds), Communities and Crime (pp. 313-341). Chicago: The University
of Chicago Press.

Sparrow, M., Moore, M. H., & Kennedy, D. M. (1990). Beyond 911: A new era
for policing. New York: Basic Books.

Stalans, L. J ., & F inn, M. A. (1995). How novice and experienced officers
interpret wife assaults: Normative and efficiency frames. _L_aw an_cLSociety Review. 29.
287 -32 1 .

Stanton-Salazar, R. D., & Sanford, M. D. (1995). Social capital and the
reproduction of inequality: Information networks among Mexican-origin high school

students. Sociology of Education, 68, 116-135.

152

 

Stevenson, T. M. (1988). Stress among police officers: Burnout and its correlates
(Doctoral dissertation, California School of Professional Psychology). Ann Arbor, MI:
University Microfihns International.

Teachman, J. D., Paasch, K., & Carver, K. (1997). Social capital and the
generation of human capital. Social Forces. 75. 1-17.

Thouez, J -P. (1984). Cancer mortality differential by social milieu: The case of
the Montreal metropolitan area, 1971. Social Science and Medicine. 18(1). 73-81.

Truxillo, D. M., Bennett, S. R., & Collins, M. L. (1998). College education and
police job performance: A ten-year study. Public Personnel Management, 27, 269-80.

Van Maanen, J. (1978). Kinsmen in repose: Occupational perspectives of
patrolrnen. In P. K. Manning & J. Van Maanen (Eds), Policing: A View from the Street
(pp.115-128). Santa Monica, CA: Goodyear.

Van Maanen, J. (1983). The boss: First-line supervision in an American police
agency. In M. Punch (Ed.), Control in the Police Organization (pp. 275-317). Cambridge,
MA: MIT Press.

Vlahov, D., Tang, A. M., & Lyles, C. (2000). Increased frequency of overdose
deaths among HIV-infected injection drug users. Addiction Research. 8(4). 311-326.

Waldinger, R. (1995). The ‘other side’ of embeddedness: A case-study of the
interplay of economy and ethnicity. Ethnic and Racial Studies,18. 555-580.

Walker, S. (1984). Broken windows and fractured history: The use and misuse of
history in recent police patrol analysis. Justice Qparterly, 1, 77-90.

Walker, S. (1999). The police in America (3rd edition). Boston: McGraw-Hill.

Weisburd, D., & McElroy, J. E. (1988). Enacting the CPO role: Findings from the
New York City pilot program in community policing. In J. R. Greene & S. D. Mastrofski

(Eds), Communiiy Policing: Rhetoric or Realifl? (pp. 89-101). New York: Praeger.

Weisel, D. L., & Eck, J. E. (2000). Toward a practical approach to organizational
change: Community policing initiatives in six cities. In R. W. Glensor, M. E. Correia, &

K. J. Peak (Eds), Policing Communities: Understanding Crime and Solving Problems
(pp. 257-271). Los Angeles: Roxbury.

 

Wellman, B., & Wortley, S. (1990). Different strokes fiom different folks:
Community ties and social support. American Journal of Sociology, 96, 558-88.

153

Westley, W. A. (1970). Violence and the police. Cambridge: MIT Press.

White, S. E., & Marino, K. E. (1983). Job attitudes and police stress: An
exploratory study of causation. Journal of Police Science and Administration 11, 264-74.

 

Williams, H., & Murphy, P. V. (1990). The evolving strategy of police: A
minority View. No. 13 of Perspectives on Policing. Washington, DC: Government
Printing Office.

Williams, J., & Sloan, RC. (1990). Turning concept into practice: The Aurora
Colorado story. East Lansing, MI: National Center for Community Policing.

Wilson, W. J. (1987). The truly disadvantaged. Chicago: The University of
Chicago Press. .

Wilt, G. M., & Bannon, J. D. (1976). Cynicism or realism: A critique of
Niederhoffer’s research into police attitudes. Journal of Police Science ad
Administration 4, 38-45.

 

Winfree, T. L., Jr., & Newbold, G. (1999). Community policing and the New
Zealand Police: Correlates of attitudes toward the work world in a community-oriented
national police organization. Policing: An International Journal of Police Strateges and
Management, 22, 589-617.

Wood, R. L. (1997). Social capital and political culture: God meets politics in the
inner city. American Behavioral ScientisfiO. 595-605.

 

Woolcock, M. (1998). Social capital and economic development: Toward a
theoretical synthesis and policy framework. Theory and Society, 27, 151-208.

Worden, R. E. (1989). Situational and attitudinal explanations of police behavior:
A theoretical reappraisal and empirical assessment. Law apd Society Review. 23. 677-
711.

Wycoff, M. A. (1982). Improving police performance: One more voice. Urban
Interest 4, 8-16.

 

 

Zhao, J ., Thurman, Q., & Lovrich, N. (2000). Communitiy-oriented policing
across the US: Facilitators and Irnpediments to Implementation. In R. W. Glensor, M. E.

Correia, & K. J. Peak (Eds), Policing Communities: Understanding Crime and Solving
Problems (pp. 229-238). Los Angeles: Roxbury.

Zorn, C. J. W. (1998). An analytic and empirical examination of zero-inflated and

hurdle Poisson specifications. Sociological Methods and Research, 26(3), 368-400.

154

     
 

IIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

111111111(twirl

1 3 02334