. .7. .Wuwmwmmuhnmi"
.mnbnnufvmdﬂ

KW . ‘3’.

in
.wmlwwma
, .Hﬂ

i

a
Law

 

Iti... .
5303.15.55}!
§.:.¢xi.\ . {-
Kilis’l » z\

a. . y

2.5!

. :2.
3's»)! .1 it! .-

THESIS

 

’2
J _
20:19 LIBRARY
Michigan State
University

 

 

This is to certify that the
dissertation entitled

THE INFLUENCE OF WORKGROUPS ON OCCUPATIONAL

ATTITUDES OF THE POLICE: A MULTILEVEL EXAMINATION

presented by

Jason Robert Ingram

has been accepted towards fulfillment
of the requirements for the

Ph.D. degree in Criminal Justice

 

 

CL/ﬂH/W

 

Major Professor’ Signature
f/f lo

Date

 

MSU is an Affirmative Action/Equal Opportunity Employer

 

 

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

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 KzlProleccatPreleIRClDateDue.indd

 

 

 

THE INFLUENCE OF WORKGROUPS ON OCCUPATIONAL ATTITUDES OF THE
POLICE: A MULTILEVEL EXAMINATION
By

Jason Robert Ingram

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
Criminal Justice

2010

ABSTRACT
THE INFLUENCE OF WORKGROUPS ON OCCUPATIONAL ATTITUDES OF THE
POLICE: A MULTILEVEL EXAMINATION
By

Jason Robert Ingram

The study of ofﬁcers’ occupational attitudes has been a cornerstone of police
research for almost ﬁfty years. Traditionally, this line of research has been rooted within
individual level approaches in which individual characteristics are expected to explain
attitudes and individual attitudes are expected to explain behaviors. Such approaches,
however, often neglect the fact that ofﬁcers work in organizational contexts that shape
and inﬂuence their attitudes. Relying on a multilevel theoretical framework and using
data from ﬁve departments, the proposed study takes a multilevel approach that examines
the inﬂuence of ofﬁcers’ assigned workgroups on occupational attitudes. Workgroups are
important contexts within the formal organization of policing because they represent
structural boundaries by which both police work and ofﬁcer interactions are patterned. As
such, the study argues that workgroups serve as the most proximal inﬂuence on ofﬁcer
attitudes.

The current work empirically tests three research objectives: 1) To detemiine if
workgroups inﬂuence officer attitudes; 2) To examine the effect of workgroup
characteristics on attitudinal differences across workgroups; and 3) To determine whether
workgroup level attitudes impact ofﬁcer behaviors. Both research and practical

implications are discussed based upon the results of these three objectives.

COPYRIGHT by
JASON ROBERT INGRAM

2010

ACKNOWLEDGEMENTS

I would like to acknowledge a number of people who contributed in various ways
to the completion ofthis dissertation. First, I would like to thank Dr. William Terrill for
all of his assistance (and patience) that he provided throughout this process. Second, I
would like to thank Dr. Eugene Paoline for sharing his insight and ideas on this topic.
Both of you have taught me a great deal, and I truly appreciate the opportunities you have
provided me. In addition, I would also like to acknowledge my committee members, Drs.
Timothy Bynum, Christina DeJong, and Steve Kozlowski for their assistance on this
project.

A team of individuals across the country assisted me and without each of you, this
research could not have been conducted: Ofﬁcer Jack Coraci (Albuquerque Police
Department), Diane Lewis (Knoxville Police Department), Captain Eric Hendricks
(Portland Police Bureau), Sgt. Mark Stevens (Colorado Springs Police Department), and
Lt. Lisa Williams (Fort Wayne Police Department). In addition, I would like to
acknowledge Ryan Arnold, Sasha Craft, Shaﬁna F azel, Michael Hansen, Julie Nehl, and
Alyssa Tibbs for their assistance in Portland. A special thanks to Charles Scheer and
Heidi Bonner for all of their help with data collection and entry. Although sometimes
overlooked in the research process, I would also like to thank the approximately 1,000
police ofﬁcers who took the time to complete the survey.

Finally, I would like to thank my family which includes the Ingram, Georges,
Daliege, and Ondrejka households for all of their love, support, and understanding over
the last several years. Last but not least, I would like to thank my wife, Amanda, for her

unconditional love and support. This dissertation is dedicated to you.

TABLE OF CONTENTS

LIST OF TABLES ......................................................................................................... viii
CHAPTER ONE: INTRODUCTION ............................................................................. 1
Limitations of Prior Research ......................................................................................... 3
Current Study .................................................................................................................. 4
Organization of Dissertation ........................................................................................... 6
CHAPTER TWO: REVIEW OF THE LITERATURE ................................................ 7
Conceptual Development of Ofﬁcers’ Occupational Attitudes ...................................... 7
The Occupational Perspective ..................................................................................... 8
Ofﬁcer Typology Studies .......................................................................................... 15
Empirical Research on Police Ofﬁcer Attitudes ........................................................... 20
Individual Characteristics as Explanations of Ofﬁcer Attitudes ............................... 21
Attitudes as Explanations of Ofﬁcer Behaviors ........................................................ 34
Limitations of Prior Empirical Research ...................................................................... 48
Conceptual Limitations ............................................................................................. 48
Methodological Limitations ...................................................................................... 51
Chapter Summary ......................................................................................................... 52
CHAPTER THREE: THEORETICAL FRAMEWORK ........................................... 54
The Multilevel Theoretical Perspective ........................................................................ 54
Overview of Framework ............................................................................................... 56
The Where: Ofﬁcer Workgroups .................................................................................. 57
The Formal Organization of Policing ....................................................................... 58
Workgroup Conceptualization .................................................................................. 62

The Why: Workgroups as Contexts for Ofﬁcer Attitudes ............................................ 63
Environment .............................................................................................................. 63
Front Line Supervisors .............................................................................................. 65
Workgroup Composition .......................................................................................... 67

The How: Emergence of Shared Attitudes ................................................................... 68
Implications for the Study of Ofﬁcer Attitudes ............................................................ 71
Workgroup Factors Explain Attitudinal Differences ................................................ 71
Attitudes as Contextual Measures ............................................................................. 72
Current Study ................................................................................................................ 75
Research Objective #1 .............................................................................................. 75
Research Objective #2 .............................................................................................. 76
Research Objective #3 .............................................................................................. 76
CHAPTER 4: DATA & METHODS ............................................................................. 78
The Assessing Police Use of Force Policy & Outcomes Study .................................... 78
Description of Study Sites ............................................................................................ 81
City Characteristics ................................................................................................... 81
Department Characteristics ....................................................................................... 83

Patrol Ofﬁcer Characteristics .................................................................................... 84

Ofﬁcer Survey Research Design & Protocol ................................................................ 85
Survey Protocol ......................................................................................................... 86
Conﬁdentiality Protections ....................................................................................... 87
Survey Administration Results ................................................................................. 87

Workgroup Operationalization ..................................................................................... 89

Ofﬁcers’ Occupational Attitudes .................................................................................. 91
Survey Items ............................................................................................................. 91
Validity & Reliability ............................................................................................... 94

Analysis Plan ................................................................................................................ 98
Research Objective #1: Workgroup as Contexts for Ofﬁcer Attitudes .................... 98
Research Objective #2: Workgroup Composition & Ofﬁcer Attitudes .................... 99
Research Objective #3: Workgroup Aggressiveness & Ofﬁcer Behavior ................ 99

CHAPTER FIVE: WORKGROUP EFFECTS ON OFFICER ATTITUDES ....... 101

Analyses & Results ..................................................................................................... 101
Descriptive Statistics ............................................................................................... 102
Between-group Variability ...................................................................................... 105
Group Reliability & Within-group Agreement ....................................................... 112

Discussion ................................................................................................................... 120

CHAPTER SIX: WORKGROUP COMPOSITION & OFFICER ATTITUDES. 124

Data ............................................................................................................................. 124
Variables ..................................................................................................................... 125
Analyses ............................ 127
Results ................................ 129
Descriptive Statistics ............................................................................................... 129
Bivariate Results ..................................................................................................... 131
Multilevel Results ................................................................................................... 133
Discussion .......................................................... . ......................................................... l3 8
CHAPTER SEVEN: WORKGROUP ATTITUDES & FORCE BEHAVIOR ....... 142
Data ............................................................................................................................. 142
Variables ..................................................................................................................... 143
Analysis ....................................................................................................................... 151
Results ......................................................................................................................... 153
Descriptive Statistics ............................................................................................... 153
Soft Hand Force Results ......................................................................................... 156
Hard Hand Force Results ........................................................................................ 157
Impact Force Results ............................................................................................... 162
Discussion ................................................................................................................... 166
CHAPTER EIGHT: CONCLUSION ......................................................................... 170
Summary of Key Findings .......................................................................................... 172
Views toward Citizens ............................................................................................ 172
Role Orientations .................................................................................................... 173

vi

Views toward Supervision ...................................................................................... 176

Policing Tactics ....................................................................................................... 179

Job Satisfaction ....................................................................................................... 182
Study Limitations ........................................................................................................ 183
Research & Practical Implications .............................................................................. 185
APPENDIX A ................................................................................................................ 190
APPENDIX B ................................................................................................................ 191
APPENDIX C ................................................................................................................ 192
APPENDIX D ................................................................................................................ 195
APPENDIX E ................................................................................................................ 199
REFERENCES .............................................................................................................. 200

vii

LIST OF TABLES

Table 2.1: Studies Examining the Effect oflndividual Ofﬁcer Characteristics on

Ofﬁcers’ Attitudes .............................................................................. 24
Table 2.2: Studies Examining the Police Attitude-Behavior Relationship ............. 38
Table 4.1: City, Department, & Ofﬁcer Characteristics of Study Sites ................. 82
Table 4.2: Ofﬁcer Survey Administration Results ......................................... 89
Table 4.3: Workgroup Survey Results (N=61) ............................................. 90
Table 4.4: Final CFA & Reliability Results of Attitudinal Measures (N=l ,022). . 97
Table 5.1: Descriptive Statistics ofOfﬁcers’ Occupational Attitudes (N=1,022). . 103
Table 5.2: One-way Random Effect ANOVA Results .................................... 108
Table 5.3: Between-Workgroup Variability Controlling for Precinct & Department
Levels (N=988) ................................................................................. 111
Table 5.4: Group Reliability & Within-Group Agreement Results (N=6l) ............ 1 17
Table 5.5: Summary of Results for Workgroup Effects & Ofﬁcer Attitudes .......... 121
Table 6.1: Descriptive Statistics for Ofﬁcer Attitudes, Demographics, &

Workgroup Composition Variables .......................................................... 130
Table 6.2: Multilevel Regression Results Predicting Ofﬁcer Attitudes ................. 134
Table 6.3: Multilevel Ordinal Logistic Regression Results Predicting Ofﬁcer

Attitudes ......................................................................................... 135
Table 7.1: Descriptions of Force, Workgroup, Ofﬁcer, & Suspect Variables ......... 150
Table 7.2: Descriptive Statistics for Force, Workgroup, Ofﬁcer, & Suspect

Variables ........................................................................................ 155
Table 7.3: Multilevel Model Results for Hard Hand Force .............................. 159
Table 7.4: Multilevel Model Results for Impact Force .................................... 165
Table A.1: Attitudinal Survey Items ......................................................... 190

viii

Table B.1: Polychoric Correlations Among Attitudinal Survey Items .................. 191
Table C.1: Citizen Distrust .................................................................... 192
Table C.2: Order Maintenance ............................................................... 192
Table C.3: Direct Supervisors ................................................................ 193
Table CA: Top Management ................................................................. 193
Table C.5: Job Satisfaction ................................................................... 193
Table C.6: Aggressiveness .................................................................... 194
Table C.7: Selectivity .......................................................................... 194
Table C.8: Crime-ﬁghting ..................................................................... 194
Table D. 1: rWg and r*wg0) Values by Workgroup ....................................... 195
Table D2: ADM and ADM“) Values by Workgroup .................................... 197
Table E.1: Correlations among Ofﬁcer Attitudes, Demographics, & Workgroup

Composition .................................................................................... 199

ix

CHAPTER ONE: INTRODUCTION

A considerable amount of research has been devoted to police occupational
attitudes, especially those that reﬂect ofﬁcers’ responses to the challenges they face in
their work environments. Such attitudes include views toward citizens (Manning, 1997;
Skolnick, 1967; Rubinstein, 1973; Westley, 1970), policing tactics (Bittner, 1970; Van
Maanen, 1974; Worden, 1995), role orientations (Bittner, 1970; Manning, 1997),
supervision (Van Maanen, 1974; Reuss-Ianni, 1983; Worden, 1995), andjob satisfaction
(Worden, 1995). Collectively, these attitudes have been conceptualized as occupational
outlooks and have become “central constructs” in empirical studies of the police
(Worden, 1995: 49).

One line of empirical research has focused on identifying factors that explain
attitudinal differences among ofﬁcers. This research area has emphasized the importance
of individual ofﬁcer demographic characteristics (i.e., gender, race, age, and education)
or ofﬁcer length of service on attitudinal variation (Paoline, Worden, & Myers, 2000).
Results, however, indicate that demographic characteristics are rather weak predictors of
ofﬁcer attitudes. As such, the utility for examining the effect of ofﬁcer characteristics on
ofﬁcer attitudes has come into question (Carlan, 2007; Lord & Friday, 2008; Paoline et
al., 2000).

A second body of empirical research has examined ofﬁcer attitudes as
explanations of street-level behaviors (e.g., use of force, decision to arrest, or pro-activity
behaviors). Initially, it was believed that ofﬁcer attitudes would explain a substantial
amount of variation in ofﬁcer behaviors (Smith & Klein, 1984; see also Engel & Worden,

2003). Within the policing literature, however, research has shown that this is not the

case. Over the past twenty-ﬁve years, comprehensive reviews of the factors expected to
impact police ofﬁcers’ behaviors have concluded that ofﬁcer attitudes have little impact
on their actions (National Research Council [NRC], 2004; Riksheim & Chermak, 1993;
Sherman, 1980; Worden, 1989, 1995).

The lack of evidence has had a profound effect on research in this area. For
example, it has been estimated that studies on the police attitude-behavior relationship
have decreased by over twenty percent from 2000 to 2004 (Varriale, Gibbs, Ahlin,
Gugino, & Na, 2007). Research has been even less prevalent recently, as only a handful

of studies have been published since 2004 (e.g., McCluskey, Terrill, & Paoline, 2005;

Paoline & Terrill, 2005; Smith, Novak, Frank, & Lowenkamp, 2005).1 The reduction in
such studies may be related to the lack of results stemming from prior research. As Engel
and Worden (2003: 156) conclude, “it appears that variation in how ofﬁcers do their jobs
is not congruent with their occupational attitudes.”

Ofﬁcer attitudes continue to play a critical role in the efforts of police
administrators to shape and control the discretionary behaviors of ofﬁcers despite the lack
of empirical support from research studies (NRC, 2004; Mastrofski, 2004). Recently, the
National Research Council (2004) identiﬁed a number of policies implemented by police
administrators in an attempt to manage ofﬁcers’ behaviors, such as selection and hiring,
performance review, disciplinary procedures, personnel assignment, and training policies.
The council contended that the above policies should inﬂuence ofﬁcers’ street-level

behaviors due to two “intervening mechanisms: ofﬁcer attitudes and beliefs” (NRC,

 

1 Surprisingly, a few of these more recent studies have begun to uncover an attitude-behavior relationship
(see Paoline & Terrill, 2005; Terrill, Paoline, & Manning, 2003). Even though there has been a lack of
support historically, the results of recent research illustrate the importance of continuing and extending this
line of research.

2004: 129). In addition, ofﬁcer attitudes have been identiﬁed as reasons why police
reforms fail (Skogan, 2008) and why the move towards community policing is often met
with resistance (Skogan & Hartnett, 1997). While research has faced challenges
establishing an empirical link between attitudes and behaviors, ofﬁcer attitudes remain at

the forefront of police reforms.

Limitations of Prior Research

Although many studies on ofﬁcers’ occupational attitudes have been conducted,
certain limitations exist. The most notable is the reliance on individual approaches and
assumptions. Studies that aim to explain attitudinal differences among ofﬁcers have
focused primarily on individual ofﬁcer characteristics. Additionally, research on the
police attitude-behavior relationship has historically been rooted within individual
explanations for ofﬁcer behaviors that emphasize the autonomy that ofﬁcers have in
making discretionary decisions (Brown, 1988). These studies assume ofﬁcers develop
their own policing styles shaped by their attitudes and that these styles inﬂuence their
actions (Worden, 1989).

A key limitation of individual approaches is that they imply that ofﬁcer attitudes
exist in a social vacuum, uninﬂuenced by the broader environment or other ofﬁcers
(Drummond, 1976; Frank & Brandi, 1991; Klinger, 1997). However, two aspects of the
policing environment limit the utility of individual approaches: the organizational context
and police culture. Scholars have noted that ofﬁcers’ organizational contexts likely exert
a direct effect on ofﬁcer attitudes (Frank & Brand], 1991). Additionally, research on

police culture indicates that ofﬁcers share occupational attitudes (Paoline, 2003). These

two aspects point out the need to move beyond individual level approaches and toward

approaches that can incorporate the organizational and cultural contexts of policing.

Current Study

The purpose of the current study is to incorporate a multilevel approach to the
study of ofﬁcer attitudes. Relying on a multilevel theoretical paradigm from
organizational psychology as a framework (Klein, Dansereau, & Hall, 1994; Kozlowski
& Klein, 2000), the current study identiﬁes workgroups as important organizational
contexts that inﬂuence ofﬁcer attitudes. Workgroups, deﬁned as ofﬁcers’ assigned shifts
within precincts, are important boundaries of the formal organization of policing because
they provide the most immediate environment in which police work is carried out.
Ofﬁcers within workgroups are exposed to common environmental features, supervisors,
and workgroup compositional characteristics (e.g., racial composition, gender
composition, educational composition, and experience levels) which may shape their
attitudes. Additionally, workgroup boundaries structure ofﬁcers’ interaction patterns.
Collectively, it is argued that ofﬁcers’ exposure to common features of the environment
as well as each other within the workgroup context serve to produce shared
understandings of these environments, which thereby produce shared, collective attitudes.
These collective attitudes likely inﬂuence ofﬁcer behaviors.

Take for example ofﬁcers’ attitudes toward citizens. Traditionally, ofﬁcers have
been described as being distrustful and suspicious of the citizens (Skolnick, 1967;
Westley, 1970). With recent demographic changes that have created greater diversity and
civilianization into the policing occupation along with the adoption of more community

oriented policing approaches, it has also been argued that ofﬁcers may no longer hold

such distrustful attitudes (Paoline, 2001). However, the fact that ofﬁcers work in different
environments, are exposed to different community characteristics, and are exposed to
supervisors who have their own views toward citizens likely impacts levels of distrust. As
a result, organizational characteristics are important factors that should explain
differences in ofﬁcers’ views toward citizens. This is particularly salient at the
workgroup level as this is the most immediate level where ofﬁcers come into contact with
citizens. Also, because ofﬁcers within workgroups work together routinely, they are more
likely to develop common or shared views toward citizens. If ofﬁcers within workgroups
share similar views toward citizens, then ofﬁcers’ attitudes may best represent contextual
measures of the workgroup environment rather than being treated as individual attributes.
Such contextual measures should serve as important explanations for ofﬁcers’ street level
behaviors.

Using survey and ofﬁcial police data from ﬁve police departments, this study
examines workgroup effects on ofﬁcers’ oCcupational attitudes and addresses the
following three research objectives and associated research questions:

Research Objective #1 : First, the study explores the viability of workgroups as important
contexts that shape ofﬁcer attitudes by posing three research questions.

@estion 1: Are there signiﬁcant differences in ofﬁcer attitudes between
workgroups?

 

Question 2: How do these differences compare to other formal organizational
levels of policing, such asﬁprecinct and department levels?

Question 3: Do ofﬁcers within workgrorms share attitudes?

 

Research Objective #2: To the extent that workgroups serve as important contexts for

attitudes, the second research objective further tests the framework by examining

workgroup compositional characteristics (e.g., racial composition, gender composition,
educational composition, and experience levels) as explanations for attitudinal
differences among ofﬁcers.

Question 4: Do workgroup compositional characteristics explain attitudinal
differences across workgroups?

Research Objective #3 : To the extent that ofﬁcers within workgroups share attitudes, the
third research objective tests whether collective, workgroup level attitudes explain
variation in ofﬁcers’ behaviors. To better illustrate this aspect of the theoretical
framework, this objective focuses on the impact of shared attitudes on one type of
behavior (i.e., use of force).

Question 5: Do workgroup level attitudinal measuresﬁpredict ofﬁcers’ use of
force behavior?

Organization of Dissertation

Chapter Two reviews the literature on ofﬁcers’ occupational attitudes. Speciﬁc
attention is given to the conceptual deveIOpment of occupational attitudes and the results
of empirical research. Limitations of the individual approach emphasized in this literature
are also discussed, and the chapter concludes with the rationale for a multilevel approach
to the study of ofﬁcer attitudes. Chapter Three presents the multilevel theoretical
framework used in the study. This chapter provides the rationale for why workgroups
serve as important contexts for ofﬁcers’ attitudes and also discusses how ofﬁcers come to
share attitudes within workgroups. Chapter Four discusses the data and methodologies
used to test the proposed research questions. Chapters Five through Seven report the
results for each of the three study objectives. Finally, Chapter Eight discusses the

research and practical implications that arise from the ﬁndings.

CHAPTER TWO: REVIEW OF THE LITERATURE

This chapter reviews the literature on police ofﬁcers’ occupational attitudes and is
divided into four sections. First, the chapter begins by discussing two bodies of literature
that have been important to the conceptual development of attitudinal research on the
police: the occupational perspective and ofﬁcer typology studies. The second section of
this chapter reviews empirical research on police ofﬁcer attitudes. Studies that have
sought to explain attitudinal differences among ofﬁcers are reviewed ﬁrst. Next, studies
that have tested the police attitude-behavior relationship are reviewed. The third section
identiﬁes the conceptual and methodological limitations associated with empirical
research studies. Finally, a summary of the extant literature and the need for alternative

approaches to the study of ofﬁcers’ attitudes is provided.

Conceptual Development of Ofﬁcers’ Occupational Attitudes

Two bodies of literature have been important to the conceptual development of
attitudinal research on the police: the occupational perspective and ofﬁcer typology
studies. Research within the occupational perspective has served to identify a set of
occupationally relevant attitudes developed as ways for ofﬁcers to successfully cope with
the challenges and problems they face in their work environments. This line of research
has assumed that ofﬁcers hold similar attitudinal outlooks. In contrast, ofﬁcer typology
studies question the occupational perspective. These studies have highlighted variation in
ofﬁcers’ attitudes and have noted that understanding these differences is important. As a
result, ofﬁcer attitudes have become “central constructs” in police research focused on
' explaining attitudinal differences among ofﬁcers and whether these differences impact

ofﬁcer behaviors (Worden, 1995: 49).

The Occupational Perspective

The conceptual underpinnings for the study of ofﬁcers’ occupational attitudes
arose out of a set of early observational studies of the police (i.e., Bittner, 1970; Manning,
1997 [1977]; Rubinstein, 1973; Skolnick, 1967; Van Maanen, 1974; Westley, 1970).
Collectively, these studies provided some of the ﬁrst accounts of what police work was
like based on Observations of ofﬁcers in their work environments. Before, writings on the
police were largely descriptions of police departments that focused on their organization
and administration (Rubinstein, 1973). The above scholars were some of the ﬁrst to move
beyond such descriptions and, as a result, have had a profound impact on police research
as they have provided the foundation for the systematic study of policing in general and
police ofﬁcer attitudes more speciﬁcally (Manning, 1997 [1977]; Westley, 1970).

Many of these studies focused on providing insight into the unique features of
ofﬁcers’ work environments and the ways in which ofﬁcers responded to them. In doing
so, these early observational studies took an occupational approach to the study of
policing. For example, Skolnick’s (1967: 42) research portrayed ofﬁcers as having
“distinctive cognitive tendencies. . .as an occupational grouping” which comprised a
working personality among ofﬁcers. This working personality developed as a result of
the challenges ofﬁcers faced in their work environment and the manners by which
ofﬁcers responded to these challenges (Rubinstein, 1973; Skolnick, 1967). Others also
made similar occupational characterizations, as Van Maanen (1974: 87) described
ofﬁcers’ responses to their work environment as an “occupational ideology,” and Westley
(1970: 49) referred to ofﬁcers’ responses as “occupational directives.” Thus, as a result of

working in the same occupation, ofﬁcers were believed to respond similarly to their

work-related challenges. By focusing on policing as an occupation, this body of research
made two key contributions to the study of ofﬁcer attitudes.

First, early observational studies highlighted the major challenges that ofﬁcers
faced in their work environments. Speciﬁcally, research emphasized danger, authority,
uncertainty, and ambiguity issues that were associated with police work. Danger and
authority issues stemmed from ofﬁcers’ contacts with citizens in their occupational
environment, while uncertainty and ambiguity issues stemmed primarily from ofﬁcers’
contacts with supervisors and administrators in their organizational environment (see
Paoline, 2003). The challenges inherent in police work described by this body of research
are important because they gave rise to a second key contribution: the identiﬁcation of a
set of attitudinal responses developed by ofﬁcers to assist them in facing these
chanenges.

These responses have come to be conceptualized as an interrelated set of attitudes
that comprise ofﬁcers’ occupational outlooks (Worden, 1995). The most prevalent
attitudes included: suspiciousness or distrust of the citizenry (Manning, 1997; Skolnick,
1967; Rubinstein, 1973; Westley, 1970), a strong orientation to the law enforcement role
(Bittner, 1970; Manning, 1997); subscription to aggressive policing tactics (Bittner, 1970;
Van Maanen, 1974; Worden, 1995); negative attitudes toward supervision (Van Maanen,
1974; Reuss-Ianni, 1983; Worden, 1995); and job satisfaction (Worden, 1995). The
following provides a description of these attitudes as well as their relation to the

challenges associated with police work.

Views toward Citizens

As a result of the dangers and uncertainties associated with police work, ofﬁcers
have been described as being distrustful or suspicious of citizens with whom they come
into contact with on the street. Such a view has been attributed to the fact that encounters
with citizens often occur with those in violation of the law, a dynamic which fosters
hostility and resentment towards ofﬁcers (Rubinstein, 1973). This hostility and
resentment has a twofold effect on ofﬁcers. First, ofﬁcers develop a View that
characterizes citizens as derogative and uncooperative. For example, Westley (1970: 93)
reported that ofﬁcers believed that the citizens they served were against them. The
majority of ofﬁcers described the public’s view of them in negative terms with the
strongest conceptions stating that the public was “antipolice.” Thus, as a result of their
work, the police believe that the public is not supportive of their efforts.

Second, ofﬁcer contacts with citizens have the unpredictable potential to lead to
violence. Westley (1970: 105) noted that the “major types of occupational experiences”
of police ofﬁcers involve trafﬁc violations, family disputes, criminal investigations,
ﬁghts, and juvenile delinquency. Many of these experiences have been characterized as
the most dangerous and uncertain types of situations that ofﬁcers must face (Bayley &
Bittner, 1984). Because of the potential threat of danger inherent in police contacts with
citizens, ofﬁcers use their suspicions and distrust as a way to protect themselves from
threats or signs of potential violence (Manning, 1997; Skolnick, 1967). In short, ofﬁcers’
distrust of citizens serves as a way for ofﬁcers to cope with the dangers and uncertainties
associated with their encounters with citizens in their occupational environment, a major

component of police work.

10

Views toward Policing Tactics

Another challenge ofﬁcers’ face in their occupational environments concerns
issues of authority. Here, aspects of police work interact to create potential problems for
ofﬁcers during citizen encounters. For example, one of the core components of policing
is the power granted to ofﬁcers to use the amount of force necessary to carry out their
primary role of enforcing laws (e. g., Bittner, 1970). As Skolnick (1967: 44) noted
however, in enforcement situations “the policeman directs the citizenry, whose typical
response denies recognition of his authority,” often through resistant or violent means.
Thus, the potential for citizens to deny and rebel against ofﬁcers’ authority creates
problems for ofﬁcers because it serves to reinforce the threat of danger in citizen
encounters (Bittner, 1970; Skolnick, 1967; Van Maanen, 1974; Westley, 1970).

In response to these potential threats of violence and challenges to authority,
ofﬁcers hold strong views toward aggressive displays of authority in order to minimize
such threats. For example, Van Maanen (1974) noted ofﬁcers strived to “maintain the
edge” or sought to be decisive at all times in all situations. Furthermore, ofﬁcers often
seek to establish a dominant position in citizen encounters. As Bittner (1970: 26-7) noted,
“ofﬁcers believe that by overwhelming those with whom they deal at the outset, they nip
in the bud any potential resistance or opposition.” Thus, establishing authority has been
deemed a “principal concern” of ofﬁcers (Manning, 1997: 198). In essence by developing
an aggressive orientation towards displays of their authority, ofﬁcers believed they could

further reduce the potential threat of violence and danger in their contacts with citizens.

ll

Role Orientations

 

Another occupational attitude identiﬁed by early observational studies of the
police concerns ofﬁcers’ narrow attitudes toward the police role, which arise out of
ambiguous role deﬁnitions that ofﬁcers often experience. Although policing the street
requires ofﬁcers to perform a wide variety of activities, they are generally encompassed
within three primary roles: enforcing laws, maintaining order, and providing services to
citizens (Bittner, 1970). However, evaluating ofﬁcers’ performance in these areas is
rather difﬁcult and as a result, the law enforcement role often takes precedent over order
maintenance and service roles (Van Maanen, 1974; see also Paoline, 2001). In part, this is
due to the fact that the law enforcement role is the one in which both the general public
and the police organization use to judge the effectiveness of the police (Manning, 1997).

For example, in the eyes of the general public, the role of law enforcement serves
as the “core of the police mandate and the principal justiﬁcation for the existence of the
police establishment” (Bittner, 1990: 20). Furthermore, in the eyes of the police
organization, the enforcement of laws is the role that gets ofﬁcially recognized by the
department; thus ﬁghting crime is the role that gets characterized as real police work
(Bittner, 1990; Manning, 1997). In a review of the occupational perspective, Paoline
(2001) noted that the emphasis on law enforcement creates role ambiguities for ofﬁcers.
Ofﬁcers are expected to perform a broad array of duties, but are evaluated mostly on their
abilities to control crime. As a result, ofﬁcers are said to hold narrow attitudes toward the
police role that emphasize crime ﬁghting and law enforcement, as opposed to holding
role orientations that encompass the broader range of job related activities required of

ofﬁcers.

12

Views toward Supervision

In addition to ambiguous role deﬁnitions, ofﬁcers also face uncertainty in their
organizational environment in terms of how their efforts are viewed by both immediate
supervisors and top management. In conducting their duties, ofﬁcers must follow legal as
well as departmental guidelines, which sometimes are at odds with the roles they are
expected to perform. As Paoline (2001: 14) noted, “ofﬁcers are constrained, working
within an organization that demands that all problems be handled on the street. . .yet held
to excessive scrutiny by ‘watchful administrators’ at a later date.” The demand to handle
all types of problems increases ofﬁcers’ contacts with citizens, which also increases their
opportunity to violate legal and departmental guidelines.

Furthermore, police organizations focus more on sanctioning and disciplining
ofﬁcers for their mistakes rather than for rewarding ofﬁcers who are recognized for good
performance (Van Maanen, 1974). As a result, ofﬁcers were often found to hold negative
attitudes toward supervision at both the immediate and upper administrative levels. A
lack of trust in supervision develops where ofﬁcers feel that supervisors fail to support
them in times of need and that top management fails to recognize good police work yet
heavily scrutinizes ofﬁcers when mistakes are made (Paoline, 2001; Reuss-Ianni, 1983;
Van Maanen, 1974).

Job Satisfaction

A ﬁnal occupational attitude concerns ofﬁcers’ satisfaction with their jobs.
Research has emphasized that job satisfaction is key for understanding the police
occupation. As Reiss (1967: 1-2) noted, job satisfaction concerns a “major feature of

police work as an occupation” as well as being important to understanding “how ofﬁcers

13

 

 

Jr,
.‘IO

v

I‘L‘

\

 

i
.1.

orient themselves to their work. . Much of the early research has found ofﬁcers to be
satisﬁed with their occupation. Studies from the 19703 and early 19805 reported that only
one in ten ofﬁcers were dissatisﬁed with their jobs and most ofﬁcers had a high level of
commitment toward their work (Worden, 1995; Sharp, 1982, Van Maanen, 1975). This
led to the conclusion that “[m]ost police ofﬁcers, it appears, like their jobs and the
organizations in which they work” (Worden, 1995: 68). Thus, even though ofﬁcers were
often described as having to be distrustful of citizens and supervisors, as well as having to
cope with the uncertainties, dangers, and ambiguities of their work environments, the
occupational perspective traditionally characterized ofﬁcers as being highly satisﬁed with
their occupation. 2
Summary of the Occupational Perspective

Results from early observational studies on the police described ofﬁcers’ work
environments as being characterized by issues of danger, authority, uncertainty, and
ambiguity (Bittner, 1970; Manning, 1997 [1977], Rubinstein, 1973; Skolnick, 1967; Van
Maanen, 1974; Westley, 1970). In providing such descriptions, the studies took an
occupational perspective that depicted ofﬁcers “as a homogenous occupational
group. . .having in common particular psychological attributes that distinguish them from
the rest of the population” (Worden, 1995: 49). Such common psychological attributes

took the form of a set of occupational attitudes developed by ofﬁcers in response to the

 

2

It should be noted that the inclusion of job satisfaction as an occupational attitude of ofﬁcers departs
somewhat from traditional accounts of the occupational attitudes of the police. Most of these accounts have
not explicitly included job satisfaction as a part of their discussion. Additionally, as will be seen in the next
section that reviews empirical research on ofﬁcer attitudes, job satisfaction is examined independently of
other ofﬁcer attitudes. However, Worden (1995) characterized ofﬁcer job satisfaction as part of ofﬁcers’
pccupational outlooks in his synthesis of the police occupational attitude literature and so in this study it is
Included as part of the set of attitudes examined.

14

issues of their work environments. This set of attitudes included views towards citizens,
policing tactics, the police role, supervision, and overall job satisfaction. The
identiﬁcation of these attitudes within this occupational framework has been an important

contribution to policing research.

Officer Typology Studies

Ofﬁcer typology studies have also provided important insight into ofﬁcers’
occupational attitudes. Much like the early observational studies associated with the
occupational perspective, typology studies were also based on observations of ofﬁcers in
their work environments (Broderick, 1977; Brown, 1988; Muir, 1977; White, 1972).
While the occupational perspective essentially assumed ofﬁcers’ held similar attitudinal
outlooks as a result of working in the same occupation, the results from typology studies
have questioned this assumption. Unlike the occupational perspective, typology studies
argue that ofﬁcers responded differently to these challenges, which lead to attitudinal
differences (Brown, 1988).

The typology studies offered four primary explanations for why such differences
may exist among ofﬁcers as well as the attitudes associated with these differences. For
example, Broderick’s (1977) typology focused on differences in ofﬁcers’ working
personalities. Similar to Skolnick’s (1967) occupational characterization of ofﬁcers’
working personality, Broderick (1977) found evidence of multiple working personalities
based on ofﬁcers’ attitudes associated with the police role and policing tactics.
Personality differences, then, in ofﬁcers’ views toward their work environment have been

offered as one explanation for attitudinal variation.

15

A second explanation focuses on differences in the ability of ofﬁcers to cope with
the coercive authority granted to them. This explanation focuses on ofﬁcers’ moral
qualities and their views toward authority. For example, Muir (1977) found differences in
ofﬁcers’ abilities to morally reconcile the use of coercion as well as their ability to
empathize with the citizens they serve. Ofﬁcers were grouped in different types based on
the differences in these views (Muir, 1977). While the above two reasons focus on
individual differences of ofﬁcers, the ﬁnal two explanations focus on changes in the
policing occupation as well as the nature of police work.

A third explanation concerns the move to professionalize policing. White (1972)
argued that the move to professionalize the police served to change ofﬁcers’ perceptions
of the police role to encompass broader role orientations other than those solely related to
crime ﬁghting. For example, White (1972: 64) noted that “several different sets of role
perceptions can develop within the context of professionalization,” and these sets vary by
ofﬁcer. Thus, professionalization affects ofﬁcers in different ways and allows for
variation in their views toward the police role.

A ﬁnal explanation is based on the vast amount of autonomy afforded to ofﬁcers.
Ofﬁcers working the street have largely been described as individualistic and their
decisions about how to conduct their work are made primarily on their own. A result of
this autonomy is that ofﬁcers develop distinctive policing styles (Brown, 1988;
Goldstein, 1977). As Brown (1988: 87) highlighted, “the signiﬁcance of...individualism
for police discretion is that [it serves] to decentralize decision-making,” and such
decentralization allows “patrolmen the freedom to fashion distinctive approaches to

police work and to handle their job as they see ﬁt.” The autonomy afforded to ofﬁcers

l6

‘ \
{5.4a I

'\ ~ R
‘Ksm

>"I

Us“ .
s

P.
.,\\‘

Lu

 

’1-

will "

'Ir

,.1 ‘
,Hvl‘

leads them to develop distinctive policing styles and that these styles are shaped by their
occupational attitudes (Brown, 1988).

While the occupational perspective focused on the attitudinal similarities among
officers, typology studies focused on attitudinal differences and classiﬁed ofﬁcers based
upon the attitudinal dimensions under study. Ofﬁcers with similar attitudinal dispositions
were grouped together into a common type (Worden, 1995). In a synthesis of this
typology research, Worden (1995) identiﬁed the attitudinal components that served as the
basis for ofﬁcer types. The most common attitudes included many identiﬁed by the early
observational studies of the police (e.g. views toward the police role, citizens, policing
tactics, supervision, coercion, and job satisfaction). In conducting such a synthesis,
Worden (1995) found evidence for ﬁve common types of ofﬁcers based on attitudinal
differences (e.g., professionals, tough cops, clean-beat crime-ﬁghters, problem solvers,
and avoiders).

Similar to the occupational perspective, ofﬁcer typology studies have also made
two important contributions to the study of the occupational attitudes of police. First,
typology studies further reﬁned the attitudes relevant to ofﬁcers. The most notable
contribution was made by the work of Brown (1988). Based on his observations, Brown
(1988) offered additional insight into ofﬁcers’ views toward the police role by
introducing attitudes associated with aggressiveness and selectivity.

For example, Brown (1988: 223) noted that ofﬁcer aggressiveness was comprised
of two characteristics, “taking the initiative on the street to control crime and a
preoccupation with order that legitimizes the use of illegal tactics.” As a result, ofﬁcer

aggressiveness and views toward legal restrictions were introduced as additional attitudes

17

of ofﬁcers. Selectivity on the other hand relates to how ofﬁcers view their priorities in
controlling crime. Brown (1988) found that not every ofﬁcer believed that all laws should
be enforced. Some ofﬁcers preferred focusing on more serious felony offenses while
ignoring less serious misdemeanors. Selectivity as an attitudinal measure came to be
deﬁned as “the conscious and systematic use of the law to control a particular crime
problem” (Brown, 1988: 157).

Second, the ﬁndings from ofﬁcer typology studies indicated that variation existed
in ofﬁcers’ occupational attitudes. As Worden (1995: 49) concluded, this research has
generally indicated that “police ofﬁcers are not psychologically homogenous.” While the
occupational perspective placed an emphasis on the homogeneity of ofﬁcers, typology
studies emphasized the variance (NRC, 2004; Paoline et al., 2000). Thus, typology
studies are useful for showing variation in ofﬁcers’ attitudes and in demonstrating why
such variation is theoretically important. Police scholars, however, have criticized the
ability of ofﬁcer typology studies to test explicit hypotheses regarding the impact of these
differences (Hassell, 2006; Herbert, 1998; Hochstedler, 1981; Worden, 1995). As such,
the results from typology research on the police have garnered theoretical interest for
ofﬁcer attitudes as both dependent and independent variables in studies of the police
(Worden, 1995).

Summarvof Typology Studies

Unlike the occupational perspective, which assumed ofﬁcers responded to the
challenges of their work environment in similar ways, ofﬁcer typology studies
highlighted differences in ofﬁcers’ responses (Broderick, 1977; Brown, 1988; Muir,

1977; White, 1972). These differences were due to variation in ofﬁcers’ attitudes toward

18

their work, and officers were placed in different types based on these attitudes. As such,
typology studies have been important to the conceptual development of ofﬁcer attitudes
as they have “provide[d] some of the best written sources of such information” (Worden,
1995252)

It should be noted that the ofﬁcer types proffered by the early typology
researchers were rooted within an individual level perspective. For example, Broderick’s
(1977) and Muir’s (1977) evidence for different ofﬁcer types were based on differences
in individual ofﬁcers’ working personalities or moral qualities, respectively. Furthermore,
White’s (1972) and Brown’s (1988) typologies were based on the individualistic nature
of police work and the autonomy granted to ofﬁcers to perform their work. Such an
emphasis on these individual level explanations for attitudinal differences has had a
strong impact on empirical research on ofﬁcers’ attitudes discussed in the next section.

In addition to the explanations offered by typology studies for attitudinal
differentiation, typology research made two important contributions to the study of
ofﬁcer attitudes. First, this line of research offered additional insight into the attitudes
relevant to ofﬁcers. For example, Brown’s (1988) observational work identiﬁed two
additional attitudes important to ofﬁcers: their views toward aggressively controlling
crime as well as their selectivity in enforcing laws. Second, typology research provided
evidence of variation in ofﬁcer attitudes and that this variation is theoretically
meaningful. As a result, typology studies have had a profound inﬂuence on empirical
research on ofﬁcer attitudes that have sought to explain attitudinal variation as well as on
research that has sought to ascertain if attitudinal differences impact ofﬁcers’ street-level

behaviors.

l9

Empirical Research on Police Ofﬁcer Attitudes

Empirical research on ofﬁcer occupational attitudes generally falls within one of
two categories: 1) studies that aim to explain differences in ofﬁcer attitudes and 2) studies
that aim to explain ofﬁcer behaviors as a function of their attitudes, which is more
commonly referred to as the police attitude-behavior relationship (Frank & Brandl,
1991). Both lines of research are rooted within an individual level approach to the study
of ofﬁcer attitudes. For example, studies aimed at explaining attitudinal differences place
an emphasis on the signiﬁcant demographic changes that the occupation of policing has
undergone over the last forty years. These studies seek to determine if such changes
impact how ofﬁcers view their work and primarily focus on the effect of individual
ofﬁcer demographic characteristics (i.e., gender, race, age, and education) or length of
service on ofﬁcers’ attitudes (Manning, 2005; Paoline et al., 2000).

Research on the police attitude-behavior relationship has also been subsumed
under individual explanations of ofﬁcer behaviors. Individual theories of police behavior
propose that “who a police ofﬁcer is makes a difference in how he acts” and ofﬁcer
attitudes are important explanatory characteristics (Sherman, 1980: 71). Similar to the
views of the typology studies, ofﬁcers are believed to develop distinctive policing styles
and that these styles are shaped by their occupational attitudes (Worden, 1996). As
Worden (1996: 25) has noted, “differences in attitude are presumed to manifest
themselves in ofﬁcers’ behavioral patterns.” Thus, variation in individual ofﬁcers’

attitudes is expected to explain variation in individual ofﬁcers’ behaviors.

20

 

' I

.\..-

H‘,

,n

I

.

‘.
,1

Hit‘

1'“

In“.

I

~ 1

.1 .‘
“

l“~

Individual Characteristics as Explanations of Ofﬁcer Attitudes

The focus on individual ofﬁcer characteristics as explanations of ofﬁcer attitudes
is based on the idea that differences in ofﬁcers’ backgrounds lead to differences in how
they view police work. Two often cited examples regard ofﬁcer gender and race. Female
ofﬁcers are said to have “fundamentally different priorities and perspectives on human
relationships” and these differences lead to variation in how male and female ofﬁcers
view their jobs (A. Worden, 1993: 204). Differences in cultural backgrounds between
minority and white ofﬁcers are also expected to lead to differences in how they View their
work (Paoline et al., 2000; Sun, 2003). Furthermore, the long-held resistance to the
integration of minority and women ofﬁcers into the police occupation is believed to have
an adverse impact on job satisfaction as the discrimination experienced by minority and
women ofﬁcers leads to dissatisfaction (Carlan, 2007). In essence, the occupation of
policing has evolved to include a greater diversity of social backgrounds and this
diversity is expected to affect ofﬁcers’ occupational attitudes (Paoline et al., 2000).

Examining the relationship between ofﬁcer characteristics and occupational
attitudes is important for two reasons. First, it offers potential insight into why ofﬁcers
hold certain types of attitudes. Second, this insight can aide in explaining differences in
ofﬁcers’ actions. For example, prior research has noted that “it is critical to understand
ofﬁcers’ perceptions. . .and what factors inﬂuence these attitudes, since these attitudes
could affect the ways ofﬁcers interact with citizens” (Moon & Zager, 2007: 485). The
following reviews empirical research that has examined the effect of individual ofﬁcer
characteristics on occupational attitudes of the police. Only studies that focused on

explaining differences in the ﬁve attitudinal areas discussed previously in this chapter

21

,1 .
I" :1 n

L..u.u

A I

u'us

V; n
i» 5..

ilJr (

sung
s

r 3'

AI‘L

d

 

 

were included in the review (i.e., views toward citizens, role orientations, views toward
supervision, policing tactics, and job satisfaction). Additionally, only studies that
employed a multivariate framework in their analyses were sought to ensure that potential
confounding effects were controlled for. In all, ﬁfteen studies ﬁt these criteria and are
reviewed here. Table 2.1 provides a summary of the studies and includes the ofﬁcer
characteristics used as predictors, the attitudes used as outcomes, study sample sizes, and
a description of the relevant ﬁndings. Many of the studies included multiple ofﬁcer
characteristics in their analyses. Thus, the following discussion also highlights the
research results organized by each ofﬁcer characteristic.

gawk:

Thirteen of the studies included ofﬁcer gender as a variable expected to explain
attitudinal differences. Only two of these studies focused primarily on the effect of
gender on ofﬁcer attitudes. DeJong (2004) examined the role of gender in explaining
ofﬁcers’ distrust toward citizens using data from over 6,000 encounters from the Project
on Policing Neighborhoods study (POPN). While the results indicated that gender had no
direct effect on distrust, two signiﬁcant interaction effects were reported: nonwhite males
were found to be less distrustful of citizens and college educated females were more
likely to be distrustful of citizens. The effects were rather weak however, as the entire
model including all ofﬁcer demographic variables and interaction terms explained only
3.6 percent of the variation in citizen distrust (DeJong, 2004).

A second study focused on the role of gender differences in explaining variation
in ofﬁcer attitudes toward the police role, policing tactics, citizens, and supervisors. In a

re-analysis of survey data from 740 ofﬁcers from the Police Services Study (P88),

22

I .v‘
Ii. L,
.-
‘n. ,
',_..ss
o-.. .
.LM.
0),...
,s...u
man
i..,,

I
t
_,--i .

~..

'7’:
”Hull

F l

ilk“

H;
'lel

9\\ ,
x

q.”
‘i.

\u
n

y

Worden (1993) also reported a handful of signiﬁcant direct and interactive effects for
ofﬁcer gender. For example, female ofﬁcers were more likely to hold positive attitudes
towards citizens. Additionally, when compared to white males, both minority male and
female ofﬁcers were more likely to hold favorable attitudes towards legal restrictions
imposed on police work, and minority male ofﬁcers were more likely to prefer uniform
enforcement of laws over selective enforcement practices. Despite these ﬁndings, there
were few signiﬁcant attitudinal differences between male and female ofﬁcers. This led
Worden (1993: 229) to conclude that, “[t]he most striking ﬁnding in this study is the
failure of gender to explain much or any variation in the array of attitudes examined, even
when potentially confounding variables are controlled.”

The above statement summarizes much of the work that has tested gender effects
on ofﬁcers’ occupational attitudes. Of the remaining eleven studies, most report weak or
null results and many indicate inconsistent ﬁndings. With respect to ofﬁcer attitudes
toward the police role for example, some studies report female ofﬁcers to hold broader
role orientations in general (Sun, 2003) or hold more positive attitudes towards
community policing (Schafer, 2002), while others note that female ofﬁcers hold more
negative attitudes towards order maintenance roles (Paoline et al., 2000). Other studies
report female ofﬁcers to prefer rigid enforcement of the law (Worden, 1990; cf. A.
Worden, 1993) or to hold more negative views towards the use of force (Brooks, Piquero,
& Cronin, 1993). Yet, other studies have found no gender effect on ofﬁcers’ attitudes

towards management or citizens (Sun, 2002; cf. DeJong, 2004; A. Worden, 1993).

23

 

.95:va ..:o_55::5

:5 5m :9: :23 55585 :5 : 55555.3:
5:585 55 95 .555.— .m5oc3 :23 :55585
59559:: 55:9: 3:5 555: 53 .:9: .=55>O:

55520 3 5355
558.55
55535553:
553250 53%

55_>5m 3 Ewcoq
.:o_55::m 5:50 55¢

@850 35qu
8:95:35: N

page
.5 5 5:35:

 

.255. 3:55.555 E :o:5_5> 9: 3 £5: 33 :535
55:59; 235525: 9: 55353 .:o:55.5_55

:2. o. :555: 35355: 55355 .3 555% 55550
JMWNJB 3:55:55 :5“ $3535 5 ..55:5_5>
9: 3 m5. 3 .xb 3555355.. 553593 5.: .5533:
.:o:55.5_55 .3 £5.35. :9: 2 55555 3353335
553 55355 3 ~:w:5_ :5 .55 .535 55550
.85

.m8 ..:5EE55 595:5 3 >85 5.: E :o:5_5> .35
3 :59: :535 9 535 3 5.5.5.: 5.: 53.53::
wExEm 50E: 9: .5535: .9535 9532 5:33:
53:55 53E :9. 555:3 5:33 9.955335

:o:55.5_5m :2.

3:55:38 :3.
5:83:36 3 5355
555: 3 5355
555:0 3 5355
m:o:2:55m 5:51—
b_>:55_5m

55_>5m .3 9:554
.:o_55::m .5950 655%

55_>5m

3 5951— 525525
.59... .5950 .83:

55m>5m .3 £954

aaoEo oauzv
35.5555: _

385.5 Xumuzv
9:53.555: 3

555:3 ovnnzv

8%: a a 8.5

Awaa_5
:33: a. 5555

 

 

535555 55.: 8 .255:— mm5_ 555:3 552 5:35:50 53% .5w< .*5::50 55% 8:53.55: vm 5 $3: .5953 .<
.82 was :32 53 55:5_5> 553.: 9539 5355
3:535 9: :5 5:53 5553 5:25—5:35 9: £5535 b_>:55_5m 55_>._5m

553 5:25.589? 5735595: :5 59:95
55:3 55:05 :5535: 9.29355: 9:8 5:55..
5% .53 .825

>5 E 59533 9: 3 53:: 5:535 .5535;
625525 3955535 5255.55 :5 53:85

:8 55:55:53 555% 95 £3535: 55. 95539
@353 53:55 .53: 553: 9538 5353 35:5: :3:

955:0 3 m35_>
3:59.250 53%
5532

5353595 3 5355
53:835.
b_>:55_5m
5,3335% 55:

3 5554 65.5535
.5w< 5:50 555%

55_>5m 3 9:954

@850 GTE
$55.55: N

555:3 com. T20

23: a a £85

 

 

9 355.: 53.: 3:55:3w5 555:3 55525 55:50 353250 53m .*:o:55::m .5350 .553— 358935: 5.5. 5 833 5:33
.255. 3:55.53
:_ 3:55.» 9: .3 3: 3:5 5:535 35355255
55:25:55.5 55:3 .5 553 5553 555:5 55_>5m
9: .5535: 555353 955:3w5 5: 8 958 3 9:554 .:o:5::m 555:3 mom. TZV Amway
:5 553 55355 3 £52 95 .55 .555: 555550 3:55:58 3.. .5w< .5950 .555”: 3:58.555: 0 595052 a. 3:3
$59.; 355.5% 595:3. 5v— mo:m_555.5__0 .5550 555 53.5% .355 Sam

 

59:53. 55550 :5 5255555520 5ch 5:239: .3 55cm— 9: 9.33.83: 5:55 "ﬁm 53:.

‘

24

5553. 3 535.5.» 353.3 53:85 5553:. 5535: 55:3 *

3:55 555959352 3.53: :5 55.3.5 9: 35: 55: 553.55 55.:3m 5

5.5m 555.?5m 55.3: 9: 35.... 55: :55355 535% 5

 

.wEozon. 33:39:55

 

 

 

 

 

 

5: 555523555: 3 3 53535555 555:3 55.?5m 555:3 EH2.
E 55359.5 E535 5: :25: 5.555555 55.....0 A.00 3 535.> 3 £35.. 5% 55.55::m. 35:99.5: . @553 5:9". 5% 951.
555.5 59535: 535:» 53.59. :3: 5: 55.>5m 555:3 unvuzv
3.55... 533 3.359.335 555:3 555.555 53$. 353.0 3 535.> .3 5m5u. 5% :5:55::m. 55.3: 5:252 55.3. $558 55M 5% :553.
II 55.?5m n
.31. was ..:5:55.5:55 55.5 3.955595 3 9:35.. .:5:55::m. 555:3 v. ....I.2.
5: 55:3 59.. 3 55 559355535? 5550 3.55533 55.. .5w< .5350 .553. 5395.55: 5. .5558 5.50
555:5 3 3:55:53
9. 5: 3.55... 5:53 3352.235 5535: 55:55 55.?5m .3 535.. 5535555 mm..5uzv
9:.3 555:3 5.535: :55 555:3 5.53 5:23:52 353.0 3 5355 .:5:55::m. .*5::50 .555”. 5353.555: m 9.35:3 35.50
555:3 53.3 :9: 35.5.55: .55. 9532 wE>3m 35.55:. h.
5:535 53:55: :3: 5: 3.55.. 533 :5 .35 35.5353. 55:.
35535 E 5355.55 555. 5:5:535u5 5.55:5:559” b.>:.55.5m 55_>5m 3 9:951. 555:3 womuzv
:3: 5: 3.55... 533 9.353335 555:3 5.555 3535.50 5.5M. .:5:55::m. .5950 .*555.~. 395555: . 5308. ::m
3530 3 m35.> |
.8w 39:55:53. 3 m35.>
.wav 5:535 5:5:555555 555:3 E :5:5_5> 5353553 3 5355 55.>5m 3 535.. 555:3 mmmnzv
9: 3 5:3 E535 5.355555 55.. 9: 3 5:52 555: 3 5355 .:5:55::m .5350 .555”. 395.555: . 955.com. :5m
:00 3 5353555555 .5553 9532 5:235 555:3 372.
52:55: :3: 5: 3.55... 533 555:3 5.535”. n.00 3 m35.> :5:55::m .553. .5350 353.555: . Amoomv 53:5m

 

.:5339 59:35:55: 355_...:w.m 52
5w:.::.m 355.5%

:5:55.m:5m 55..
855:3 >3.

55.>5m

3 9:35.. .5w< .5950
55.5.55330 55.....0

5.85.5 oEuzv
5:55.30

55.35 555 525.553. N

5:5 533% >.:=.m

:85. .5. .5 5255
:55

353 85:52 38:55 .8 8.55.2225 52.55 36.25.: 5.. 55:: 2: 55558.: 8.555 .3 535,—.

25

A similar pattern is found when examining the role of gender in explaining ofﬁcer
job satisfaction. Many studies have reported no effect, indicating that gender is not
related to ofﬁcer satisfaction (Carlan, 2007; Davey, Obst, & Sheehan, 2001; Hunt &
McCadden, 19895; Zhao, Thurman, & He, 1999). Only a single study reported a gender
effect on ofﬁcers’ job satisfaction. Dantzker and Kubin (1998) found male ofﬁcers to
have higher levels of job satisfaction than female ofﬁcers. Although, signiﬁcant, the
authors note the relationship was weak as the combined effect of ofﬁcer demographics
explained only 6 percent of the variation in satisfaction.

Bag

Twelve studies used ofﬁcer race as a predictor of occupational attitudes. Only
Sun’s (2003) analysis of 398 ofﬁcers from the POPN study, however, focused
speciﬁcally on the effect race might have on ofﬁcers’ attitudes. The study examined
differences between black and white ofﬁcers on their views toward the police role, legal
restrictions, and selective enforcement practices and reported three signiﬁcant ﬁndings.
Black ofﬁcers were more likely to hold broader role orientations, to hold more positive
views towards legal restrictions, and to hold more positive views toward selective
enforcement of laws. Although the effect of race was one of the strongest reported across
the three attitudes, the combined effect of all variables included in the analysis explained
only between 6 percent (views of legal restrictions) to 16 percent (role orientations) of the
variation in attitudes. Despite this, the ﬁndings did indicate that there were some racial
differences in ofﬁcer attitudes (Sun, 2003).

Results from the remaining studies, however, highlight a number of

inconsistencies for the effect of race on ofﬁcer attitudes. For example, some studies

26

report ﬁndings similar to Sun (2003) with respect to selective enforcement of the law
(Paoline et al., 2000; Worden, 1990), but not in all cases (cf. A. Worden, 1993).
Additionally, ﬁndings from these studies indicate that minority ofﬁcers have more
positive views towards the legal restrictions placed on police work (Worden, 1990; A.
Worden, 1993) but that ofﬁcer race has generally little affect on their views toward
citizens (DeJong, 2004; Paoline et al., 2000; Sun, 2002), supervisors (Sun, 2002), or job
satisfaction (Carlan, 2007; Dantzker & Kubin, 1998; Hunt & McCadden, 1985; Zhao et
aL,l999)

Another inconsistency often reported in the literature concerns the effect of race
on ofﬁcers’ views toward the police role. While Sun (2003) found minority ofﬁcers to
hold broader role orientations, this ﬁnding was only supported by one other study.
Paoline et al. (2000) also found minority ofﬁcers to hold more positive attitudes towards
order maintenance and community policing roles. Other studies, however, have reported
no relationship between ofﬁcer race and attitudes toward the police role in general (e.g.,
broad versus narrow orientations) (Brooks et al., 1993; Worden, 1990) or more
speciﬁcally, such as acceptance of community policing (Schafer, 2002). Thus, as with
much of the research reported in this section, the effect of race on ofﬁcer attitudes
remains mixed and unclear.

Education

All but one of the studies (i.e., A. Worden, 1993) included a measure of ofﬁcer
education as an explanation of occupational attitudes. Although the educational measures
used varied somewhat across studies (e.g., highest degree earned or comparing college

educated ofﬁcers to non-college educated ofﬁcers), all tested the effect of higher

27

education on ofﬁcers’ attitudes in some manner. Of the fourteen studies, only one
focused speciﬁcally on educational effects. Worden (1990) re-analyzed PSS data to
determine if ofﬁcers’ highest degree earned impacted their occupational attitudes. The
results indicated that ofﬁcers who had earned bachelor degrees were signiﬁcantly more
likely to hold narrower views toward the police role (i.e., a crime—ﬁghting orientation),
more positive views towards legal guidelines, and to have signiﬁcantly greater
preferences for autonomy and selective enforcement. Once again, although statistically
signiﬁcant, the effects were small. As Worden (1990: 584) noted, “to dwell on the
handful of statistically signiﬁcant relationships that emerge from these analyses would be
to overemphasize the impact of education on ofﬁcer attitudes. . .little of the variance in
any attitude is explained by education.”

Results from the remaining studies on ofﬁcer education mirror those discussed for
ofﬁcer gender and race in the sense that the ﬁndings are often reported to be weak and,
when examined collectively, highlight inconsistencies. For example, while Worden’s
(1990) study indicated that college educated ofﬁcers held narrower, crime-ﬁghting
orientations to the police role; other studies have reported that more highly educated
ofﬁcers were less likely to favor crime-ﬁghting orientations (Brooks et al., 1993; Paoline
et al., 2000). On the other hand, some studies have reported no relationship between
ofﬁcers’ education and either broader measures of their role orientations (Sun, 2003) or
more speciﬁc attitudes, such as ofﬁcer acceptance of community policing (Lord &
Friday, 2008; Schafer, 2002).

With respect to other types of occupational attitudes, research has indicated that

ofﬁcer education is generally unrelated to their views toward citizens (DeJong, 2004;

28

Kyu~vn

.1»
J

{I

l
r

Moon & Zager, 2007; Paoline et al., 2000; Sun, 2002) as well as to their views toward
supervisors (Sun, 2002; Worden, 1990). Although there is some evidence to suggest that
college graduates hold more negative views towards upper police management, the effect
was small (Sun, 2002). Additionally, of the studies that have examined the effect of
education on ofﬁcers’ job satisfaction none have uncovered a signiﬁcant relationship
(Carlan, 2007; Datzker & Kubin, 1998; Davey et al., 2001; Hunt & McCadden, 1985;
Zhao et al., 1999).

The only consistent, signiﬁcant effect found for ofﬁcer education concerns their
attitudes toward selective enforcement. Worden’s (1990) ﬁnding that college educated
ofﬁcers prefer selective enforcement has been supported in two subsequent studies. Using
POPN data, both Paoline et al. (2000) and Sun (2003) found college educated ofﬁcers to
hold more positive views toward selective enforcement of the law. Although signiﬁcant,
the effects reported explained only a small amount of variation (Sun, 2003). Overall, the
results from this research echo Worden’s (1990: 584) sentiment that there is “little to
support to arguments that higher education will improve the quality of street-level
policing by changing the attitudes of patrol ofﬁcers.”

Age

Only six studies used ofﬁcer age as a predictor of ofﬁcer attitudes and four of the
six examined age in relation to job satisfaction (Carlan, 2007; Dantzker & Kubin, 1998;
Davey et al., 2001; Hunt & McCadden, 1985). Although a signiﬁcant effect was reported
in three of the studies (Carlan, 2007; Dantzker & Kubin, 1998; Hunt & McCadden,
1985), it should be noted that for all three studies the combined effect of all demographic

variables explained only four to six percent of the variation in satisfaction levels. Thus,

29

gr 1

J .
yaks

r——<
r11
1

l

(-‘.
.«

,4.

Vv—‘

CA.‘

L
"I
.lJ‘:

ofﬁcer age appears to be only weakly related to job satisfaction. With respect to other
ofﬁcer attitudes, ofﬁcer age has not been found to be related to ofﬁcer role orientations
(Brooks et al., 1993) or views toward citizens (Worden, 1993). Older ofﬁcers, however,
were found to hold more positive attitudes toward legal restrictions and greater
acceptance of authority, as well as more positive views towards their peers and their work
environments (Worden, 1993).

Despite the signiﬁcant relationships reported by the above studies, it should be
noted that it is difﬁcult to separate the unique effects of age from those associated with
ofﬁcers’ years of experience. For example, Sherman (1980) has pointed out that while
younger ofﬁcers may hold more aggressive attitudes toward policing, their attitudes may
also reﬂect experience or training levels (Sherman, 1980). The difﬁculty in isolating
these effects is likely one reason why ofﬁcer age has been included in so few of the
studies reported in Table 2. ‘1.

Length of Service

The ﬁnal individual characteristic that has received considerable attention as an
explanation of ofﬁcer attitudes is length of service (i.e., years of experience). All but one
of the studies included this characteristic as a predictor of attitudes. Much like the results
for the previous ofﬁcer demographic characteristics, the ﬁndings reported by the studies
also point to weak and inconsistent relationships. Once again, the most inconsistent
results are reported for the relationship between length of service and ofﬁcer role
orientations. For example, some studies found less experienced ofﬁcers to accept broader
role orientations (Sun, 2003; Worden, 1990), while another study indicated the opposite

(Paoline et al., 2000). Other studies, however, reported no relationship between length of

30

service and ofﬁcer attitudes towards the police role (Brooks et al., 1993; Lord & Friday,
2008; A. Worden, 1993).

Such inconsistencies are also illustrated when examining other types of attitudes,
such as views toward citizens, policing tactics, and views toward supervision. With
respect to length of service and views toward citizens, some studies have found less
experienced ofﬁcers to be more distrustful of citizens (Moon & Zager, 2007; Sun, 2002),
some have found more experienced ofﬁcers to be more distrustful (DeJong, 2004), and
others have reported no relationship (Paoline et al., 2000). Although most of the studies
report no relationship between officer length of service and views toward legal
restrictions (Sun, 2003; A. Worden, 1993; Worden, 1990), results for attitudes toward
other policing tactics, such as selective enforcement and aggressiveness, are not as clear.

Of the four studies that examined length of service and selective enforcement,
three found less experienced ofﬁcers to hold more negative attitudes towards selective
enforcement (Paoline et al., 2000; Sun, 2003; Worden, 1990). Only one study has
examined the effect of length of service on ofﬁcer aggressiveness and found an inverse
relationship; ofﬁcers with more experience were less aggressive (Paoline et al., 2000).
Finally, length of service has generally been found to be negatively related to views
towards supervision (A. Worden, 1993; Worden, 1990), but not in all instances (Sun,
2002)

Additionally, ﬁve studies have examined the effect of ofﬁcer experience on job
satisfaction with mixed results. Two of the studies found no relationship (Carlan, 2007;
Davey etal., 2001). Although the remaining three studies found more experienced

ofﬁcers to be less satisﬁed. with their jobs, the effects accounted for only small

31

percentages of variation in satisfaction levels (Dantzker & Kubin, 1998; Hunt &
McCadden, 1985; Zhao et al., 1999). Thus, when viewed in relation to the effects of
ofﬁcer characteristics as a whole on job satisfaction, the results are mostly unsupportive.
As such, it has been noted that ofﬁcer demographic characteristics “are of little value to
understanding job satisfaction” (Carlan, 2007: 74).

Summary of Research on Individual Characteristics & Ofﬁcer Attitudes

 

Empirical research that has sought to explain differences in ofﬁcers’ occupational
attitudes has placed an emphasis on the inﬂuence of ofﬁcer demographic characteristics
and length of service as predictors. Based on the notion that the occupation of policing
has become increasingly diverse with the inclusion of more women, minority, and college
educated ofﬁcers, these studies have examined the effect that these individual
characteristics along with ofﬁcer age and experience have on ofﬁcers’ occupational
attitudes - the rationale being that “[a]s police forces have become more diverse, then,
one might expect to ﬁnd greater variation in ofﬁcers’ adaptations to their working
environment and greater covariation between their [attitudes] and their backgrounds”
(Paoline et al., 2000: 584). In all, from the ﬁfteen studies that examined the effect of
ofﬁcer characteristics on their occupational attitudes two important conclusions can be
drawn.

First, the results reported by these studies point out a number of inconsistencies
with respect to the nature of the relationship between individual ofﬁcer characteristics
and occupational attitudes. Although inconsistencies between the effect of ofﬁcer
characteristics and their attitudes towards citizens, policing tactics, supervisors, and job

satisfaction all existed, the most notable discrepancies occurred when looking at ofﬁcer

32

attitudes toward the police role. For example, discrepancies were noted for the effects of
ofﬁcer gender, race, education, and length of service on ofﬁcer role orientations as some
studies reported no relationships while others reported signiﬁcant relationships.
Furthermore, even studies that reported signiﬁcant relationships often reported results in
opposite directions (e. g., the effect of length of service on ofﬁcer role orientations).
These inconsistencies make it difﬁcult to determine the exact nature of the relationships
between individual ofﬁcer characteristics and occupational attitudes.

Second, the results indicate that ofﬁcer characteristics are rather weak predictors
of ofﬁcer attitudes. Even though this set of studies has found ofﬁcer characteristics to
signiﬁcantly predict attitudinal differences among ofﬁcers, the exact contribution these
factors offer for explaining variation in attitudes is small. For example, the studies on job
satisfaction note that ofﬁcer characteristics explain four to six percent of the variation in
satisfaction levels which calls into question the effect of ofﬁcer characteristics on ofﬁcer
attitudes (Carlan, 2007; Dantzker & Kubin, 1998; Hunt & McCadden, 1985; Zhao et al.,
1999). This sentiment was mirrored by many of the studies reviewed here (e.g., DeJong,
2004; A. Worden, 1993; Worden, 1990).

The inconsistent and weak ﬁndings regarding the effect of ofﬁcer characteristics
on their occupational attitudes has led to calls for research to begin exploring other types
of factors that might explain attitudinal differences among ofﬁcers. Since individual
characteristics have been found to have little effect, researchers have noted that
differences in ofﬁcers’ work environments, such as organizational and environmental
factors, should be examined as potential explanations of ofﬁcer attitudes (Brooks et al.,

1993; DeJong, 2004; Paoline et al., 2000; Sun, 2002, 2003). Although, studies have

33

a.“ '

\.

w"

 

"11

h

\

ll"

lu

g.“-

chm

begun to explore the effect of such factors (e.g., Hassell, 2006; Moon & Zager, 2007),
research in this area is currently underdeveloped. The rationale for taking such a

perspective is discussed later in this chapter as well as in Chapter Three.

Attitudes as Explanations of Officer Behaviors

Research on the police attitude-behavior relationship has also taken an individual-
level approach to explaining ofﬁcer behaviors. For example, it has been proposed that
ofﬁcers who hold narrow orientations towards the police role, who are aggressive, and
who are distrustful of citizens are more likely to use force (e.g., McCluskey et al., 2005;
Terrill & Mastrofski, 2002; Terrill, Paoline, & Manning, 2003; Worden, 1996), to make
arrests (e.g., Mastrofski, Worden, & Snipes, 1995), or to conduct searches (e.g., Gould &
Mastrofski, 2004; Terrill & Paoline, 2005). Additionally, ofﬁcer job satisfaction has also
been argued to be an important component for understanding ofﬁcer behavior. The more
satisﬁed ofﬁcers’ are with their jobs, the greater the likelihood that they will strive to
achieve the goals of the department. As the National Research Council (2004: 134) noted,
“maintaining adequate levels of job satisfaction is essential for sustaining good or
satisfactory police performance.” Thus, ofﬁcer attitudes have also been deemed important
predictors of ofﬁcer behaviors.

This section reviews empirical studies that have examined the effect of ofﬁcer
attitudes on behavior. To be included in the review, only studies that tested the impact of
attitudes on behavior using multivariate analyses were sought. All studies included in the
review have controlled for some combination of situational, organizational, or
Community level factors which also affect ofﬁcer behavior (NRC, 2004; Riksheim &

Chermak, 1993; Sherman, 1980). Additionally, the outcome of interest measured some

34

39‘
ml-

“1‘

I!

w...

4
1M1

 

 

‘ in};

EU

\">

“K;

LL'

type of observable or self-reported behavior. Studies that used ofﬁcer intentions to
engage in behaviors, such as vignette scenarios, were excluded as prior research has
noted that factors that explain behavioral intentions are not the same as those that explain
actual behavior (Ajzen & Fishbein, 1977; Schuman & Johnson, 1976). In all, eighteen
studies ﬁt these criteria and are included in this review. Table 2.2 provides a summary of
the studies and includes the type of behavior under study, the attitudinal measures
hypothesized to impact the behavior, study unit and sample size, and a summary of the
relevant ﬁndings. The following discussion also highlights the research ﬁndings by
behavioral area.
Use of Force

Six studies tested the effect of ofﬁcers’ attitudes on their use of force behavior.
These studies examined a wide range of attitudes including job satisfaction, views
towards citizens, attitudes towards force, ofﬁcer role orientations, views toward legal
guidelines, and aggressiveness. The studies do use different measures of force behavior,
ranging from the use of reasonable and improper force (Friedrich, 1980; Worden, 1996)
and the highest level of force used in encounters (McCluskey et al., 2005 ; Terrill &
Mastrofski, 2002; Terrill et al., 2003) to predicting speciﬁc types of force, such as verbal
force and impact force (Novak, Smith, & Frank, 2003). Each of the studies, however,
focused on using attitudes as explanations of force behavior.

For example, Friedrich (1980) examined the effect of job satisfaction and views
toward minority citizens on ofﬁcers’ use of reasonable and unreasonable force. Job
satisfaction was found to be unrelated to use of force behavior. A signiﬁcant relationship

was found between ofﬁcers’ views toward minority citizens and their use of force. The

35

.41 ‘

111‘.

Clint

ﬁnding, however, was weak as Friedrich (1980: 95) noted that there was only a “slight
tendency for the more prejudiced white patrolmen to employ more force against black
citizens.” Two additional studies also examined the relationship between ofﬁcers’ views
toward citizens and their use of force. While Friedrich’s (1980) measure was an indicator
of ofﬁcer prejudice against minority citizens, these studies employed measures of citizen
distrust and perceptions of citizen hostility. Similar to Friedrich’s ﬁndings, Worden
(1996) found a signiﬁcant, but weak relationship between citizen distrust and use of force
behavior. Distrustful ofﬁcers were more likely to use improper force. Ofﬁcer distrust of
citizens, however, was not found to be related to the severity of forced used in citizen
encounters (Terrill & Mastrofski, 2002).

Four studies also examined whether ofﬁcers’ attitudes toward the police role,
force, or aggressiveness affected force behavior. While Worden (1996) reported a
signiﬁcant relationship between attitudes towards force and force behavior, this ﬁnding
was also weak. Studies examining ofﬁcers’ subscription to a narrow, crime-ﬁghting role
orientation reported no signiﬁcant effects on force behavior (Worden, 1996; Terrill &
Mastrofski, 2002). Similarly, ofﬁcer views toward aggressive policing were also found to
be unrelated to their force behavior (McCluskey et al., 2005). Thus, ofﬁcers’ attitudes
toward the police role and policing tactics have generally been found to be unrelated to
use of force behavior.

The effect of ofﬁcers’ attitudes toward legal restrictions and use of force behavior
has also been largely unsupportive. For example, Terrill and Mastrofski (2002) reported
no signiﬁcant effect between ofﬁcers’ beliefs regarding overlooking legal guidelines to

conduct their job and the severity of force used in encounters. Similarly, ofﬁcers’ views

36

 

'1‘“ .,

AL

I"!

.1..-

\\1

,u
I.|§

.—

Ubi-

"311,-

r“,

“L.

l l

.lri

a5”

 

towards civil liability were found to have only a weak effect on ofﬁcers’ use of verbal
force and no effect on their usage of impact force (Novak et al., 2003).

While the above studies examined the impact of each attitudinal measure
separately on use of force outcomes, one ﬁnal study combined a set of attitudes into a
single measure to distinguish subgroups of ofﬁcers. As will be discussed in the next
section, many of the aforementioned occupational attitudes have been described as part of
the police culture. Terrill, Paoline, and Manning (2003) examined whether ofﬁcers’
adherence to the attitudes associated with this culture affected their use of force behavior.
Using cluster analysis, ofﬁcers with similar dispositions on a set of attitudes (e.g., views
toward citizens, views toward supervisors, views toward the police role, and views
toward policing tactics) were grouped together. They found that ofﬁcers in groups that
held attitudes in accordance with the culture were more likely to use force in citizen
encounters than ofﬁcers in groups whose attitudes did not align with the traditional view
of culture.

Decision to Arrest

Four studies tested the effect of ofﬁcer attitudes on the decision to arrest. Smith
and Klein (1983) were two of the ﬁrst scholars to examine whether ofﬁcers’ attitudes
were related to their arrest behavior. Although their primary research interest was in
identifying the impact of organizational-level properties on the probability of arrest, the
authors also conducted an encounter-level analysis for 950 police-citizen encounters to
control for individual factors. Included in the analysis was a measure of ofﬁcer job
satisfaction. Results from the model, however, reported no signiﬁcant relationship

between ofﬁcer satisfaction and their decision to arrest.

37

 

.343 .wmv :280 0:88 0:: E 080:0E

:8 :08 :05 mo. 50 8:85 a a: «8:8 50 8:0

0:: .0888 :0 86200: 889820 E 033:?

2:: 50 8050 of. .8080 :m 300—: 80— 05 .wE0:0m

8:25:00 0830: :850 0:: 03:8: 0:08 06..

Aw».— .wﬁ 10305 b98298 £80.: 0:: 8 2::

0:58:00 5:: anew a 8. . .8388 582880

:8 3:588 50 8:: :8: $2 :88 E 88::

E05 50 87. 0:: :wzoﬁﬁ $9482.30: 885830

3.8888 30:0 8:38 800-0330: 0::

E 880 0:: 08 9:088:88 50 3:50:50 .008

802:6 E 88050: 8830: 285.: 080m:
.3? was :ngqaﬁ: 0:05 0:20 8003:0900

E05 :85 @0588 0:9: 08 0:3 8850.. .N
.8? .wmv :88:an0 50 8::

E05 805a 8% E05 @830: 88:8 .2850: .:

LOU 50 32>
808080.80 ~DQ
50 8828830:
50 0.303 .8850 .N
8088880
5D 50 050:0:
13:08:58:
.3850 .—

88885
82385 .N
838588 :2. A

80t<

68820.80
52.2 50

.0885:
825 0E5.

221.: 32:85

@816 .850

3:3 “20 85

A38 .3 a 30832

:8: a a 02882

$8: 880 0 5:20

 

.03: .w& ..m8:0>_m8:wwm E05

:0 8:85: 2:: 8:3: 08: 88:8 .8850:
.200 .w& :88:me E

8:: E 85.0: 88555 3:85:90 08550:
02:? 35.8.5805 0:08 08 0.030808 3
95 8:00 0:: 3 8:083 08 8:0: :2: 08:0:
0:? 8850 .8835. . 02888 0:08 08
35088: 08 88:6 8:: 08:0: 0:3 2850:

3:828 .m
808885 Ewe:— .v
80858— Bwoq .m

88:5 .m
8882.0 0.05 .—

88028833:
5:58-05

$883823)": ”own HZV
9:38:05 ”EmuZV

.850

«3?: SEE:
«53: 22: 0 5:5.

 

 

52808.0: EmocEwi 02 8:850:90; :2. “motxx Smonzv 883005:
.30 was :88E0
0.83 H888 8:05 0.5:: >2qu 0: 88—088
8:»: 8060.08 0:08 0:: :05 x8082 Ew:m: .N 8:03.85 .N
8:80:20: 885880 02 ._ 88.583 :2. ._ 8:0“. A 3o. T7: :2:=00:m Swm C :Eﬁotm
me—EE 0:56.3— moczzzax 50:22—3— oEm 0.95% 0% 3:: hczum zuam

 

3:805:01 ..0_>a:0m-0_055< 855 05 wEEEaxm— 00:35 "N.N 030R.

38

 

NI:III.1I Ilr I 1!.I.I-..

 

 

6%: .wmv ..:E>m:0: E0:: :0>0 00:0:cE
8808 m 08: wE>_0m 8030:: :0: 0028::
.m:0m_>:0::m E0:: .:0 80::00:0: .8850: .m
.:_:80::_0: 2:85:30 02 .N
8:80:00: 885:8: 02 .:
.88:
.w:: ...m:00:80 :0>0 80:00 030000 E 880056
00:60:: 05:60 00:0: :0 303 38:60::
0:: 6:030: 880856 3:60:60 .8850:
6:800 :05 800:0: .88:_0 :EﬁmT 0: .200:—
0:0E .:00 :0: 53:00:: 8:88 :EB 0:850

.Ammm was ...00:0.: 0: 60:20: 38:05:80
0:03 00:88:: 063:6 00E: 0:: .:0 0:02:
03:80:20: ::8_.::w_m 02 .N
.33 .w:: .1030:
50:80—90 £069: 0:: .:0 8:08: 80:08: 0::
:0: 6083000: 5:00: :0: :0 E: .:0 >03 0:: 6:0

868::
.m:0m_>:0:=m .m
88:5 .N
:0:S:0_:0 0:0: ._

0:315 00:0:

:00 a: 53:85
08:88: :80: .m

88:5 .m
:0:S:0_:0 0:0: .—

wEZOm 8030::
:0: 3888830:
:0 @503 .2850 .m

wEZom 8030::

00:0 :

$00300: :8:5
0: 880:8”:

00:0 :

AomHZ: :0m_>:0::m
Amvmuzv :0050
Emuzv :20

Ammmﬁuzv :0::300:m

Gang
60:0 8050:2880:

G: :.mHZV 88:08.:

3:80.: 5:083 a. :80

: Amoomv :0 :0 580:.

.0 :80: .3 :0 82:32

:80: 82:32 a £5:

 

 

.0: 630:: 0:03 00:0: :83 50 32> 788:0: wEZom
.8850 0:: 8:850: 830:8: .:0>0>:0: 8030:: :0: 50:0:
...60:0E..:06 830:0.» $5388.88: 23:08:88:
.80:E06:800 8:900:00 8:05:08 :0 :0: ._ .8850 ._ wEZom 8030:: clouzv 5_:m :283 :a :0 8030
A3 .wmv ..b:8080::= 6::
3:80.80: .80.: 08 0: 30:: 0:8: 0:: 88:6
6:030: 862:2: 02:80: 0:08 :85 0:850: .m
.?m .w:: 30:03:00: :0::0:w 88:5 .m
:85 38:08:: 00:0: 08 0: 6:0: 00:05 .:0 08 00:0:
0:: 6:030: 863:8 03:00: 0:08 :65 0:850: .N 6:030: 063:3. .N
.:E:80:0_0: 885:8: 02 .: :0:S:0_:0 0:0: .: 00:0..— Amcvuzv :0050 a 3:2 .35: 86:03
amicﬁm 2:30—3— :0:=:_::< 20:52:3— 0N_m ENE—am 0% 3:: 56:5 Exam

 

3:00: 3:822:01 :0_>::0mu0—0=:_::< 00:0: 0:: WEE—wam— 865m :N.N 03.3.

39

x603 800838: 62 86:0: :0 80.00:: 0:: 80:: 0:06 68380 :0_60:m :
>605 :003:0m 00:0: 0:: 80:: 0:06 683080 :0_60:m 0

 

.83 was ..:00_0:0 33:00 6080058
0>0: :08 86:0: 3808800 ::0_ 88:8: 0:
80:: 803 ::0:3:0:0: :0:: :0:: 800::0 :00: 080:

 

x: :0::00:0: 0:: 6:0 :::0: 608—8088 80:8: :00

0: 80:: 803 88:6 :0:: :0:0: 0:: .8850 0:::0::.

:00: :0 : .:0:3:00 60:0_0:-08_:0 ::0_ 0: 60:06: 00:02.8: 003:0m

3800508: 0:03 86:0: 3808800 6:030: :8:5 .m 08:0

86:80 03:8: :0:: :0::00x0 0:: ::_3 .8068 00:00.3: 8:02 880

33:00 8050 3808800 0: 60:06: 3:000:88: :0:3:0:0m .N 3:0:
:0: 0:03 88:88 06880 :0 3:008 0: :.: :00 :0 32> .: 80 80:: 08$: 3:012:33 $003 ._0 :0 ::_8m

:8: 0::

..::0:0>_::0:ww0 :303 :00:w :00: 0:: :033
0: 00:86 0:: :0 60:0: 00:0: :0 :0:00:::0
0:: ::00::0 3_0::0:0:::6 880:8 :00:w

 

 

 

:00:: ”808:: :00::0 8:086:00 88:88: ::0:0>_::0:ww<
.:0>030: :00::0 8086 88:83: 02 .N :00: .m .
8:80:06: 88:88: 02 ._ ::0:0>_::0:ww< ._ 00:0: £8. 72: 88000:: : $003 .0 :0 :0:—8502
8080080 :0:: 03600:: 8 8:800:
:006000 0: 30:3 0:08 3:80:88: 08:00 00:0: :0:0:00m :0:m

::_3 60:w:0 3.086.800 0:08 0:03 0:3 80050 80:00 00:0: 058:: 03800:: onmuzv 88000:: :Amoomv 380:: 0:: 8:00:
822282 2802:: oz :00 :o 32> 8:88: 0; T2: 5.5.8:: 080: 008.32 :0 2.50

.2 .w:: ..::8:6 0: ::0:800 8630838050 88:5 )
0: 60:06: :0: :.: 88:6 :00:0 068::0 8050: 88:5 w:::0:800 Amm : 0H2: 88000:: :AVOONV 0:030

 

.00: 0:: 08:22:
:0>0_-:00::: 3:30. :00::0 :0: 06 :0::00:0:
_36 :00:0 ::0:0::.3:30: :0 8:00:00 :: 80:::0

 

_0::0:::0: .80 0: 608206 0: 0:800: :0: 06 3:305
08:80:: 80380: 038880 60:03:: :00::0: _35 6:030: :30_> ::0:0>_::0:ww< ARV—HZ: :00::0 383 :0 :0 :0>0Z
$15—05...— ::0>0_0~_ 8:00:54: 830—30 85 0.:80m 0% :8: Exam .5005

 

0830:5830: 332008502 8:0: 2:: 8:08.08 8:03: "0.0 030:.

40

Research has also examined the effect of ofﬁcers’ views towards civil liability
and the decision to arrest. For example, Novak et al. (2003) hypothesized that ofﬁcers
who believed that they should be exempt from civil liability would be less inclined to
make arrests. Using interview and observational data from 147 ofﬁcers, they found a
signiﬁcant, but weak effect. Ofﬁcers who believed they should be exempt made
signiﬁcantly less arrests. Another signiﬁcant relationship has been reported between
ofﬁcers’ views towards DUI enforcement and their self-reported DUI arrests. Using
expectancy theory as a framework, it was found that ofﬁcers who believed that DUI
enforcement was instrumental to their career advancement and ofﬁcers who believed that
the rewards of DUI enforcement outweighed the costs were signiﬁcantly more likely to
make DUI arrests (Mastrofski, Ritti, & Snipes, 1994). The relationships were weak in
magnitude as the authors noted, “intrinsic rewards (reﬂected in attitudes about uniformity
of enforcement) and the costs in the reward-cost balance show statistically signiﬁcant
relationships, although. . .as a group they contribute little to the model’s explanatory
power” (Mastrofski et al., 1994: 138).

A ﬁnal study examined the effect of ofﬁcers’ attitudes toward a speciﬁc aspect of
the police role, community policing, and their arrest behavior. They found that ofﬁcers
with more positive views towards community policing were signiﬁcantly less likely to
make arrests. As the authors concluded, “the effect of this variable is substantial,
reducing, on average, the odds of arrest by a factor of .63 for each unit increase in the
attitudinal scale” (Mastrofski, Worden, & Snipes, 1995: 551-2). This is one of the few

studies on the police attitude-behavior relationship to ﬁnd a strong effect.

41

l “*9

'7‘
(X.

Hr

UTA

\lsh

Q‘Qrﬂ
» W“

Proactive Stops & Searches

Four studies examined the effect of attitudes on ofﬁcers’ proactive stop and
search behaviors. Similar to use of force and arrest studies on the police attitude-behavior
relationship, these studies have also tested a wide range of occupational attitudes
including ofﬁcer views toward the police role, citizens, legal issues, and policing tactics.
Also similar to the aforementioned studies, ﬁndings from this line of research also
provide mixed support for the police attitude-behavior relationship.

For example, two studies tested the effect that ofﬁcers’ views toward the police
role might have on both stop and search behaviors. Worden (1989) found that ofﬁcers
who subscribed to broader role orientations engaged in slightly more suspicion stops than
those who subscribed to a narrower, crime-ﬁghting role. Role orientations, however, had
no effect on the number of trafﬁc stops ofﬁcers engaged in. In a second study, Gould and
Mastrofski (2004) examined whether ofﬁcer attitudes toward community policing
affected their proclivity to engage in unconstitutional searches. Although the study used
observational data to assess ofﬁcers’ views toward community policing rather than
survey data, the results indicated no signiﬁcant effect between whether ofﬁcers had
positive or negative views towards community policing and their search behavior.

Ofﬁcers’ attitudes toward legal issues have also rarely been found to have an
impact on stop or search behaviors. In a recent study examining the effect of ofﬁcers’
views toward civil liability on stops and searches, the results were largely unsupportive.
No relationship was found between ofﬁcers who believed they should be exempt from
civil suits and their self-initiated stopping behavior. Although ofﬁcers who supported

exemptions from civil suits were found to be involved in signiﬁcantly fewer searches,

42

this relationship was weak in magnitude (Novak et al., 2003). Furthermore, Worden
(1989: 691) found that ofﬁcers who believed that the court system was supportive of their
efforts were more likely to engage in both trafﬁc and suspicion stops; however, it was
noted that “while both are statistically signiﬁcant, neither is large in magnitude.”

Additionally, two other ﬁndings from Worden’s (1989) study bear mentioning.
First, ofﬁcer attitudes toward citizens, while signiﬁcant, was also found to be weak in
magnitude. Ofﬁcers who had a more positive outlook towards citizens were only slightly
more likely to engage in trafﬁc stops. Second, ofﬁcers’ attitudes toward selective
enforcement of the law were found to have no effect on either the number of trafﬁc or
suspicion stops conducted by ofﬁcers. Overall, this led to the conclusion that the
“ﬁndings further indicate that attitudinal variables account for a very small part of the
variation in ofﬁcers’ behavior” (Worden, 1989: 668-9).

Only one study that has examined the police attitude-behavior relationship on
ofﬁcer pro-activity outcomes has uncovered a substantial effect. Paoline and Terrill
(2005) examined the effect of ofﬁcers’ attitudes in relation to ofﬁcers’ search behavior
stemming from trafﬁc stops. Using a classiﬁcation scheme similar to the one constructed
for their study on police culture and coercion (see Terrill et al., 2003), the authors
reported a positive relationship with ofﬁcers whose attitudes were more aligned with the
police culture to engage in signiﬁcantly more searches. In contrast to previous studies
that found a signiﬁcant effect, the results reported here were much more profound as it
was noted that “encounters involving ofﬁcers with pro-culture orientations were over two
and a half times more likely to conduct a search in comparison to encounters involving

con-culture carriers” (Paoline & Terrill, 2005: 467).

43

r];

l

y.

[3'
.\u

" '.' ‘n
I‘IU“

Willi}

'33.! ,,..

I H“.

CQTLHJ'

Kw
[H‘i .
Jxmﬂk

I» .
..i:.~.c Eu

Service & Other Discretionary Behaviors

A ﬁnal set of studies has examined the relationship between ofﬁcer attitudes and
service or other discretionary behaviors. These studies have focused on ﬁve behavioral
outcomes: time spent problem solving (DeJong, Mastrofski, & Parks, 2001; Engel &
Worden, 2003), providing comfort to citizens (DeJong, 2004), responding to citizens’
requests during encounters (Mastrofski, Snipes, Parks, & Maxwell, 2002), time spent
gooﬁng off or “shirking” (Brehm & Gates, 1993), and overall performance of
discretionary activities while on duty (Smith et al., 2005). Types of ofﬁcer attitudes
expected to have an impact on these behaviors primarily include attitudes toward the
police role, views toward citizens, views toward supervisors, and job satisfaction.

Of the two studies that tested the impact of ofﬁcer attitudes on problem solving
behavior, mixed results have been reported. Ofﬁcers who believed that supervisors
placed a priority on problem solving behavior were more likely to engage in problem
solving activities (DeJong et al., 2001; Engel & Worden, 2003). Ofﬁcers’ acceptance of
broader police roles and their level of distrust towards citizens, however, had no effect on
their time spent on problem solving activities (Engel & Worden, 2003).

Ofﬁcers’ attitudes toward citizens have also been examined in relation to
comforting behavior and time spent on general discretionary activities. DeJong (2004)
reported no signiﬁcant relationship between ofﬁcers’ level of distrust of citizens and their
comforting behavior. Similarly, Smith and colleagues (2005) found little evidence that
ofﬁcers’ views of citizens affected their discretionary behaviors. Thus, the results from
these two studies indicate that ofﬁcers’ views toward the citizenry are largely unrelated to

service and more general behavioral outcomes.

44

Ofﬁcers’ views toward the police role have been examined in relation to ofﬁcers’
responses to citizen requests and time spent on general discretionary activities. Results
from these studies have found that acceptance of community policing roles signiﬁcantly
increase ofﬁcers’ likelihood of granting citizens’ request to control a suspect during an
encounter (Mastrofski et al., 2002) and signiﬁcantly less likely to engage in activities
aimed at addressing crime while on duty (Smith et al., 2005). Although these studies have
provided some evidence that views toward the police role impact behavior, the results,
once again, appear to be weak as both effects were minor in magnitude.

A ﬁnal study tested the effect of ofﬁcers’ satisfaction with their jobs on the
amount of time spent shirking while on duty. Using data from Black and Reiss’s (1967)
study of ofﬁcers within the Washington, Boston, and Chicago police departments, Brehm
and Gates (1993) found that ofﬁcers who were dissatisﬁed with their job or ofﬁcers who
were more satisﬁed with their supervisors were more likely to “shir ” during a shift. As
Engel and Worden (2003) would note ten years later, the results from this study is one of
the few to report a signiﬁcant relationship between ofﬁcer attitudes and their behavior
within this speciﬁc type of behavioral outcome.

Summary of Research on the Police Attitude-Behavior Relationship

Prior research on the police attitude-behavior relationship has examined the effect
of ofﬁcers’ occupational attitudes on four behavioral outcomes: the use of force, the
decision to arrest, pro-activity measures, and service related behaviors. Based on the
notion that variation in individual ofﬁcer attitudes explains variation in street level
behavior, the results of the studies reviewed here have generally found attitudes to

explain little variation in behaviors. In fact, only three of the eighteen studies that have

45

tested this relationship have reported ﬁndings that would indicate a substantial effect
exists (e.g., Mastrofski et al., 1995; Paoline & Terrill, 2005; Terrill et al., 2003).
Collectively, the results have led comprehensive reviews of the factors expected to
impact ofﬁcer behaviors to conclude that ofﬁcer attitudes have little impact on their
actions (NRC, 2004; Riksheim & Chermak, 1993; Sherman, 1980; Worden, 1989, 1995).

Such a conclusion regarding the police attitude-behavior relationship has led
police researchers to offer potential explanations for the lack of signiﬁcant ﬁndings.
Police scholars have often related their ﬁndings to the broader social psychology
literature, which has also had a history of non-signiﬁcant ﬁndings (see Ajzen & Fishbein,
1977; Schuman & Johnson, 1976). Although reviews of the social psychological
literature note that there is a signiﬁcant amount of evidence that indicates the relationship
is theoretically important (e. g., Schuman & Johnson, 1976), the weak empirical evidence
linking attitudes to behavior has made it “hardly surprising that the attitude concept has
come under increasingly strong criticism” (Azjen & F ishbein, 1977: 912). Such a
sentiment has been echoed within the policing literature as many of the studies on the
police attitude-behavior relationship have also pointed out this limitation (Engel &
Worden, 2003; Novak et al., 2003; Smith et al., 2005; Worden, 1989, 1996).

Others have pointed out measurement and theoretical deﬁciencies of prior
research. Speciﬁcally, calls have been made to include a much broader range of attitudes
in ofﬁcer behavioral studies (Worden, 1995). As can be seen from Table 2, earlier studies
included only a few attitudinal measures. Additionally, an emphasis was placed on
providing a more explicit theoretical link between attitudes and behavior. As Frank and

Brandl (1991: 85) noted, “if the attitude measure is not relevant to the target behavior, it

46

\

.‘A.‘.

an

in.

mu

1
-—~ii

would be improper to anticipate that the attitude measure would predict the behavior
under analysis.” For example, it is unclear in earlier studies why ofﬁcer job satisfaction
should predict ofﬁcers’ force or arrest behavior (i.e., Friedrich, 1980; Smith & Klein,
1983). Both other types of attitudes as well as situational factors would theoretically be
more closely related to these types of police behavior. Thus, police scholars have
indicated that research is needed to address these issues before concluding that attitudes
are unrelated to behavior (Frank & Brandl, 1991; Worden, 1995).

Recent research has begun to address many of the above deﬁciencies and the
results are promising for future inquiries into the police-attitude behavior relationship.
For example, the fact that a few studies have reported substantial relationships between
ofﬁcer attitudes and behaviors (e.g., Mastrofski et al., 1995; Paoline & Terrill, 2005;
Terrill et al., 2003) indicates that the relationship remains theoretically important. The
framing of the relationship from a cultural perspective (e.g., Paoline & Terrill, 2005;
Terrill et al., 2003) is an important approach that addresses some of the conceptual
shortcomings of prior research.3 Furthermore, as Table 2.2 indicates more recent studies
have begun to employ a broader range of attitudinal measures (e.g., Paoline & Terrill,
2005; Smith et al., 2005; Terrill et al., 2003) as well as providing more explicit
theoretical links between attitudes and behavior (e.g., Engel & Worden, 2003;
McCluskey et al., 2005).

Despite the progress that has been made, one area that has received little attention

is the conceptual and methodological approach taken by studies. Speciﬁcally, the

 

3 These studies conceptualize attitudes as being shared by ofﬁcers. This is important because these studies
begin to move beyond the individual level assumptions of ofﬁcer attitudes towards a more collective
approach. Although attitudes are characterized from a cultural perspective, these studies are still rooted in
the individual approach, however, since ofﬁcers’ cultural alignments are treated as individual attributes
rather than collective properties. This distinction is discussed in more detail in Chapter Three.

47

 

1“
1

‘1:

limitations associated with taking an individual level approach to the study of ofﬁcer
attitudes has not been systematically incorporated into studies, especially those on the
police attitude-behavior relationship. Such an oversight offers a potential alternative
explanation for why the studies have historically reported null or weak relationships

between attitudes and behavior.

Limitations of Prior Empirical Research

The theoretical focus of past empirical research on ofﬁcers’ occupational attitudes
has emphasized individual-level relationships. Studies that have sought to explain
variation in ofﬁcer attitudes have traditionally emphasized the effect of individual ofﬁcer
characteristics. Studies that have examined the effect of ofﬁcer attitudes on behavior have
relied heavily on the assumption that the individualism and autonomy afforded to ofﬁcers
enables them to develop distinctive styles of policing which are manifest by their
attitudes. These attitudes are then hypothesized to impact ofﬁcers’ behaviors. There are,
however, conceptual and methodological limitations associated with the individual level
approaches taken by these bodies of literature which point to the need for alternative

approaches to be undertaken and applied to the study of ofﬁcer attitudes.

Conceptual Limitations

Conceptually, the emphasis placed on individual level approaches by prior
research neglects the importance of contextual factors that might also inﬂuence attitudes.
Ofﬁcers perform their duties in occupational and social environments that likely shape
their attitudes. By focusing on individual level relationships, however, the bodies of
empirical research reviewed treat ofﬁcer attitudes as if they occur in a social vacuum

(Drummond, 1976; Frank & Brandl, 1991; Klinger, 1997). The deﬁciencies associated

48

with overemphasizing the individualistic nature of policing were noted over thirty years
ago when Drummond (1976) pointed out that police research concentrated too heavily on
individual ofﬁcers while largely neglecting the larger contexts in which they work. For
Drummond (1976: 7), the focus on individual ofﬁcers is “not so much an illogical
proposition as it is an oversimpliﬁcation which neglects the interrelationships of people
in an organization structure.” The limitations of relying on individual level approaches to
the study of ofﬁcer attitudes have also been given attention more recently (See Frank &
Brandl, 1991; Klinger, 1997). Thus, research has pointed out the limitations of
overemphasizing individual level approaches to the study of policing and has identiﬁed
the need for incorporating broader organizational and social environments in which
ofﬁcers work. In essence, this necessitates a move from a micro-level perspective that has
dominated research on ofﬁcer attitudes to a more macro-level approach, which
incorporates higher level factors into research on ofﬁcer attitudes (Klinger, 2004). Two
macro-level factors important to the study of ofﬁcer attitudes are the organizational
context and police culture.
Organizational Context

Police scholars have stressed the importance of examining how organizational
characteristics impact ofﬁcer attitudes. Organizational characteristics have been argued to
have a direct effect on ofﬁcer attitudes because they represent an immediate context for
attitudes (Frank & Brandl, 1991). As will be discussed in the following chapter, the
organizational contexts in which ofﬁcers work likely serve to shape and constrain
ofﬁcer’s occupational attitudes. Research in the organizational sciences has shown that

such constraints reduce the likelihood that individual characteristics will predict

49

attitudinal variation (Kozlowski & Klein, 2000). This questions the ability of individual
characteristics to explain variation in ofﬁcer attitudes and substantiates the calls of
previous research to incorporate the organizational environment into the study of ofﬁcer
attitudes (Brooks et al., 1993; DeJong, 2004; Hassell, 2006; Moon & Zager, 2007;
Paoline et al., 2000; Sun, 2002, 2003). Although such an importance has been stressed,
few studies have been conducted which tests the effect of the police organization on
ofﬁcer attitudes (i.e., Hassell, 2006; Wilson, 1968). Attention will be given to these
studies in Chapter Three.

Police Culture

 

Police culture is commonly deﬁned as a set of shared norms that ofﬁcers use to

cope with the challenges inherent in police work (Herbert, 1998; Manning, 1995; Paoline,

2003).4 Central to the deﬁnition of police culture is the notion of shared attitudes and
values derived from the unique characteristics of policing. These commonly shared
attitudes have been collectively presented as the “street cop culture” and include a
number of the occupational attitudes described above (See Reuss—Ianni, 1983: 13-15).
Shared attitudes among ofﬁcers may provide a reason for the lack of ﬁndings for the
police attitude-behavior relationship as it questions the likelihood that ofﬁcers develop

individual styles that inﬂuence their behavior. The notion that attitudes are shared is

 

4Conceptualizations of police culture have also been undertaken from both an occupational perspective
based largely on the ideas of the early observational studies of the police as well as a subcultural
perspective based on typology approaches. Congruent with the view of the occupational perspective,
ofﬁcers are said to share similar attitudes because they work in similar environments (Crank, 2004). As a
result, police culture is also treated as a “cohesive whole” (Herbert, 1998: 344). On the other hand,
typological studies have also been conducted which found evidence of variation in cultural attitudes. This
subcultural perspective provides evidence of variation in ofﬁcers’ attitudes that differed from traditional
views of the police culture offered by the occupational perspective (Cochran & Bromley, 2003; Jermier,
Slocum, & Fry, 1991; Paoline, 2001, 2004). Regardless of the conceptual approach taken, the key notion
here is that ofﬁcers are described as sharing attitudes.

50

theoretically important because it may indicate that attitudes represent collective
measures of the context of policing. For example, research on organizational culture has
shown that if attitudes are shared, then a multilevel approach is necessary. As Ostroff et
al. (2003: 577) note, “by deﬁnition, a multilevel process takes place in [attitudes],
moving from individual constructions of the situation and sensemaking to the creation of
shared meanings across people.”

The occupational attitudes of the police are sometimes framed from a cultural
perspective in which attitudes are shared by ofﬁcers (Paoline et al., 2000; Paoline &
Terrill, 2005; Terrill et al., 2003). Although this is not always the case, studies on the
police attitude-behavior relationship that have taken such a perspective are some of the
few to uncover substantial relationships (Paoline & Terrill, 2005; Terrill et al., 2003). As
such, the relevance of discussing the relationship between ofﬁcer attitudes and police
culture is necessary because the idea that ofﬁcers share attitudes is important for the
methodological approaches taken by prior research, especially those that test the police

attitude-behavior relationship.

Methodological Limitations

Methodologically, the individual level approaches have employed analysis
techniques that also assume ofﬁcers attitudes exist in a vacuum. The studies reviewed in
Tables 2.1 and 2.2 rely on regression-based analytic techniques as ways for controlling
for other potentially confounding factors related to either attitudes or behavior. Such
techniques assume that observations are independent of each other (Bliese & Hanges,
2004; Kenny & Judd, 1986; Raudenbush & Bryk, 2002). If attitudes are indeed shared by

ofﬁcers, which research on police culture suggests, this assumption is violated because

51

the attitudinal measures would not be independent of each other. The independence
assumptions associated with regression-based analyses are violated and modeling
techniques that correctly model non-independence are needed.

A second methodological limitation involves the sources of data used for
analyses. For both studies that attempt to explain attitudinal differences and studies on the
police attitude-behavior relationship, research has relied on data from only a few
departments. This is more problematic for recent studies as many of them have relied on
data from the POPN study, which only included two departments. Thus, one reason for
the weak ﬁndings in both these bodies of literature may be due to the fact that it is not
possible to ﬁnd adequate variation in attitudes due to the small number of research sites

included by studies (see also DeJong, 2004).

Chapter Summary

The study of ofﬁcers’ occupational attitudes has been a cornerstone of research on
the police for almost ﬁfty years. Early observational studies of the police identiﬁed a set
of occupational attitudes ofﬁcers developed to deal with the challenges of their work
environments. This set of attitudes included views towards citizens, policing tactics, the
police role, supervision, and overall job satisfaction. Ofﬁcer typology studies further
reﬁned the set of attitudes relevant to ofﬁcers and documented variation in these
attitudes. As a result, ofﬁcers’ attitudes became important factors in empirical studies of
the police that sought to explain ofﬁcers’ attitudinal differences, as well as in studies on
the police attitude-behavior relationship.

Empirical research that has sought to explain differences in ofﬁcers’ occupational

attitudes has placed an emphasis on the inﬂuence of ofﬁcer demographic characteristics

52

and length of service as predictors. Fifteen studies that examined these relationships were
reviewed and the results indicated that ofﬁcer characteristics were rather weak predictors
of ofﬁcer attitudes. Prior research on the police attitude-behavior relationship was also
reviewed. The results of the eighteen studies reviewed also found that in most cases
attitudes had little effect on behaviors.

The extant literature relies on individual level assumptions that limit the
conceptual and methodological development of research in this area. For example,
research has indicated that the organizational context in which ofﬁcers work inﬂuences
their attitudes (e. g., Frank & Brandl, 1991). Furthermore, research on police culture
highlights the fact that ofﬁcers are said to share occupational attitudes (e.g., Paoline,
2003). Collectively, these limitations point out the need to move beyond individual level
approaches and toward an approach that can incorporate these two characteristics.

Unfortunately, such approaches are not readily identiﬁable in the policing
literature. Police scholars have pointed out that theory in police research is generally
limited and theory that moves beyond individual explanations is scarce (Klinger, 2004;
Manning, 2005). As Klinger (2004: 124) recently noted, “police scholars have devoted
hardly any effort to developing theories about how organizational and environmental
properties might affect how the police Operate as they do” (Klinger, 2004: 124). The
multilevel paradigm in organizational psychology, however, provides a useful theoretical
framework for accomplishing this. This is the framework taken in the current work and is

the subject of the next chapter.

53

CHAPTER THREE: THEORETICAL FRAMEWORK

This chapter presents the theoretical framework that serves as a guide for the
proposed study. Chapter Two pointed out the need for research on ofﬁcers’ occupational
attitudes to incorporate macro-level phenomena, such as the organizational context and
police culture, with the individual approach that has dominated this line of research.
However, the integration of such approaches is limited within the policing literature. As
such, the proposed study uses the multilevel theoretical perspective in organizational
psychology (Klein et al., 1994; Kozlowski & Klein, 2000) as a way to integrate both
perspectives. The chapter begins by providing a description of this approach. Next, an
overview of how this theoretical framework is applicable to the study of occupational
attitudes of the police is provided. The implications ofthis approach for the study of
ofﬁcer attitudes are also discussed. Finally, the chapter concludes with the speciﬁcation

of the research objectives and questions the proposed study aims to address.

The Multilevel Theoretical Perspective

The multilevel perspective in organizational psychology has been offered as a
way to combine both micro and macro-level perspectives into a single framework that
can be applied to the study of organizational phenomena. A key assumption of the
multilevel perspective is that organizational constructs are tied to, and affected by,
different levels of an organizational system. As Klein et al. (1994: 198) note, “[n]o
construct is level free. Every construct is tied to one or more organizational levels or
entities, that is individuals, dyads, groups, organizations, industries, markets, and so on.

To examine organizational phenomena is thus to encounter levels issues.” By viewing

54

constructs in relation to various organizational levels, this perspective is rooted within
organizational systems theory (Kozlowski & Klein, 2000).

Organizational systems theory offered initial insight as to why the incorporation
of multiple levels into the study of organizations was necessary for gaining understanding
of organizational phenomena. The perspective sought to overcome the weaknesses of
relying solely on either micro-level approaches driven by individual level explanations or
macro-level approaches driven by sociological explanations. It did so by viewing
organizational phenomena as part of the broader social system in which they operated
and by making theoretical connections between system parts (Katz & Kahn, 1978;
Kozlowski & Klein, 2000). Although the inﬂuence of systems theory on organizational
science has been important for conceptualizing relationships among organizational
constructs, its usefulness has primarily been metaphorical as others have noted that it
provides “heuristic value but has contributed relatively little to the development of
testable principles in organizational sciences” (Kozlowski & Klein, 2000: 7).

Unlike the criticism often associated with the systems framework, the multilevel
perspective offers a more direct way of integrating macro and micro-level perspectives. It
achieves this by offering a set of principles to assist in the application of multilevel
theory to organizational research. These principles prescribe the need to explicitly state
the nature of the expected multilevel relationships, where within the system the
relationships are hypothesized to be most salient, the rationale for why such relationships
exist, and the processes by which the relationships are formed. Multilevel theorists have
more appropriately termed these principles the what, where, why, and how and have

argued that researchers should address these issues when applying the framework

55

(Kozlowski & Klein, 2000). As a result, the goal of the multilevel perspective is “to
identify principles that enable a more integrated understanding of phenomena that unfold
across levels of organizations” (Kozlowski & Klein, 2000: 7).

The application of the multilevel perspective offers a potentially fruitful approach
to the study of police ofﬁcers’ occupational attitudes. Multilevel theorists have noted
such an approach is useful when the outcome of interest represents behaviors or
cognitions of individuals that are affected by organizational factors (Kozlowski & Klein,
2000). Occupational attitudes represent individual ofﬁcers’ responses to the challenges of
their work environment. Yet, police scholars have argued that individual ofﬁcer attitudes
are inﬂuenced by higher level organizational factors such as the organizational context
and police culture (Drummond, 1976; Frank & Brandl, 1991; Klinger, 1997; Paoline,

2003). Thus, the study of ofﬁcers’ attitudes ﬁts within this multilevel framework.

Overview of Framework

The current framework identiﬁes workgroups, deﬁned as ofﬁcers’ assigned shifts
within precincts, as the aspect of the organizational context of policing most likely to
inﬂuence ofﬁcer attitudes. Due to the structure of the formal organization of policing,
ofﬁcers within workgroups are exposed to similar features of the occupational
environment (e.g., crime rates, workload, and community structural characteristics) and
common leadership practices of supervisors. Additionally, ofﬁcers assigned to the same
workgroup are likely to interact with each other more often and more directly than with
ofﬁcers outside of their workgroups. As a result, contextual features of workgroups
inﬂuence ofﬁcers’ attitudes and ofﬁcers within workgroups develop shared attitudinal

perceptions regarding the police occupation. These shared perceptions emerge as a result

56

of ofﬁcers’ interactions. As such, emergence is the key process by which attitudes
become shared and is the reason why attitudes can represent measures of the social
environment. The following illustrates how this perspective is applied in the proposed

study and is discussed in terms of the basic principles of the multilevel framework: the

where, why, and how.5

The Where: Ofﬁcer Workgroups

A key component in applying the multilevel framework to the study of ofﬁcers’
occupational attitudes is specifying the “where” or the aspect of the police organization
most likely to inﬂuence ofﬁcer attitudes. Research within the multilevel perspective has
suggested that such inquiries begin with the entity expected to have the most immediate
and proximal effect on the outcome of interest (Katz & Kahn, 1978; Kozlowski & Klein,
2000; Ostroff et al., 2003). In identifying the appropriate organizational entity, two types
of organizational units are considered: formal and informal units. Formal organizational
units include the entities deﬁned by the structures or tasks of the organization while
informal units consist of social groupings of organizational members (Kozlowski &
Klein, 2000). With respect to ofﬁcers’ occupational attitudes, both types of units are
potentially applicable. Prior research has stressed the importance of organizational
structural boundaries (e.g., Hassell, 2006; Wilson, 1968), as well as informal groupings
of ofﬁcers based on typology studies (e.g., Cochran & Bromley, 2003; Jermier et al.,

1991; Paoline, 2001) in shaping ofﬁcer attitudes.

 

5 . . . . . .

It should be noted that the rationale for the “what” prmcnple was discussed at length in Chapter Two in
terms of the limitations of prior research and the need for multilevel approaches to be applied to the study
of ofﬁcers’ occupational attitudes. As a result, the discussion is not repeated here.

57

In this section, I argue that formal structural boundaries are important
organizational entities that shape attitudes. Formal structures specify boundaries that
pattern interactions among organizational members and the nature of work tasks. As a
result, members bounded by these structures interact with each other more often and are
exposed to common features of the work environment (Kozlowski & Klein, 2000). As
Louis (1985: 79) noted, formal organizational structures can inﬂuence members’ attitudes
because “they are regularly convening settings, they impose structural interdependencies
among people performing tasks, they provide opportunities for afﬁliation, and they
constitute constellations of interest or purposes. As such they serve as breeding grounds
for the emergence of local shared meanings.” In policing, organizational structure serves
as important boundaries of police work. As a result, ofﬁcers’ occupational attitudes are
more likely to be inﬂuenced by structural boundaries rather than informal social

groupings (Hassell, 2006; Klinger, 1997).

The Formal Organization of Policing

Police work is structured both geographically and temporally. In most municipal
departments, patrol is divided into smaller geographic jurisdictions so that services can be
provided in an efﬁcient and effective manner (Klinger, 1997; Reiss, 1992). In larger
jurisdictions, departments are ﬁrst divided into precincts that serve a speciﬁc area of the
larger jurisdiction (e. g., city) and have distinct command personnel.6 Patrol operates out
of these precincts, but it is also connected to the larger administrative structure (Hassell,

2006). Precinct boundaries are also further subdivided into beats, or smaller geographic

 

6 . . . . .

The term precmct 1S used throughout, however departments may use different terms to srgmfy these
smaller geographic jurisdictions, such as area commands, divisions, as well as precincts. Although different
terms are used they represent the same type of organizational entity.

58

units patrolled by individual ofﬁcers. Thus, patrol work is territorially deﬁned where the
department serves as the central organizing entity, precincts represent distinct command
structures, and beats are where individual ofﬁcers patrol. Generally speaking, patrol
ofﬁcers are assigned to precincts, and, in some instances, patrol beats (Hassell, 2006;
Klinger, 1997; Reiss, 1992). Although some scholars focus on the geographic nature of
policing (Hassell, 2006; Klinger, 1997), police work is also structured temporally. Within
each precinct, ofﬁcers are also assigned to shifts. Traditionally, ofﬁcers are assigned to
work one of three shifts (i.e., day, afternoon, or night). As Crank (2004: 64) notes,
“[w]ith minor variation, this shift pattern is omnipresent across municipal police
organizations today.”

With respect to the formal organization of policing, ofﬁcers are embedded within
shifts, shifts are embedded within precincts, precincts are embedded within departments,
and departments are embedded within the municipality. Each respective level of the
formal organization of policing could plausibly impact ofﬁcers’ attitudes, and prior
studies have examined how the department (Wilson, 1968), precinct (Hassell, 2006;
Klinger, 1997) and lower organizational levels (Haarr, 2001; Sun, 2002) inﬂuences
ofﬁcer attitudes. As such, it is important to brieﬂy discuss this body of literature in order
to show how the current study builds upon and extends work in this area.

In his classic study of police organizations, Wilson (1968) offered an
organizational perspective that hypothesized that ofﬁcers’ views toward the police role
are inﬂuenced by features at the department level. As Wilson (1968: 48) noted, “[t]he
patrolman’s conception of the police role will vary to some extent with the character of

the community and with the duty assignment of the ofﬁcer, but the general features of the

59

patrol problem-coping with disorder by exercising wide discretion over vital matters in an
apprehensive environment-tend to impart to the police department as a whole its special
character.” His study found evidence of departmental styles and differences in
departmental styles were based on the attitudes and policies emphasized by the chief (see
also Paoline, 2003).

For Wilson (1968), styles vary based on differences in the attitudes and policies
directed at the police role. For example, some departments emphasized the crime-ﬁghting
role and were categorized as legalistic departments while other departments emphasized
order maintenance or service roles and were categorized as watchmen or service style
departments. The key notion here is that the “ways in which ofﬁcers cope with their
organizational environments might differ across different types of departments” (Paoline,
2003: 205). Thus, individual ofﬁcers within departments develop similar attitudes toward
aspects of police work, such as the police role, and as a result will behave consistently in
similar situations (McCluskey et al., 2005).

Prior research has also emphasized the precinct level as an important inﬂuence on
ofﬁcer attitudes. For example, Klinger (1997: 283) stressed the importance of
“territorially based workgroups” in explaining patrol practices in his negotiated order
theory of policing at the precinct level. Using this perspective as a framework, Hassell
(2006, 2007) argued that precincts inﬂuence ofﬁcer attitudes because they serve as
“distinct and relatively independent communities within the infrastructure of the police
organization.” Based on interviews, her study found that ofﬁcers believed that similar
situations would be handled differently by ofﬁcers in different precincts (Hassell, 2006).

Empirical tests, however, found few signiﬁcant differences in ofﬁcer attitudes across

60

precincts. While this may have been due to the methodological design which examined
ofﬁcers from a single department, the work is conceptually important for the idea that
precincts produce variation in attitudes among ofﬁcers and these differences translates
into differences in behavior.

Other studies have taken a peer group or shift perspective. For example, Sun
(2002) argued that ofﬁcers within peer groups were more likely to develop stronger
bonds with each other and adopt similar attitudinal outlooks than with those outside their
peer group. As Sun (2002: 72) concluded, “[i]t would be very difﬁcult for police ofﬁcers
to survive on the job if they did not share many of the views of their peer group.” Peer
groups have also been used as explanations for use of force behavior. Here, peer groups
(ofﬁcers working the same shift and sector) were noted as being “an overlooked gap in
explaining decision making in general, and speciﬁcally in decisions to use force”
(McCluskey et al. 2005: 22). Shifts have also been identiﬁed as being an important
referent for ofﬁcers in conducting their daily work (Crank, 2004; Rubinstein, 1973), and
research has found shifts to inﬂuence their attitudes. For example, Haarr (2001) found
that changes in ofﬁcers’ views toward community policing were partially explained by
the type of shift (i.e., day, afternoon, or night) ofﬁcers were assigned to upon graduation
from the training academy. Thus, attitudinal differences toward police work were
attributed to shift assignments (see also Paoline, 2001; Sun, 2003).

Although the above line of research has examined the impact that the formal
organization of policing has on ofﬁcer attitudes, a limitation of previous work is that each
study focuses solely on a single level. The current study extends this line of research by

incorporating each organizational level into a systematic framework; however it is

61

hypothesized that ofﬁcers’ shifts within precincts (i.e., workgroups), likely have the
greatest impact on attitudes. As will be seen, it is within workgroups where much of
police work is carried out and where ofﬁcer interactions are patterned. In short,
workgroups are the most proximal entity where attitudes are shaped and where ofﬁcers

are most likely to develop shared understandings of them.

Workgroup Conceptualization

The above studies illustrate the importance of formal organizational groups to the
study of ofﬁcer attitudes. Although such groups have been applied in police research,
some discrepancies exist in their conceptualization. Peer groups have been deﬁned
differently, but have been labeled the same (e.g., McCluskey et al, 2005; cf. Sun, 2002).
Furthermore, groups based solely on geographic boundaries (e.g., Hassell, 2006) miss the
importance of the temporal nature of shifts. Each shift leads to interactions with different
types of citizens and different work obligations (Rubinstein, 1973). Thus, even ofﬁcers
working the same precinct likely have different experiences based on the time of day that
they work. As a result, it appears fruitful to examine ofﬁcers’ attitudes in relation to their
assigned shifts within precincts. For the purposes of the proposed study, workgroup is the
term used to represent ofﬁcers’ shifts within precincts in order to avoid confusion with
the peer group and shift concepts of prior studies.

It is important to note that just being assigned to the same workgroup does not
fully capture the manners by which groups inﬂuence ofﬁcer attitudes. Additionally,
certain contextual features associated with workgroups facilitate interactions and expose
ofﬁcers to common features of the work environment. In order to have a full

understanding as to how workgroups inﬂuence ofﬁcer attitudes, it is also necessary to

62

discuss these factors (i.e., the environment, supervisors, and other ofﬁcers). The speciﬁc
aspects of the workgroup environment which expose ofﬁcers to common situations,
constrains their attitudes, and produce shared attitudinal responses serve as contextual

features that have impact ofﬁcers’ attitudes.

The Why: Workgroups as Contexts for Officer Attitudes

Workgroups are argued to inﬂuence ofﬁcer’s occupational attitudes because they
represent immediate and proximal contexts of their work environments. It is within
workgroups where police work is carried out. These contexts shape ofﬁcers’ attitudes and
certain characteristics of workgroups serve as important attitudinal inﬂuences. In
identifying such factors, Hackman (1992: 250) found that, “[g]roups provide the
immediate context for individual thought and action. The attributes of the people in the
group, the characteristics of the tasks, and the properties of the place where the group
works collectively provide a stimulus rich context for individual(s).” Applied to policing,
these three types of factors associated with ofﬁcers’ workgroups are important in shaping
occupational attitudes. Speciﬁcally, characteristics of ofﬁcers’ environment, frontline
supervisors, and the compositional attributes of the group are important characteristics of

the workgroup context.

Environment

The idea that environmental characteristics inﬂuence ofﬁcers’ attitudes was
offered by the early observational studies of the police. For example, Rubinstein (1973:
186) observed that “[t]he patrolmen’s knowledge of people develops haphazardly. He is
constantly recording bits of information about people he meets which he uses in making

judgments about them if he encounters them again. He remembers places where he has

63

had trouble or where trusted colleagues have met resistance. These recollections can
affect his attitudes toward the people he meets even before his arrival at the scene.” Thus,
within workgroups ofﬁcers interact with different types of citizens in their routine
encounters, and these interactions inﬂuence their attitudes, such as their views toward
citizens (Crank, 2004; Rubinstein, 1973).

In addition, limited empirical evidence within the individual approach also
suggests that the characteristics of ofﬁcers’ occupational environments inﬂuence their
attitudes. For example, it has been found that ofﬁcers in low crime and racially
heterogeneous neighborhoods have broader role orientations (Sun, 2003). Furthermore,
ofﬁcers who perceive their beat to be more economically afﬂuent have been found to
have more positive attitudes toward citizens, while those who perceive their beat to have
a crime problem are more likely to hold more negative attitudes toward citizens (Moon &
Zager, 2007). Finally, reported crimes and call for service workloads have also been
found to inﬂuence ofﬁcer attitudes. Speciﬁcally, ofﬁcers with higher workloads were
found to be more crime-control oriented while ofﬁcers with lower workloads more
service oriented. and more positive views toward citizens (Brooks, Piquero, & Cronin,
1994).

The above environmental characteristics, such as interactions with citizens, levels
of crime, and work load rates are all salient factors likely to impact ofﬁcers’ attitudes
within the contexts of workgroups. Prior research has found some evidence that such
characteristics impact ofﬁcers’ views toward citizens and the police role (Brooks et al,
1994; Moon & Zager, 2007; Sun, 2003), but the relationships may also readily apply to

other attitudes. For example, it is possible that ofﬁcers who work in occupational

64

environments characterized by higher workloads and higher levels of crime also
subscribe to aggressive policing tactics and selective enforcement of the law.
Additionally, ofﬁcers in these conditions may also be more likely to feel that the
administration is unsupportive of their efforts to address these workload and crime issues.

Collectively, this may also reduce ofﬁcers’ satisfaction with their jobs.

Front Line Supervisors

A second workgroup characteristic that shapes ofﬁcer attitudes are the inﬂuence
of ofﬁcers’ frontline supervisors (i.e., sergeants). The role of sergeants and their
leadership practices in shaping ofﬁcer attitudes are important for two reasons. First,
sergeants have direct control over ofﬁcers’ daily work routines (Skogan, 2008). Second,
sergeants have been described as mediators between the line level and top management
(Fielding, 1988; Sharp, 1982). As such, sergeants have been described as the
“transmission belts” of police departments (Skogan, 2008: 25).

The daily interactions between sergeants and their ofﬁcers provide for a
relationship where sergeants’ views and attitudes can have a direct inﬂuence on ofﬁcers’
attitudes. Supervisors can affect ofﬁcers’ behaviors by shaping attitudes to coincide with
their own views on what constitutes acceptable police work (Engel & Worden, 2003;
Muir, 1977; Sharp, 1982). For example, Skogan and Hartnett (1997) found that line
ofﬁcer attitudes toward community policing were similar to those of their supervisors.
Other studies have also found line ofﬁcers’ attitudes to be similar to those of sergeants
(Paoline, 2001). Thus, the views of sergeants likely exert a direct inﬂuence on ofﬁcer

attitudes.

65

The idea that sergeants serve as mediators between the larger organization and the
line level also illustrates their importance to the development of ofﬁcers’ attitudes. For
example, the unique role of sergeants allows them to interpret the meaning of the policies
and practices implemented by top management and communicate their own
interpretations to their ofﬁcers (Sharp, 1982; Skogan, 2008). As Sharp (1982: 169-170)
noted, “[s]ergeants reinforce [policies] by using their considerable resources at their
disposal to mold their squad of ofﬁcers to their own image of good policing. Thus. . .the
immediate supervisor’s role is crucial.” This role provides an additional more indirect
inﬂuence on ofﬁcer attitudes.

The above research implies that the role of sergeants within the workgroup
context impacts the occupational attitudes of ofﬁcers. Sergeants’ own views toward
factors such as the citizenry, police role, and policing tactics are likely to have a direct
inﬂuence on how ofﬁcers’ view these factors. Additionally, sergeants’ interpretations of
department policies regarding treatment of citizens, the scope of the police role, or
appropriate policing tactics are also likely to inﬂuence ofﬁcers’ own interpretations.
Finally, sergeants own leadership styles are likely to inﬂuence ofﬁcers’ views toward
supervision as well as job satisfaction. For example, Engel (2000) documented four
distinct types of sergeants’ leadership styles: traditional, innovative, supportive, and
active, and argued that each type would inﬂuence ofﬁcers’ attitudes and behaviors in
different ways. Applied to the current context, ofﬁcers with supervisors who are less
supportive of them may have more negative attitudes toward supervision and may be less

satisﬁed with their jobs as a result.

66

 

ll:

Elli

Workgroup Composition

A ﬁnal workgroup factor that likely shapes ofﬁcers’ attitudes is the composition
of ofﬁcers within the groups. In short, the attitudes of ofﬁcers reﬂect the type of ofﬁcers
who compose the workgroup. Within the workgroup literature in the organizational
sciences, it has been found that phenomena “often reﬂect the number and type of people
who are its members” (Kozlowski & Bell, 2003: 337). Characteristics such as diversity
and experience at the workgroup level serve to inﬂuence organizational phenomena. This
is attributed to the fact that “people with similar backgrounds and experiences tend to
behave in similar ways and to view their worlds from similar vantage points” (Schneider,
1981: 23).

The policing literature has offered insight into how the composition of
workgroups can inﬂuence ofﬁcer attitudes. For example, Haarr (1997) observed racial
and gender barriers to integration into shift work. Speciﬁcally, white male ofﬁcers
interacted almost exclusively with other white male ofﬁcers during their shifts. Female
ofﬁcers, on the other hand, had fewer types of interactions with other ofﬁcers on their
shifts. Diversity alters interaction patterns, as well as how ofﬁcers are socialized into the
culture of policing. Such increasing diversity is argued to change “the patterns of
interactions among ofﬁcers because minorities and women have not been accepted
readily into the traditional culture” (Paoline et al., 2000: 583). As such, compositional
characteristics of workgroups may also inﬂuence ofﬁcer attitudes.

Workgroups composed of ofﬁcers similar to each other may have more positive
views toward supervision and job satisfaction because they receive more support from

each other. On the other hand, workgroups that have higher concentrations of minority

67

5nd {critic
urnlrnu:
ltrkgnit;
manta

' l
l ‘ I 7‘
lChCX JILL!

inpacis
.431. “it.
CHhJHtt
hunts-
eddeUt
“orlgn

Inﬂucn.

lhellu

mama
antudc,
ﬁUnﬂjnc
COnchlU

haS Ol‘l‘cr

and female ofﬁcers may have more negative views toward supervision and job
satisfaction because they have not been adequately socialized into the police culture.
Workgroups with higher concentrations of white, male ofﬁcers may hold attitudes similar
to those expected by the occupational perspective (e.g., distrust of citizens, crime-ﬁghting
roles, and aggressive policing tactics).

It should be noted that the rationale for the inﬂuence of workgroup composition
characteristics on ofﬁcer attitudes is similar to that of the individual approach that
assumes demographic differences among ofﬁcers explains attitudinal differences. Here,
the difference lies in the process by which workgroup compositional characteristics
impacts ofﬁcer attitudes. It is not the characteristics of individual ofﬁcers’ that explains
attitudinal differences, but rather the idea that differences in such characteristics either
enhance or constrain the interaction patterns among officers. Therefore, contextual
factors such as diversity (e. g., gender composition, racial composition, and average
educational levels) and experience (e.g., average length of service of the group) at the
workgroup- level, not the individual-level impacts ofﬁcer attitudes. These factors

inﬂuence ofﬁcers’ interactions, a social process that requires ﬁirther discussion.

The How: Emergence of Shared Attitudes

So far, the multilevel perspective offered has identiﬁed ofﬁcer workgroups as
important structural boundaries of the organizational context that inﬂuences ofﬁcer
attitudes. Additionally, speciﬁc features of the workgroup, such as the environment,
frontline supervisors, and the composition of workgroups have been identiﬁed as salient
contextual factors that explain why workgroups inﬂuence ofﬁcer attitudes. In doing so, it

has offered a perspective for identifying where within the police organization ofﬁcer

68

attitudes are affected and why. The ﬁnal component of the framework addresses the how
ofﬁcers within workgroups come to develop shared attitudinal perceptions. Central to this
discussion is the importance of ofﬁcer interactions within workgroups. The key concept
for establishing how ofﬁcers share attitudes is emergence.

Many concepts central to theories in the organizational sciences have been
characterized as emergent phenomena. A concept is emergent “when it originates in the
cognition, affect, behaviors, and other characteristics of individuals, is ampliﬁed by their
interactions, and manifests as a higher-level, collective phenomenon” (Kozlowski &
Klein, 2000: 55, italics in original). Emergent concepts have two essential features:
elemental content and interactive processes. Elemental content denotes the substantive
components of the concept while interaction is the key process by which the concept
becomes a collective phenomenon (Kozlowski & Klein, 2000). Applied to the current
topic, ofﬁcers’ attitudes comprise the elemental content while interactions among ofﬁcers
reﬂect the process by which emergence takes place. As Ostroff et a1. (2003: 578) notes
emergence occurs, “through communication and information exchange, sharing of ideas,
exchanging work products, and other forms of interactions of employees.” Over time,
these interactions lead to the development of shared occupational attitudes among

ofﬁcers within workgroups. Thus, the current study views emergence from a

compositional perspective (Kozlowski & Klein, 2000).7 Ofﬁcers within workgroups are

 

It is important to recognize that different types of emergence have been identiﬁed based on how
interactive processes are assumed to relate to elemental content. For example, compositional emergence is
based on the idea that individuals will share common perceptions of the elemental content because
interactions lead to a convergence of perceptions. On the other hand, compilational emergence assumes that
individuals will develop different perceptions of the elemental content as a result of their interactions. Sec
Kozlowski & Klein (2000: 66) for a typology of the different types of emergence.

69

argued to develop shared attitudes as a result of their interactions and exposure to similar
environments and supervisors.

The concept of emergence and the importance of ofﬁcer interactions are not new
within the policing literature, as research has included an emergent component to
deﬁnitions of police culture and has also focused on how patterns of interaction among
ofﬁcers in their work environments shape police practices (Crank, 2004; Fielding, 1988;
Haarr, 1997). However, much of the literature on police culture has not focused on how
attitudes become shared, but rather has focused on assessing the amount of variation in
these cultural attitudes (Paoline, 2003). Despite this, a number of studies have discussed
speciﬁc types of interactions important to ofﬁcers which can be drawn on.

For example, prior research has discussed the importance of jokes, stories, and
metaphors to police work (Bayley & Bittner, 1984; Pogebrin & Poole, 1988). As
Pogebrin and Poole (1988: 205) observed ofﬁcers’ use of humor and concluded that
“[t]hrough humor the police reinforce their work perspective. . ., promoting a collective
self-conﬁdence conducive to the maintenance of group autonomy and the exercise of
individual discretion.” In addition to jokes, stories, and humor, interactions based on
negotiations (Klinger, 1997), the communication of myths and rituals (Manning, 1997),
and the sharing of ideas and experiences (Fielding, 1988) have also all been found to be
salient in the policing literature. Collectively, the nature of these interactions produces
shared meanings regarding occupational attitudes. Because ofﬁcers within workgroups
are more likely to interact on a routine basis and are exposed to the same contextual

factors (i.e., environment, supervisors, and other ofﬁcers), shared attitudes emerge.

7O

The concept of emergence is important because it offers insight into how ofﬁcers’
attitudes move from an individual level perspective to higher, workgroup level
constructs. Based on the individual approach, ofﬁcers may develop their own styles of
policing based on their attitudes (e.g., Brown, 1988). However through processes of
interaction, the variation in these interpretations lessens as ofﬁcers form common
interpretations of their work environments. As Kozlowski & Klein (2000: 63) note,
“[v]ariations in individual interpretations dissipate as a collective interpretations
converges.” As such, ofﬁcer attitudes become shared constructs through the process of

emergence at the workgroup level.

Implications for the Study of Ofﬁcer Attitudes

The above framework has two important implications for the study of ofﬁcers’
occupational attitudes. First, the notion that workgroups inﬂuence ofﬁcer attitudes
because ofﬁcers are exposed to common aspects of the work environment suggests that
contextual factors (i.e., environmental, supervisory, and compositional factors) of the
workgroup can explain attitudinal differences. Second, the idea that ofﬁcers within
workgroups develop shared attitudes as a result of their daily interactions implies that
attitudes may also represent contextual measures of the workgroup environment, which is

important for research on the police attitude-behavior relationship.

Workgroup Factors Explain Attitudinal Differences

In explaining why workgroups inﬂuence ofﬁcer attitudes, three key features of the
workgroup were identiﬁed: the occupational environment, supervisors, and the
composition of workgroups. These factors were argued to have a direct inﬂuence on

attitudes because ofﬁcers within workgroups are exposed to the same characteristics on a

71

daily basis. To the extent that environments, supervisors, and workgroup composition
vary across workgroups, workgroup-level factors represent important measures of the
organizational context that can explain attitudinal differences (Kozlowski & Klein, 2000;
Ostroff et al., 2003). These workgroup level characteristics should explain differences
over and above individual ofﬁcer demographic characteristics that have dominated prior

research.

Attitudes as Contextual Measures

Ofﬁcers’ attitudes may also be viewed as contextual measures of the workgroup
environment if they exhibit valid and reliable properties that allow for their aggregation
to the workgroup level. These collective measures, then, may be used as predictors of
ofﬁcer behaviors. In this regard, “groups affect behavior indirectly by shaping members’
beliefs and attitudes” (Hackman, 1992: 251). This notion is not as straightforward as the
previous one and is discussed in more detail based on the idea of workgroup climates in
the organizational psychology literature.

Workgroup climates are deﬁned as “sets of perceptually based descriptions of
organizational features, events, and processes. At the individual level, these perceptions
represent cognitive interpretations of the context that arise from individuals’ interactions
with context and with each other” (Ostroff et al., 2003: 572). These climates represent
macro-level constructs that are based on individuals’ perceptions of their work
environment (James, 1982). As such, they serve as mediators between individuals and
their organizational contexts. As Kozlowski and Hattrup (1992: 162) note, “[c]limate
perceptions represent cognitively based descriptions of these contextual features, linking

contextual factors and events to psychological interpretation and individual responses.”

72

Measures of workgroup climates are based on aggregated responses of
individuals’ attitudes (e. g., group means) and represent emergent constructs. Workgroup
climate measures are designed to “capture or freeze” the social process of emergence and
are used as indicators of collective properties (Kozlowski & Klein, 2000: 64). Such a
conceptualization is important because they can have an impact on individuals’ actions.
For example, James (1982: 220) has argued that “aggregate climate perceptions may
provide a powerful explanatory and predictive tool. . .because describing an environment
in psychological terms. . .may enhance, in comparison to situational descriptors. . .the
understanding of how individuals in general impute meanings to environments and,
especially, how individuals in general will respond to environments.”

Workgroup climate perceptions have been found to signiﬁcantly predict
employees’ behaviors within the organizational psychology literature. For example,
workgroup climate for safety has been examined in relation to the frequency of unsafe
behaviors and accidents within organizations (Hoffman & Stetzer, 1996; Zohar, 2000;
Zohar & Luria, 2005). Safety climates represent individual perceptions about
management’s commitment to safety and individuals’ involvement in safety related
activities and were hypothesized to be shared by individuals within each workgroup. As
Hoffman and Stetzer (1996: 314) noted, “individuals attach meanings to and interpret the
environment within which they work. These meanings and perceptions then inﬂuence the
way in which individuals behave within the organization through their attitudes, norms,
and perceptions of behavior-outcome contingencies.” Results from this set of studies
have found workgroup safety climate perceptions to be signiﬁcantly related to unsafe

behaviors by employees and accidents.

73

Additionally, workgroup climates for procedural justice have been found to
inﬂuence group members helping behaviors (i.e., behavior that assists organization or
other group members). This has led to the conclusion that “procedural justice climate is
an important contextual variable expected to inﬂuence work attitudes and behaviors.”
(Nauman & Bennett, 2000: 888). Finally, reviews of group climate perceptions have also
found shared perceptions to be related to individual job performance, harassment
incidents, and absenteeism (See Ostroff et al., 2003: 574). Thus to the extent that ofﬁcers’
within workgroups share attitudes, aggregate measures of these attitudes may serve as
important predictors of behavior within the workgroup context.

To date, only a single study on the police attitude-behavior relationship has taken
an approach similar to that of the workgroup climate literature. McCluskey et a1. (2005:
21) proposed that ofﬁcers’ use of force behavior was inﬂuenced by a “broader context”
associated with ofﬁcers’ peer groups (i.e., ofﬁcers working the same shift and sector).
Peer groups were expected to have a direct effect on force behavior because they
provided a context that deﬁned how to handle citizens during encounters. The deﬁning
aspect of the peer group was ofﬁcers’ attitudes toward aggressiveness. The study noted
that aggressiveness has “at its core, the presentation of self as an offensive force against
law breakers; and. . .it holds the coercive power of police (aggression/taking the initiative)
at its core” (McCluskey et al., 2005: 23).

In the study, peer group aggressiveness was measured by aggregating ofﬁcers’
reSponses within peer groups and creating a summary score. This was then related to
individual ofﬁcers’ use of force behavior. Ofﬁcers in peer groups that had highly

aggressive attitudes were hypothesized to use higher levels of force than ofﬁcers in peer

74

groups that held less aggressive attitudes. The results from the study, however, indicated
that peer group aggressiveness had no direct effect on use of force behavior. Some
marginally signiﬁcant (p < .10) interaction effects were present among ofﬁcers in highly
aggressive workgroups compared to those in less aggressive work groups. This led to the
conclusion that “the search for direct effects of police peer groups on behavior is unlikely
to explain a great deal when compared with situational aspects of the police-suspect
encounter” (McCluskey et al., 2005: 32).

Although this is one of the only studies to conceptualize ofﬁcer attitudes as a
contextual effect on behavior, the approach taken was limited in two ways. First, even
though peer group aggressiveness was conceptualized as a contextual variable, scores
were assigned as individual ofﬁcer attributes. Therefore, an individual level approach
was taken. The small number of peer groups available in the study for analysis (N=31)
may have precluded treatment of the measure as a macro-level construct. Second,
aggressiveness scores were aggregated without considering whether ofﬁcers within peer
groups shared similar attitudes. As such, the validity of the measure as a contextual

construct was not determined before it was related to force behavior.

Current Study

The purpose of the current study is to provide an empirical test of the multilevel
framework discussed in this chapter. Speciﬁcally, the study tests whether ofﬁcers’
assigned workgroups inﬂuence their occupational attitudes. In order to accomplish this,
three research objectives are addressed:

Research Objective #1 : The primary research objective is to determine if

Workgroups affect ofﬁcer attitudes. Three research questions are examined: 1) are there

75

signiﬁcant differences in ofﬁcer attitudes between workgroups?; 2) how do differences at
the workgroup level relate to differences at the precinct and department levels?; and 3) do
ofﬁcers within workgroups share attitudes? Examination of the above questions is
important for two reasons. First, in order for workgroups to be considered salient contexts
for ofﬁcer attitudes, signiﬁcant differences between workgroups must be found. If there
are no differences in ofﬁcer attitudes between workgroups, then workgroup level factors
will not explain attitudes. Second, in order for attitudes to be viewed as collective
measures of the workgroup environment, it must be demonstrated that ofﬁcers within
workgroups share similar attitudes. If attitudes are not shared, then they cannot represent
valid, workgroup level constructs (e.g., climates) (Bliese, 2000; James, 1982; Kozlowski
& Klein, 2000). To the extent that the results indicate that workgroups serve as valid
contexts for ofﬁcer attitudes, two additional objectives are proposed to further illustrate
the framework.

Research Objective #2: The second objective examines whether workgroup level
factors explain attitudinal differences. Although a number of environmental, supervisory,
and compositional factors were offered as potential explanations in the prior discussion,
the study will focus on workgroup compositional characteristics. Characteristics such as
the racial, gender, educational, and experiential compositions of workgroups will be
measured to test the question: do workgroup compositional characteristics explain
attitudinal differences across workgroups?

Research Objective #3: The ﬁnal objective tests whether workgroup level
attitudes explain ofﬁcer behaviors. The focus here centers on measures of workgroup

attitudinal climates and one type of behavior- ofﬁcers’ use of force. Thus the ﬁnal

76

research question is: do workgroup-level attitudes (e.g., climates) predict ofﬁcers’ use of
force behavior? The following chapter outlines the methodology for how the above three

research objectives will be carried out in greater detail.

77

CHAPTER 4: DATA & METHODS

Despite the need identiﬁed by police scholars to incorporate multilevel
approaches into the study of police ofﬁcer attitudes (Frank & Brandl, 1991; Hassell,
2006; Klinger, 1997; Worden, 1995), there is currently a lack of research in this area.
Much of the extant literature has been rooted within the individual approach to the study
ofpolice attitudes. Additionally, there currently is a lack of data available to adequately
incorporate an organizational approach to the study of policing in general (Klinger,
2004), and ofﬁcer attitudes in particular (DeJong, 2004; Hassell, 2006; Paoline, 2001), as
much of recent policing research has been based on data from only one or two police
organizations. The current study is designed to address these issues in two ways. First,
drawing on the multilevel theoretical perspective within organizational psychology, the
study employs a framework for examining ofﬁcer attitudes from a multilevel perspective.
Second, the study uses data from multiple departments and a methodology designed to
allow for an adequate examination of the research objectives. The remainder of this
chapter describes the data and methodology used to test these objectives. Speciﬁcally,
this chapter discusses the data sources, study departments, study design, measures and

variables, and the analysis plan.

The Assessing Police Use of Force Policy & Outcomes Study

The current study uses data collected as a part of a National Institute of Justice
funded project entitled, Assessing Police Use of Force Policies and Outcomes (2005-IJ-
CX-OOSNIJ). This ongoing project is a multi-phase, multi-site study designed to examine
use of force policies and practices across police departments in the United States. The

S13€Cific goals of the study are twofold: 1) to identify the types of use of force policies

78

that are currently being used by police departments and 2) to determine whether certain
types of policies produce more beneﬁcial outcomes, such as providing better assistance
and guidance to ofﬁcers in making decisions, being associated with less force, fewer
injuries to both citizens and ofﬁcers, and being associated with fewer force-related

complaints and lawsuits ﬁled against departments. In order to accomplish these research

goals, the study was conducted in two phases which are summarized below.8

Phase 1 consisted of an agency survey administered to a stratiﬁed random sample
of 1,083 municipal police departments and sheriff’s ofﬁces across the United States. The
primary purpose of this survey was to assess variation in the types of use of force policies
implemented as well as to identify the most common types of policies currently being
used by departments. Based on the results of the agency survey, eight departments were
selected to be studied more extensively in Phase II. The primary selection criterion for
the eight departments was the type of force policy each agency had implemented. In
essence, each of the eight departments had a different type of force policy.

The agency selection process, however, also sought to enhance the external and
internal validity of research results. With respect to external validity, agency selection
focused on departments of similar sizes. As such, departments that employed between
250 to 1,000 sworn ofﬁcers were targeted. With respect to internal validity, agency
selection focused on departments with similar city and departmental characteristics.
Departments that served similar social and economic jurisdictions, and had similar
Workload (e.g., calls for service and arrests) and crime levels, were targeted. As a result

0f the selection process, participating agencies for the project’s second phase included the

 

 

F 01‘ a detailed description of the objectives and methodologies for both phases of the Assessing Police
Use Of Force Policies and Outcomes Study, see Terrill, Paoline, & Ingram, forthcoming.

79

departments from Columbus, Ohio (CPD), Charlotte-Mecklenburg, North Carolina
(CMPD), Portland, Oregon (PPB), Albuquerque, New Mexico (APD), Colorado Springs,
Colorado (CSPD), Fort Wayne, Indiana (F WPD), St. Petersburg, Florida (SPD), and
Knoxville, Tennessee (KPD).

The purpose of Phase II of the study was to determine if the policies in each of the
eight departments were associated with the outcomes described above (i.e., providing
assistance, less force, injuries, complaints, and lawsuits). As such, data collection for
Phase II of the project focused on two types of data. First, departmental records including
use of force reports, citizen complaints, arrest data, call for service data, and reported
crime data were collected for a two-year period. Second, an ofﬁcer survey was
administered which asked ofﬁcers their views towards their department’s force policy,
their attitudes towards their work and organizational environments, and their background
characteristics (e. g., educational, military, and career backgrounds).

Data for the proposed study come primarily from the ofﬁcer survey administered
in ﬁve of the eight departments (APD, PPB, CSPD, FWPD, & KPD) that participated in
Phase II of the Assessing Police Use ofForce Policy and Outcomes Study.9 Although the
Primary purpose of the project is to assess the impact of force policies on force-related
outcomes, the design of the study as well as the multitude of data components collected

allows for the examination of much broader research issues such as the ones examined

 

 

9

Three of the departments (CPD, CMPD, & SPD) are excluded from the current study. First, CPD was
organized differently from the rest of the sites. For example, CPD not only assigned ofﬁcers to shifts and
pr eCincts, but also to speciﬁc zones across the city. Because the current study examines workgroups in
relation to other structural levels, it is important that the hierarchical nesting of ofﬁcers be similar across
srtes. Second, CMPD was excluded from the current study because the use of force data was not completely
COIIected at the time of the dissertation. In an effort to remain consistent as possible across the study,
CMP D was excluded; however, when the force data becomes available, the site can be incorporated for any
SUbSExluent papers that would come out of Research Objective #3. Finally, the administration at SPD did

“fit“ allow the section of the survey that measured ofﬁcer occupational attitudes to be administered to
0 rcers.

80

here. Speciﬁcally, the multi-site design along with the methodology of the ofﬁcer survey

allows for an examination of the effect of workgroups on ofﬁcers’ occupational attitudes.

Description of Study Sites

City, department, and patrol ofﬁcer characteristics for each of the sites are
discussed to provide some context behind the types of jurisdictions each department
serves and were gathered from three sources: the 2000 US. Census, the 2007 Uniform
Crime Reports (UCR), and agency reports and rosters. As mentioned above, both the city
and department characteristics were used as selection criteria for the larger NIJ project.
Since the current study focuses on the attitudes of patrol ofﬁcers, the demographic
characteristics for ofﬁcers assigned to patrol at the time of the study are also presented. A
summary of the city, department, and patrol ofﬁcer characteristics are provided in Table

4.1.

City Characteristics

Based on the city characteristics reported in Table 4.1, the ﬁve study sites shared
similar demographics. All of the departments served cities ranging in population from
over half a million people (APD & PPB) to just under 200,000 people (KPD). In terms of
racial demographics, Albuquerque (APD) had a slightly higher percentage of non-white
residents than the other sites; however all ﬁve sites had a minority population ranging
approximately between 20 and 28 percent. It should be noted that although serving the
smallest population, Knoxville (KPD) did experience higher poverty levels and Part 1
crimes than the other sites. In terms of similarities, all ﬁve sites had closely comparable
Percentages of female-headed households as well as percentages of residents below the

Poverty line.

81

Table 4.1: City, Department, & Ofﬁcer Characteristics of Study Sites

 

 

 

Characteristics APD PPB CSPD FWPD KPD
City
Population 3 513,124 538,133 374,112 248,423 182,337
% Nomwmteb 28.4 22.1 19.3 24.5 20.3
% Female Headed Householdsb 8'0 6'3 7'1 9'8 8'0
% Below Povenyb 10.0 8.5 6.1 9.6 14.4
% Unemployed!) 3.8 4.5 3.1 4.3 3.9
UCR Part 1 Crimes a 34,373 35,287 18,539 10,845 14,920
UCR Part I Crimes/1,000 66.98 65.57 49.55 43.65 81.38
residents a
Department .
Total # Sworn Ofﬁcersa 986 989 669 457 382
# Ofﬁcers/1,000 residents 1.92 1.84 1.78 1.84 2.09
Total # Calls for Sewicec 335,133 219,840 286,534 182,912 225,064
Total # part1 Arrest; 3,428 7,083 4,604 2,495 2,960
UCR Part l/ofﬁcer 34.86 35.67 27.71 23.73 39.05
Calls for service/ ofﬁcer 339.89 222.28 428.30 400.25 589.17
Part I Arrests per/ofﬁcer 3.48 7.16 6.88 5.46 7.75
# of Precincts 5 5 4 4 2
# of Shifts 4 3 3 3 4
Patrol Ofﬁcerd
Total # Patrol Ofﬁcers 429 364 314 212 141
Gender
Male 369 (86.0) 312 (85.7) 273 (86.9) 193 (91.0) 128 (90.8)
Female 60 (14.0) 50 (13.7) 41 (13.1) 18 (8.5) 13 (9.2)
Race
White 245 (57.1) 315 (86.5) 254 (80.9) 181 (85.4) 129 (91.5)
Black 10 (2.3) 13 (3.6) 19 (6.1 ) 21 (9.9) 7 (5.0)
Hispanic 165 (38.5) 12 (3.3) 29 (9.2) 7 (3.3) 0 (0.0)
Other 6 (1.4) 21 (5.8) 12 (3.9) l (0.5) 5 (3.5)
Age
Median (years) 33.87 36.58 36.87 36.17 34.00
Length of Service

Average 06298)

5.62

8.87

9.27

9.67

6.83

a

bSource: United States Department of Justice, FBI (September, 2008). Crime in the United States 2007
cSource: United States Census Bureau (2000)
dSource: Agency 2006-2007 Annual Reports

Source: Departmental Master Rosters. Note: Numbers in parentheses represent percentages. Totals may
"Qt equal total number of patrol ofﬁcers due to missing data.

\

82

Department Characteristics

Both Albuquerque (APD) and Portland (PPB) were the largest departments in
terms of number of sworn ofﬁcers, while Knoxville (KPD) was the smallest. Despite the
size differences, all ﬁve departments were comparable when examining the number of
sworn ofﬁcers per 1,000 residents as the range was 1.78 (CSPD) to 2.09 (KPD). Thus,
even though city population and department size varied, stafﬁng levels were similar
across sites. Furthermore, even though KPD had the highest crime rate, it had similar Part
I crimes per ofﬁcer as APD and PPB. In terms of workload measures, KPD did vary from
the other four departments in terms of calls for service. KPD had over 100 more calls for
service per ofﬁcer than CSPD, the next highest department. Based on these numbers,
KPD appears to have slightly higher workload levels than the other departments. The
sites did have comparable Part I arrests per ofﬁcer.

The ﬁve study departments are also structured similarly. Each department is
broken down geographically into precincts. Table 4.1 shows that the number of precincts
per department varies according to department size with the larger departments having
four to ﬁve precincts while KPD, the smallest department, having only two precincts.
Each of the ﬁve departments had similar shift allocations. Patrol ofﬁcers in PPB, CSPD,
and F WPD were assigned to one of three shifts (i.e., Day, Afternoon,or Night) while
ofﬁcers in APD and KPD could be assigned to one of four shifts (i.e., Day, Afternoon,
Split, or Night). With respect to shift assignments, ofﬁcers assigned to shifts generally
worked four days a week and eight to ten hour days depending on the site. Each site used
a bid system to allocate ofﬁcers to shifts primarily on an annual basis. Each year ofﬁcers

were allowed to bid on what shift and precinct assignment they wished to be assigned

83

with one exception; FWPD ofﬁcers could bid for a shift assignment whenever one
became open.

Finally, it should be noted that ofﬁcers were not necessarily assigned to their own
beats. In fact, only ofﬁcers in KPD and FWPD had stable beat assignments. Ofﬁcers in
the other three sites were not regularly assigned to work a speciﬁc beat. This is likely due
to the fact that KPD and F WPD police smaller jurisdictions, while the other sites have
larger areas to cover. Reductions in resources were also identiﬁed by department
personnel as another reason why ofﬁcers were not assigned their own beats. Rather

ofﬁcers may have been needed to work a number of beats based on shift needs.

Patrol Officer Characteristics

The last set of characteristics in Table 4.1 depicts the demographic characteristics
of patrol ofﬁcers across the ﬁve study sites. Across all sites, most ofﬁcers were male and
white. This is consistent with the demographic characteristics of police departments in
general, as policing has historically been dominated by white males (NRC, 2004;
Manning, 2005). APD, however, did have a sizeable population of Hispanic ofﬁcers
(38.5%). Additionally, patrol ofﬁcers were between 33 and 36 years of age. Ofﬁcers’
length of service did vary across sites ranging from approximately 6 years in APD to
approximately 10 years in FWPD. Although data on ofﬁcers’ educational levels were not
PI‘OVided in the departmental master rosters, educational requirements for ofﬁcers also

varied by department. Based on the hiring requirements listed on each agency’s website,
AP D, FWPD, and KPD required only a high school education or equivalent (i.e., GED)

While PPB and CSPD required and Associates degree or 60 hours of college credit.

84

Ofﬁcer Survey Research Design & Protocol

The ofﬁcer survey was administered during roll call sessions at the start of each
shift and targeted patrol ofﬁcers assigned to a patrol shift. Surveys were administered
during roll calls at every site except FWPD as this site did not use a roll call system.
Here, the ofﬁcer survey was coordinated with the department’s annual in-service training
schedule, and the surveys were administered at this time by the research team. The group
administration method and protocol, however, remained similar to the other sites. The
instrument contained 116 questions and was divided into three main sections. The ﬁrst
section asked ofﬁcers their views towards their departments’ use of force policy. The
second section asked ofﬁcers about their attitudes regarding their organizational and
work environment. The ﬁnal section asked some basic background characteristics. The
instrument was pre-tested on a sample of current and former police ofﬁcers in Michigan
and Florida. These ofﬁcers provided feedback regarding both the content and structure of
the survey. Based upon the comments from the pre—test, revisions were made which
incorporated this feedback before the survey was administered at each site.

The goal of the design was to survey the entire target population within each of
the ﬁve study sites. This was done for three reasons. First, sufﬁcient resources were
available to the research team to survey every patrol ofﬁcer. Second, the organization of
departments and their roll call system facilitated this approach. Finally, the multilevel
nature of the study’s research objectives requires large sample sizes as well as sufﬁcient

Sanlple representativeness across workgroups. Thus, instead of using a sampling frame, a

CensuS was conducted.

85

Survey Protocol

In order to carry out the survey research design, a standardized protocol was
implemented to ensure that the survey was administered consistently across sites. First, a
departmental master roster listing all sworn ofﬁcers, their demographic characteristics
(i.e., race, gender, date of birth, date of hire), and their current work assignments (i.e.,
rank, precinct, shift, and work schedule) was obtained for each department approximately
two weeks to one month before surveying was to take place. Based on the master roster a
survey schedule was created for each site that afforded the research team the opportunity
to survey every patrol ofﬁcer assigned to a patrol shift. Second, a memorandum
providing background information about the research project and the survey schedule for
each site was disseminated department wide to alert command personnel of the dates that
the research team would be administering the survey at their roll calls. Third, a member
of the research team attended each designated roll call to administer the survey.

During roll call, ofﬁcers were given a brief description of the research project and
instructions for completing the survey. The survey administrator also informed ofﬁcers
that participation was voluntary during this time. Ofﬁcers were to complete the surveys
during roll call before going out to work their shift. The surveys took approximately
twenty minutes to complete. Upon completion, the research team member collected the
survey directly from the ofﬁcer. Across sites, survey administration took place over a
Period of a week to ten days depending upon the organization of the department. In most
instances, the survey administrators visited each roll call twice to ensure that every

Ofﬁcer was given the opportunity to complete the survey.

86

Conﬁdentiality Protections

In addition to standardized protocol, several precautions were taken to ensure the
conﬁdentiality of ofﬁcers’ responses. First, ofﬁcers were briefed about the conﬁdentiality
protections afforded to them during roll call sessions. Because surveys were administered
while a researcher was present, any questions regarding the conﬁdentiality of responses
could also be directly answered. Second, ofﬁcers were assigned unique identiﬁcation
numbers speciﬁcally for the study. These identiﬁcation numbers were generated
randomly and were not associated with any characteristics of the ofﬁcers (e.g., badge
numbers). The numbers were placed on the survey instrument and served primarily as a
way for the research team to remove ofﬁcers from the survey list upon completion.
Finally, the informed consent signature page was placed on a cover sheet separate from
the survey instrument. After completion of the survey the informed consent cover sheet

was removed from the survey instrument and stored separately.

Survey Administration Results

Table 4.2 reports the survey administration results across each of the ﬁve study
sites. The numbers in Table 4.2 represent the total number of ofﬁcers assigned to patrol,
the number of patrol ofﬁcers present at roll calls when the survey was administered, the
number of ofﬁcers surveyed, the percentage of ofﬁcers surveyed, and the response rates
for each site. Total numbers and averages are also provided. On average, approximately
seventy percent of patrol ofﬁcers in the ﬁve sites were surveyed. CSPD had the lowest
Percentage of patrol ofﬁcers surveyed; however, the percentage was still above sixty
Percent. FWPD and KPD had the highest percentage of ofﬁcers surveyed (approximately

85%). It should be noted that the 407 patrol ofﬁcers on work rosters but absent from roll

87

calls were not available due to illness, vacation, training, court, or leave. For those
ofﬁcers present during the administration times, almost all took the survey. On average,
the survey response rate was considerably high (i.e., approximately 97 percent). Even
PPB, the site with the lowest response rate, still had a completion rate of almost ninety-
ﬁve percent.

In addition to the results presented in Table 4.2, additional analyses were
conducted to ensure representativeness across each department. For each of the ﬁve sites,
chi-square and t-tcsts were conducted to determine if there were any signiﬁcant
demographic differences (i.e., race, gender, age, and experience) between those surveyed
and those not surveyed.IO The only signiﬁcant ﬁnding was a gender difference for APD.
Speciﬁcally, slightly fewer females were surveyed than expected (chi square: 4.663;
p<.05). With respect to all other demographic characteristics, there were no signiﬁcant
differences between ofﬁcers surveyed and those not surveyed for any site.

Collectively, the results suggest that the ofﬁcers surveyed in each site are
representative of the patrol population for each department. The high percentage of
ofﬁcers surveyed, high response rates, and the presence of only one demographic
difference between ofﬁcers surveyed and those not surveyed are indicative of sample
representativeness across each site. These results can be attributed to the attention given
to the planning and coordination required by the survey protocol. The on-site
administration method coupled with the attempt to survey every available patrol ofﬁcer is
a critical reason for the high percentages and response rates.

Bivariate tests (not presented) were conducted for each site and for each demographic characteristic
U_Slng data from each department’s master rosters. Tests for differences in education could not be conducted
smce this data was not provided for all ofﬁcers. Results can be provided upon request.

88

Table 4.2: Ofﬁcer Survey Administration Results

 

 

Site Patrol #1: # at Roll # Surveyed Patrol % Response
Call Rateb
APD 429 304 301 70.16 99.01
PPB 364 245 232 63.74 94.69
CSPD 314 194 189 60.19 97.42
FWPD 212 189 182 85.84 96.30
KPD 141 121 120 85.11 99.36
Total #/Avg. o/o 1,460 1,053 1,022 70.14 97.25

 

a . .
Represents the total number of patrol ofﬁcers assrgned to a patrol shift.

b . .
Response rates are based on the total number of patrol ofﬁcer surveyed d1v1ded by the number of patrol
ofﬁcers present at roll calls.

Workgroup Operationalization

Workgroups are deﬁned as ofﬁcers’ assigned shifts within precincts. This
conceptualization is similar to prior research that has examined the effect of peer groups
(e.g., McCluskey et al., 2005; Sun, 2003) and shifts (Haarr, 2001; Paoline et al., 2000;
Sun, 2002) on ofﬁcer attitudes. The current work has incorporated aspects from each of
these entities to capture both the spatial and temporal boundaries important to the
structure of police work. Using data from the master rosters provided by each department
as a part of the survey protocol, ofﬁcers were coded into their respective workgroup.
Thus, each ofﬁcer assigned to the same shift in the same precinct is considered to be a
member of the same workgroup. For the current study, ofﬁcers’ workgroup membership
serves as a key independent variable. It is hypothesized that workgroup membership
inﬂuences each of the occupational attitudinal measures.

Since ofﬁcer workgroups represent an important organizational level in the
current study, survey administration results were also calculated to ensure sample
representatives across workgroups. Based on the shift and precinct information from

Table 4.1, there are 67 total workgroups available for analysis. However, there were six

89

workgroups that had low survey response rates (e. g., < 50%). As a result, all workgroup-
level analyses in the subsequent chapters will be based on 61 workgroups. The number of
workgroups for each of the ﬁve sites is as follows: APD (18 workgroups), PPB (13
workgroups), CSPD (10 workgroups), FWPD (12 workgroups), and KPD (8

workgroups).

Table 4.3: Workgroup Survey Results (N=61)

 

 

 

Site Workgroup Avg. # of Avg. # Avg. % Min. % Max. %
N Ofﬁcers Surveyed Surveyed Surveyed Surveyed
APD 18 22.61 16.16 72.00 55.55 100.00
PPB 13 25.46 17.00 67.91 50.00 81.25
CSPD 10 27.00 17.60 66.31 56.10 83.33
FWPD 12 17.67 15.16 85.77 76.90 91.67
KPD 8 17.63 15.00 84.64 70.00 100.00
Averages 12.20 22.07 16.18 75.32 61.70 91.25

 

Table 4.3 displays the survey administration results at the workgroup-level across
the ﬁve sites. Workgroups had approximately 22 ofﬁcers assigned to them, of which
approximately sixteen (i.e., 75.32%) were surveyed on average across sites. After
excluding the six workgroups with low response rates, the minimum percentage of
ofﬁcers surveyed within workgroups were all at 50 percent or above. The maximum
percentage of ofﬁcers surveyed was high, ranging from approximately 80 to 100 perCent
across the ﬁve sites. Overall, these results suggest that the survey data should provide an
accurate representation of ofﬁcers’ attitudes within workgroups. Again, these results can
be attributed to the attention given to the planning and coordination required by the

SuWey protocol.

90

Ofﬁcers’ Occupational Attitudes

Survey Items

Nineteen questions were taken from the section of the ofﬁcer survey measuring
ofﬁcers’ attitudes toward their work environments (See Appendix A). The questions are
based on Likert scales ranging from 1 (Strongly Agree) to 4 (Strongly Disagree) and
measure the extent to which ofﬁcers agree with the statement posed. The survey items
used as measures for ofﬁcers’ occupational attitudes are identical to those used by prior
research in this area. This is beneﬁcial for two reasons. First, the measures used in prior
research have a lineage indicating that they are reliable and valid constructs. For
example, many of the survey items were also used by the POPN study, and the research
generated from them (see Chapter Two) has generally indicated that the psychometric
properties of the attitudes are reliable and valid. Second, police scholars have noted the
importance of developing consistent measures for ofﬁcers’ attitudes so that the results
can be compared and ﬁndings can become cumulative (Paoline et al., 2000; Worden,
1995). As such, it is expected that these items will reﬂect the seven occupational attitudes
outlined previously. Also, by using the same measures as prior research the proposed
study builds upon the results in this area.
Ews toward Citizens

Two survey items are used as indicators of the citizen distrust measure.
Speciﬁcally, ofﬁcers were asked the extent to which they agreed with the following
Statements: 1) “Police ofﬁcers have reason to be suspicious of most citizens.” and 2)
“Police ofﬁcers have reason to be distrustful of most citizens.” Higher values indicate

ofﬁcers have a greater distrust of citizens.

91

Policimz Tactics

The survey items used as indicators of aggressiveness and selective enforcement
were taken from Brown’s (1988) study. To measure aggressiveness, ofﬁcers were asked
the extent to which they agreed with the following statement: “A good police ofﬁcer is
one who patrols aggressively by stopping cars, checking out people, running license
checks, and so forth.” To measure selective enforcement, ofﬁcers were asked the extent
to which they agreed with the statement: “An ofﬁcer is more effective when s/he patrols
for serious felony violations rather than stopping people for minor trafﬁc violations and
misdemeanors.” Higher values on both measures indicate ofﬁcers are more aggressive
and selective.
Role Orientations

Prior research has measured ofﬁcers’ role orientations in two different ways.

Some studies use three separate measures for each of the three primary policing roles
which consist of enforcing laws, maintaining order, and providing services (e.g., Brooks
et al., 1993; Lord & Friday, 2008; Paoline etal., 2000; Schafer, 2002). Essentially, this
method measures ofﬁcers’ acceptance of each role type. Other studies, however, include
all items associated with the police role under a single measure (e.g., Sun, 2003; A.
Worden, 1993; Worden, 1990). This method is used to assess whether ofﬁcers hold
narrower role orientations geared toward crime ﬁghting or hold broader role orientations
that indicate acceptance of other roles. To be consistent with the conceptual development
Of this occupational attitude described in Chapter Two, the proposed study will follow the

latter approach.

92

The narrow role orientation measure consists of seven items that indicate the
degree to which ofﬁcers feel that law enforcement is the primary role of the police.
Ofﬁcers were asked the extent to which they agreed with the following statement:
“Enforcing the law is by far a patrol ofﬁcer’s most important responsibility.” The ﬁnal
six items measure how ofﬁcers view order maintenance and service job roles.
Speciﬁcally, ofﬁcers were asked the extent of their agreement in having to address the
following six order maintenance/service duties: public nuisances, neighbor disputes,
family disputes, parents who don’t control their kids, nuisance businesses, and litter and
trash. For these items, higher values indicate a narrower role orientation.

Views toward Supervision

Five survey items were used to measure ofﬁcers’ views toward two types of
supervision. Ofﬁcers’ views toward direct supervisors were measured with the following
questions: 1) “My supervisor looks out for the personal welfare of his/her subordinates.”
and 2) “My supervisor’s approach tends to discourage me from giving extra effort.” Both
items were coded so that higher values indicate more negative views toward supervision.
ED Management

Views toward top management consisted of three questions: 1) “When an ofﬁcer
does a particularly good job, how likely is it that top management will publicly recognize
his or her performance?”, 2) “When an ofﬁcer gets written up for minor violation of the
rules, how likely is it that he or she will be treated fairly?”, and 3) “When an ofﬁcer
contributes to a team effort rather than look good individually, how likely is it that top
management here will recognize it?” Items were coded so that higher values indicate

more negative views toward top management.

93

Job Satisfaction

Finally, ofﬁcers’ job satisfaction consisted of three questions: 1) “I ﬁnd real
enjoyment in my job,” 2) “I like my job better than the average police ofﬁcer does,” and
3) “I would not consider taking another job.” Here, higher values on each item indicate

greater job satisfaction.

Validity & Reliability

Although the attitudinal items used for the study have displayed valid and reliable
properties in previous studies, three sets of analyses were conducted to ensure the validity
and reliability of the constructs in the current study. First, polychoric correlations were
estimated to determine the strength and nature of associations among individual survey
items.ll Ideally, items hypothesized to represent the above attitudinal constructs would be
highly correlated with each other and not highly correlated across items hypothesized to
comprise the other constructs. The results from the correlation analysis presented in
Appendix B generally indicated these patterns. For example, the two items hypothesized
to measure supervisor distrust were strongly correlated (r=.64, p<.001). Additionally, the
items hypothesized to measure the constructs of top management, role orientations,
citizen distrust, and job satisfaction were more strongly associated with each other than
with the other items. Two items, however, displayed some properties that need to be
explored further.

With respect to the role orientation measures, the item, “enforcing the law is by

far a patrol ofﬁcers’ most important responsibility,” was not signiﬁcantly correlated with

 

 

l 1
Polychoric correlations were estimated since the survey items were ordinal variables (i.e., Likert

measures with four reSponse categories). When measures are ordinal but expected to follow an underlying
continuous distribution as survey items usually are assumed to be, polychoric correlations are appropriate
(Muthen & Muthen, 2007).

94

any of the other six hypothesized role orientation items. This is a preliminary indication
that the item is not a valid indicator of ofﬁcers’ role orientations. This may be due to the
fact that the current study operationalizes the role orientation measure as the extent to
which ofﬁcers hold narrow or broad orientations collectively rather than assessing
ofﬁcers’ acceptance of each of the three major roles of policing separately (i.e., enforcing
laws, order maintenance, and providing services). Ofﬁcers’ attitudes toward enforcing the
law was also moderately correlated with aggressiveness (r=.4 1 , p<.001). This may
indicate that enforcing the law and aggressiveness represent a single underlying

construct.

Second, conﬁrmatory factor analysis (CFA) was conducted to assess the construct

validity of the survey items as well as to examine the three issues discussed above.12
Here, CF A was used to determine how well the proposed measurement model ﬁt the data.
The results of the ﬁnal CFA model are presented in Table 4.4. Overall, the model ﬁt
statistics indicate good ﬁt between the model and the data. Although the chi-square
statistic was signiﬁcant suggesting poor model ﬁt, prior research has shown this statistic
to be sensitive in models with large sample sizes (Kline, 1998; Yu, 2002). As such, the
comparative ﬁt index (CFI=.95), Tucker-Lewis index (TLI=.96), and the root mean
square error of approximation (RMSEA=.06) were also examined, and the values suggest
acceptable ﬁt between the model and the data. Based on the model ﬁt statistics, there is

support that the proposed attitudinal items represent ﬁve distinct constructs: citizen

A number of potential measurement models were tested to examine the measurement issues identiﬁed in
the correlation analysis. For example, an initial model tested whether ofﬁcers’ views toward law
enforcement loaded on the role orientation factor. Results revealed a small and insigniﬁcant standardized
factor loading of -.1 1. This measure was subsequently dropped from the role orientation construct. A
second model tested whether attitudes toward enforcing laws, aggressiveness, and selectivity best
represented a single measure. The results indicated that this was not the case. As a result, these three items
Were dropped from the ﬁnal CF A model.

95

distrust, narrow role orientations, supervisor distrust, top management, and job
satisfaction.

Table 4.4 also reports the unstandardized loadings, standardized loadings, and the

observed R2 values for each item. All of the unstandardized factor loadings were

signiﬁcant, and all of the standardized factor loadings were large (i.e., >.50) (Kline,
1998). Additionally, the observed R2 values indicate that the underlying constructs

accounted for at least a third and up to almost ninety percent of the variance across
individual survey items. Based on the ﬁnal CFA model results, the third analysis assessed
the reliability of the items as measures of the underlying constructs. Four of the ﬁve
constructs displayed acceptable reliability coefﬁcients of .70 or greater. Even though job
satisfaction (0:163) was somewhat lower, each of the three items will be retained based
on the results of the CF A model.

Overall, the results of the above sets of analyses suggest that the items reliably
and validly measure ﬁve multi-item constructs. Speciﬁcally the model ﬁt statistics, the
signiﬁcant unstandardized loadings, the magnitude of standardized loadings, and the
magnitude of the reliability coefﬁcients suggest that ﬁve distinct constructs can be
derived from the data: citizen distrust, order maintenance, supervisor distrust, top
management, and job satisfaction. In order to be consistent with the prior research

discussed in Chapter Two, these constructs will be measured in the current study as

additive indices.” In addition to the ﬁve multi-item constructs, three single-item

measures will also be retained. Although ofﬁcers’ views toward law enforcement,

 

 

Unless otherwise noted, the results reported in the succeeding chapters will be based off the additive
index measures. The factor scores from the measurement model were saved and were also used as variables
in subsequent analyses in order to ensure the results are consistent across different measures. However,
unless there is a discrepancy, only the results from the additive index measure are reported.

96

aggressiveness, and selectivity did not reliably reﬂect one of the above constructs, these
attitudes are still theoretically important to the occupational attitude literature discussed
in Chapter Two. As such, these measures will still be included in subsequent analyses as
single-item measures and are referred to as crime ﬁghting, aggressiveness, and selectivity
hereafter. The current study includes the examination of eight occupational attitudes as
key outcomes of interest.

Table 4.4: Final CFA & Reliability Results of Attitudinal Measures (N=1,022)
Constructs & Measures Unstandardized Standardized Observed

 

 

 

 

 

 

Loadings Loadings R2
Citizen Distrust (a=.81)
Ofﬁcer suspicious of citizens 1.00 (.00) 0.94 0.87
Ofﬁcer distrustful of citizens 0.86 (.16)*** 0.81 0.64
O_rder Maintenance (a=.75)
Ofﬁcer required to...
Public nuisances 1.00 (.00) 0.61 0.37
Neighbor disputes 1.30 (.05)*** 0.79 0.63
Family disputes 1.15 (.06)*** 0.70 0.49
Litter and trash 1.00 (.05)*** 0.61 0.37
Parents 0.94 (.05)*** 0.57 0.33
Nuisance businesses 0.94 (.06)*** 0.57 0.32
mewisor Distrust (a=.70)
Looks out for personal welfare 1.00 (.00) 0.93 0.86
Approach discourages effort 0.75 (.06)*** 0.69 0.48
ED Management (a=.78)
Recognizes good job 1.00 (.00) 0.71 0.50
Treats fairly for minor violation 1.1 1 (.04)*** 0.79 0.62
Recognizes group contribution 1.26 (.05)*** 0.89 0.80
Job Satisfaction (a=.63)
Not consider taking another job 1.00 (.00) 0.63 0.40
Like job better than avg. ofﬁcer 0.94 (.07)*** 0.59 0.35
Find enjoyment in job 1.38 (.10)*** 0.87 0.76

Model Fit Statistics

x2 =304.32, df= 54, p<.001
CFI=.95; TLI=.96; RMSEA=.06

***p<.001

97

Analysis Plan
Research Objective #1 : Workgroup as Contexts for Oﬂicer Attitudes

The primary research objective of the proposed study is to examine the extent to
which ofﬁcer workgroups represent important contexts for examining ofﬁcer attitudes.
Three sets of analyses are used to assess whether workgroups represent contexts that
inﬂuence ofﬁcers’ occupational attitudes. First, multi-level modeling procedures are used
to test whether signiﬁcant between-group variability exists on the attitudinal constructs
between workgroups. Next, the between-group variability will be compared to the
amount of attitudinal variation between precincts and departments. To the extent that
signiﬁcant variation exists at the precinct or department levels, it will also provide a way
to control for this variation. Finally, empirical tests will be conducted to determine if
ofﬁcers within workgroups share similar attitudinal perceptions. Here, there must be
adequate homogeneity of ofﬁcers’ attitudes within workgroups. If both sets of analyses
indicate signiﬁcant between-group variability and within-group homogeneity, then there
is empirical support that workgroups serve as contexts for ofﬁcers’ attitudes (Bliese,
2000; Hoffman & Steitzer, 1996; James, 1982; Kozlowski & Klein, 2000; Zohar, 2000).

Conducting the above analyses is necessary for two reasons. First, the results from
the between-group variability analyses will indicate if workgroup level compositional
factors are useful for explaining variation in ofﬁcers’ attitudes, which is the subject of
second research objective. Second, the results from the within group homogeneity
assessment will provide an indication as to whether ofﬁcer attitudes display collective,
contextual properties themselves. For example, sufﬁcient within-group agreement
indicates a “shared assignment of psychological meaning, from which it follows that an

aggregate (mean). . .score provides the opportunity to describe an environment in

98

psychological terms. . (James, 1982: 221). As such, evidence of within-group
agreement and group mean reliabilities allows for the aggregation of attitudes to the
workgroup level because the measures have valid and reliable collective properties
(Bliese, 2000; Kozlowski & Hattrup, 1992). These collective attitudes can serve as
perceptual measures of the workgroup context, which in turn may explain ofﬁcers’ street

level behaviors. This is the subject of the third research objective.

Research Objective 2: Workgroup Composition & Ofﬁcer Attitudes

The second research objective tests the effects of workgroup composition
characteristics on ofﬁcer attitudes. Using data from the survey and master rosters, this
objective examines whether the demographic makeup of workgroups explains variation
in attitudes over and above the effects of individual ofﬁcer demographic characteristics.
For example, if job satisfaction is found to be inﬂuenced at the workgroup level, then
workgroup compositional characteristics such as the proportion of male ofﬁcers or the
proportion of minority ofﬁcers present in the workgroup may aide in explaining why
some ofﬁcers are more satisﬁed with their jobs than others. Therefore, attitudes constitute
the outcomes of interest and multilevel models are proposed to address this objective.

Speciﬁc details of the analytic procedures are discussed in Chapter Six.

Research Objective #3 .' Workgroup A ggressiveness & Oﬂicer Behavior

The third research objective is proposed to demonstrate the utility of examining
Ofﬁcer attitudes as collective, workgroup constructs that impact ofﬁcer behavior. In order
to test this objective, this part of the proposed study focuses on one type of behavior,
Ofﬁcers’ use of force. For example, if ofﬁcers within workgroups share distrust toward

Citizens, then this distrust represents a collective measure of ofﬁcers’ workgroup

99

environment. This collective, perceptual measure may then explain ofﬁcers’ use of force
behavior. Ofﬁcers assigned to workgroups that are distrustful may use higher levels of
force during encounters with citizens. As such, this objective builds on McCluskey et
al.’s (2005) prior research on peer group aggressiveness and ofﬁcer use of force behavior
but extends it by taking a multilevel approach and also by determining if ofﬁcers’
attitudes represent valid, contextual constructs at the workgroup level. Speciﬁc details of

the data and analytic procedures are discussed in Chapter Seven.

100

CHAPTER FIVE: WORKGROUP EFFECTS ON OFFICER ATTITUDES

This chapter presents the results of the study’s ﬁrst research objective which tests
the following three research questions: 1) Are there signiﬁcant differences in ofﬁcers’
attitudes between workgroups?, 2) How do signiﬁcant attitudinal differences at the
workgroup level compare to other formal organizational levels of policing (i.e., precincts
and departments)?, and 3) Do ofﬁcers within workgroups share similar attitudinal
dispositions? Thus, the primary objectives of this chapter are to test for the presence and
amount of between-group variability in ofﬁcers’ attitudes and to determine if there is
signiﬁcant homogeneity of ofﬁcer attitudes within workgroups. To the extent that
empirical support exists, then there is evidence that ofﬁcers’ workgroups serve as
important organizational contexts for understanding the occupational attitudes of police.

The chapter begins by discussing the analytic techniques used to test the above
research questions. Next, the results of each of the analyses are presented. The results
from this chapter impact the types of analyses and potential results of the following two
chapters. For example, the results of the ﬁrst two research questions affect whether
workgroup level compositional characteristics might possibly explain variation in ofﬁcer
attitudes (Chapter Six), while the results of the third question affect whether occupational
attitudes can be viewed as contextual, group level measures that inﬂuence ofﬁcer
behavior (Chapter Seven). As such, the chapter concludes with a discussion of how the

results inﬂuence the approaches taken in the next two chapters.

Analyses & Results
Data analyses proceeds in three stages. First, descriptive statistics of the eight

attitudinal measures are presented. Measures of central tendency and dispersion are used

101

to provide a descriptive summary of each of the measures. The distributions of each
attitudinal measure are also discussed to assess how ofﬁcers’ attitudinal responses are
distributed across each index score or response category and to determine if there are any
potential issues with missing data. Next, multilevel modeling techniques are used to
determine the nature and amount of between-group variation in ofﬁcer attitudes across
workgroups, precincts, and departments. Finally, group-mean reliabilities and within—
group agreement values are calculated to assess the nature of the group level properties of

ofﬁcer attitudes.

Descriptive Statistics

Descriptive summaries of each of the eight attitudinal measures are provided to
show how the ofﬁcers surveyed in the study viewed citizens, order maintenance roles,
supervision, management, their job, law-enforcement tactics, and crime-ﬁghting
attitudes. In addition, attention is given to the amount of missing responses present across
each measure to see if missing data might be problematic for subsequent analyses. Table
5.1 reports ofﬁcers’ average responses, standard deviations, and the minimum and
maximum categories for each attitudinal index or item. Appendix C displays the
frequencies and percentages for each value to show the distribution of ofﬁcers’
responses.

With respect to citizen distrust, the results indicate that ofﬁcers were not overly
distrustful of citizens. Recall that this additive measure was coded so that higher values
represent greater distrust of citizens. Based on the distributions reported in Appendix C,
the majority of ofﬁcers’ responses (i.e., 71.8%) were in the middle of the index. Also,

Ofﬁcers had a mean response of approximately ﬁve on the index that ranged from two to

102

eight. This distribution is similar to ﬁndings from the POPN study in which the majority
of ofﬁcers were also not found to be distrustful of citizens (Paoline, 2001).

Table 5.1: Descriptive Statistics of Ofﬁcers’ Occupational Attitudes (N=1,022)

 

 

Attitudes N Mean SD. Min. Max.
Citizen Distrust (2 items; (1:81) 1,008 4.95 1.53 2 8
Order Maintenance (6 items; a=.75) 1,009 13.83 3.24 6 24
Direct Supervisors (2 items; a=.70) 1,012 3.67 1.47 2 8
Top Management (3 items; a=.78) 999 8.58 2.13 3 12
Job Satisfaction (3 items; a=.63) 1,001 9.12 1.80 3 12
Aggressiveness (1 item) 1,016 3.11 0.75 1 4
Selective Enforcement (1 item) 1,009 1.91 0.73 1 4
Crime ﬁghting (1 item) 1,017 2.93 0.74 l 4

 

Ofﬁcers also held broader role orientations that encompass order maintenance
type activities. Scores on the additive scale ranged from six to twenty-four with higher
values indicating less acceptance of order maintenance roles. On average, ofﬁcers’
responses were located toward the lower end of the scale (i.e., )7: 13.83) as the majority
(i.e., 71.8%) had scores of ﬁfteen or lower on the index. The results in Appendix C also
illustrate wide variation in ofﬁcers’ responses on this index, indicating that ofﬁcers did
use the entire range of response options.

With respect to ofﬁcers’ views toward supervisors and management, descriptive
results show that while ofﬁcers held more positive attitudes toward their direct
supervisors, they held more negative views toward top management. Again, both additive
indices were coded so that higher values represent more negative attitudes. For the direct
Supervisor measure, the majority of ofﬁcers (i.e., 73.8%) responded on the lower end of
the distribution. Thus, ofﬁcers held more positive attitudes toward their supervisors. On
the other hand, the majority of ofﬁcers (i.e., 68%) had values of eight or above on the top
management index. Ofﬁcers held more negative attitudes toward top management in their
reSpective departments.

103

Ofﬁcers also reported being satisﬁed with their jobs. Scores on the additive scale
ranged from three to twelve with higher values associated with greater satisfaction. On
average, ofﬁcers’ responses were located toward the upper end of the scale (i.e., f:
9.12). Appendix C also shows that the majority (i.e., 81.7%) had scores of eight or higher
on the index. The results in Appendix C also illustrate this pattern when examining
ofﬁcers’ responses across the entire distribution as few ofﬁcers had values of seven or
lower on the index.

With respect to the three single-item measures associated with ofﬁcers’ views
toward the importance of law enforcement (e.g., crime ﬁghting) and policing tactics, the
descriptive results show that the majority of ofﬁcers held highly aggressive attitudes as
over 81 percent responded as agreeing somewhat or agreeing strongly to this item. On the
other hand, the majority of ofﬁcers disagreed with selective enforcement practices as
over 82 percent disagreed somewhat or strongly to this item. Although ofﬁcers on
average agreed somewhat with the importance of law enforcement (i.e., 7c=2.93), a
substantial number of ofﬁcers also disagreed somewhat (22%) or agreed strongly (21.2%)
with this statement. Thus, ofﬁcers can be described as varying more in their attitudes
toward crime ﬁghting than in their attitudes toward aggressiveness and selectivity.

The descriptive results in both Table 5.1 and Appendix C also indicate the amount
of missing data for each attitudinal measure. Across each of the eight attitudinal
measures, missing responses ranged from less than one percent to just over two percent
0f the total number of cases, a very small amount. Because of these low percentages,

missing data are not considered to be a problem for the analyses conducted in this

104

chapter. As a result, cases with missing data will be deleted listwise from all subsequent

analyses.

Between-group Variability

Separate one-way, random effect ANOVA models for each of the eight attitudinal
measures are conducted for each formal organizational level (i.e., workgroup, precinct,
and department) using the Mplus 5.21 software. Each attitude represents the outcome of
interest, and ofﬁcer membership within each level represents the independent variable.
The random effect ANOVA models are a type of multilevel model that provides
important information about the nature and extent of between group-variability across
each of the three organizational levels. First, the models partition variation in ofﬁcer

attitudes into both within- and between- level components. Second, they test whether the

within- and between-level variance components are signiﬁcant.14 Finally, the results also
provide an intraclass correlation coefﬁcient, the ICC(1), for each of the eight models

(Raudenbush & Bryk, 2002). The ICC(l) values are calculated using the following

formula from Raudenbush and Bryk (2002) 15:

‘

4 Unlike HLM software, Mplus software conducts a signiﬁcance test for the within-group variance
Components for continuous variables. Mplus does not estimate within-group variance components for
categorical variables due to heteroskedasticity (see Muthen & Muthen, 2007). The attitudes of
aggressiveness, selectivity, and crime ﬁghting are treated as ordinal variables in the between-group
analyses because they are single-item measures based on four point Likert scales. Thus, within-group
variance estimates are not estimated. This is the reason for the “N/A” designations in Table 5.2.

The ICC(l) values for aggressiveness, selectivity, and crime-ﬁghting were calculated differently. Here,
categorical ANOVA models based on the logistic distribution were conducted in Mplus. In order to

calculate ICC(l) values for these measures, prior research substitutes 7T2 /3, or the assumed variance of
the logistic distribution, for the within-level variance component (Guo & Zhao, 2000; Hedeker & Gibbons,
2006). This procedure was used to calculate the ICC(l) values for these three measures.

105

700

7." +02

00

ICC(1)=

Where:

T00 = the between-level variance component
02 = the within-level variance component

The ICC(l) values represent the ratio of between-group variance to total variance
in the attitudinal measures. ICC(l) values greater than zero indicate the presence of group
effects or that group membership affects ofﬁcer attitudes. Interpreted another way, these
coefﬁcients represent the amount of variation in ofﬁcers’ attitudes that can be explained
by group membership (Bliese, 2000). Therefore, the ICC(I) also indicates the magnitude
of the group effects. Collectively, the results from the random effect ANOVA models and
the ICC(l) coefﬁcients will answer whether ofﬁcer attitudes differ between workgroups,
precincts, and departments as well as the amount of variation in attitudes that can be
explained by group membership.

Table 5.2 presents the between-group variability results for each of the eight
attitudinal constructs across workgroup, precinct, and department levels. Since the
Primary research question involves the amount and nature of between-group variability at
the workgroup level, discussion of the results begins here. At the workgroup level,
Signiﬁcant between-group variability was found for seven of the eight attitudinal
measures. Ofﬁcers’ attitudes measuring citizen distrust, order maintenance, direct
SUpervisor, top management, job satisfaction, aggressiveness, and crime ﬁghting were
found to vary signiﬁcantly across workgroups, indicating the presence of group effects.

Furthermore, the ICC(l) values revealed that between four (i.e., order maintenance) and

106

inner l
txp‘ainti
mirage:
trigger
found to

CEIOTCCI

“10:11:17
7111'1351
consist
p5} chi
hired l
10 1.1.21
lilllit’x

nazurc

111113111
lEl'Cl ill
malﬁlcij
Cfinite-ti

Values 1;

fourteen (i.e., top management) percent of the variation in these seven attitudes can be
explained by workgroup membership. For example, the ICC(l) value for top
management indicates that fourteen percent of the variation in ofﬁcers’ views toward top
management can be accounted for by their assigned workgroup. The only attitude not
found to vary signiﬁcantly between workgroups was ofﬁcers’ views toward selective
enforcement.

When examining the magnitude of the between-group variability results at the
workgroup-level, many of the ICC(l) values may appear to be low in magnitude with the
highest value being 0.14 for top management. The magnitude of these values, however, is
consistent with the results from other attitudinal studies within organizational
psychology. For example, Bliese (2000) notes that ICC(l) values for attitudinal items
based on Likert scales will rarely reach 0.30 and that most values will range between 0.05
to 0.20 due to range restriction associated with attitudinal items and scales. Furthermore,
values within this range have still been found to be important for understanding the
nature of group effects (Bliese, 2000). Thus, even though the ICC(l) values may appear
to be low in magnitude they are within a range of values reported by prior research to be
important.

At the precinct and department levels, the results in Table 5.2 show signiﬁcant
variation as well. Speciﬁcally, the between-level variance components at the precinct
level are similar to the results found at the workgroup level for citizen distrust, order
maintenance, direct supervisor, top management, job satisfaction, aggressiveness, and
crime-ﬁghting. The ICC(l) values, however, are lower in magnitude. Most of these

Values range from 0.04 to 0.08, and the highest being 0.12 for top management. At the

107

department level only ofﬁcers’ attitudes associated with order maintenance, direct
supervision, top management, and job satisfaction display signiﬁcant between-group
variance components and they also display lower ICC(l) values.

Table 5.2: One-way Random Effect AN OVA Results

 

Attitudinal Measures Within-Level Between-Level ICC(l)

 

Variance Variance

Workgroup Level (N=988) (N=61)
Citizen Distrust 2.22 (.08)*** 0.14 (.04)** 0.06
Order Maintenance 9.86 (.54)*** 0.47 (.14)** 0.04
Direct Supervisor 1.97 (.10)*** 0.21 (.06)*** 0.10
Top Management 3.91 (.19)*** 0.63 (.16)*** 0.14
Job Satisfaction 3.00 (.18)*** 0.27 (.08)** 0.08
Aggressivenessa N/A 0-38 (-10)*** 0-10
Selectivitya N/A 0.02 (.04) 0.01
Crime ﬁghtinga N/A 0.17 (.06)** 0.05

Precinct Level (N=1,022) (N=20)
Citizen Distrust 2.22 (.10)*** 0.12 (.04)** 0.05
Order Maintenance 10.14 (.59)*** 0.37 (.1 l)** 0.04
Direct Supervisor 2.10 (.09)*** 0.09 (.03)* 0.04
Top Management 4.00 (.16)*** 0.55 (.15)** 0.12
Job Satisfaction 3.09 (.16)*** 0.15 (.04)*** 0.05
Aggressivenessa N/A 0-27 £10)“ 0-08
Selectivitya N/A 0.01 (.02) 0.00
Crime ﬁghtinga N/A 0.15 (.06)* 0.04

Department Level (N=1,022) (N=5)
Citizen Distrust 2.22 (.12)*** 0.15 (.09) 0.06
Order Maintenance 10.19 (.87)*** 0.28 (.14)* 0.03
Direct Supervisor 2.10 (.07)*** 0.08 (.03)** 0.04
Top Management 4.06 (.26)*** 0.61 (.31)* 0.13
Job Satisfaction 3.12 (.12)*** 0.10 (.05)* 0.03
Aggressivenessa N/A 0-26 (415) 0-07
Selectivitya N/A 0.00 (.01) 0.00
Crime ﬁghtinga N/A 0.17 (.09) 0.05

 

a . . . .
***p<.001, **p<.01, *p<.05; Results based on categorical models, wrthrn-level variances are not
estimated; Note: Numbers in parentheses denote standard errors of variance component estimates.

With respect to the within-level variance component results in Table 5.2, it should

be noted that signiﬁcant within-level variation exists for the ﬁve attitudes for which

108

 

1':

t.

Elli

within-level variance components were estimated (i.e., citizen distrust, order
maintenance, direct supervisor, top management, and job satisfaction) across all
organizational levels. This ﬁnding is also important to take note of because it indicates
that individual level ofﬁcer factors may also account for attitudinal variation among
ofﬁcers. As such, these potential factors will need to be included in subsequent models
that seek to explain attitudinal differences in the following chapter.

The results in Table 5.2 provide preliminary support for the notion that
workgroups represent the part of the formal organization of policing that has the greatest
inﬂuence on ofﬁcers’ occupational attitudes. Speciﬁcally, variation in ofﬁcer attitudes
was found to be more prevalent at the workgroup level than at the precinct or department
levels. Furthermore, the ICC(l) values were generally stronger at the workgroup level.
Although this provides empirical support for the study’s hypothesis, the conclusion that
workgroups are the most salient inﬂuence on attitudes cannot be readily made. First,
signiﬁcant results are still present at the precinct and department levels. Second, the
between-group variance components are difﬁcult to compare across levels due to the fact
that there are differences in the number of organizational units analyzed at each level. As
Kozlowski and Klein (2000) have noted, it is easier to detect variation when there are a
larger numbers of groups. The stronger workgroup effects might be due to the fact that
there are 61 workgroups compared to only 20 precincts and 5 departments available for
analysis. As such, it is necessary to control for variation at the precinct and department
levels to determine if signiﬁcant attitudinal variation between workgroups remains.

Additional tests are conducted that take into account the fact that workgroups are

nested within precincts and departments. When data and organizational units are nested

109

within higher-level organizational units, such as the case is here, Mplus allows for the
clustering of data at the higher level to be accounted for and adjusts the standard errors at
the primary level of interest (e.g., workgroups) to ensure that Type I errors are not made
(Muthen & Muthen, 2007). In the current context, two sets of random effect ANOVA
models are conducted for each of the attitudinal measures: one that controls for the
nesting of workgroups within precincts and another that controls for the nesting of
workgroups within departments. 16 If signiﬁcant between-group variation in ofﬁcer
attitudes at the workgroup-level remains after accounting for clustering by precinct and
department, then the claim that workgroups are important organizational contexts that
inﬂuence ofﬁcer attitudes is strengthened.

Table 5.3 reports the results of separate one-way, random effect ANOVA models
that control for the clustering of workgroups within precincts and departments. An
interesting ﬁnding emerges when looking at the results for ofﬁcers’ views toward top
management. Speciﬁcally, the between-group variance component becomes non-
signiﬁcant when controlling for the clustering of workgroups within departments. In the
previous analyses, workgroups had the strongest effect on this attitude with the ICC(l)
value in Table 5.2 indicating that fourteen percent of the variation in ofﬁcers’ views
toward top management could be explained by workgroup membership. The results in
Table 5.3 suggest that differences at the department level likely account for this variation.

This ﬁnding makes sense when examining the nature of this attitudinal construct.
For example, the construct references the department level by asking ofﬁcers whether

‘

Mplus only allows two-level multilevel models to be analyzed. As such, it is not possible to control for

both precinct and department effects simultaneously in the same model (See Muthen & Muthen, 2007:
22 1).

110

they believe that top management supports, rewards, and treats ofﬁcers fairly. When
examining the inﬂuence of the formal organization of policing on this particular attitude,
it is differences in department policies and management practices that likely account for
this variation. Thus, differences in ofﬁcers’ views toward top management are not likely
to be important across workgroups but rather across departments.

Table 5.3: Between-Workgroup Variability Accounting for Precinct & Department
Levels (N=988)

 

Attitudinal Measures Between-Level Variance Between-Level Variance
Clustering by Precinct Clustering by Department

Citizen Distrust 0.14 (.05)** 0.14 (.06)*

Order Maintenance 0.47 (.11)"‘** 0.47 (.07)***

Direct Supervisor 0.21 (.06)** 0.21 (.06)**

Top Management 0.63 (.21)** 0.63 (.36)

Job Satisfaction 0.27 (.07)*** 0.27 (.07)***

Aggressivenessa 0.38 (.12)** 0.38 (.16)*

Selectivitya 0.02 (.03) 0.02 (.01)

Crime ﬁghtinga 0.17 (.08)* 0.17 (.08)*

 

a . .
***p<.001, **p<.01, *p<.05; Results based on categorical models; Note: Numbers in parentheses
denote standard errors of variance component estimates.

Despite the results for top management, overall the results show that even when
controlling for clustering at the precinct and department levels, ofﬁcers’ attitudes
associated with citizen distrust, order maintenance, direct supervisor, job satisfaction,
aggressiveness, and crime ﬁghting still signiﬁcantly vary across workgroups. In essence,
six of the seven attitudes with signiﬁcant between-group variance components at the
workgroup level remained signiﬁcant after controlling for precinct and department
effects. These ﬁndings strengthen the claim that workgroups represent important

organizational contexts that inﬂuence ofﬁcer attitudes.

111

Group Reliability & Within-group Agreement

The ﬁnal set of analyses assesses the reliability and validity of ofﬁcers’
occupational attitudes as group level measures. First, to test the reliability of the group
level means for each attitudinal construct a second intraclass correlation coefﬁcient is
used and is derived from information from the ANOVA models. Speciﬁcally, ICC(2)
coefﬁcients are calculated using the Spearman-Brown fomrula (Shrout & F leiss, 1979):

k(ICC1)

ICC(2) 2 1+ (k —- 1) * ICC(l) ’

 

Where:

ICC(l) = the intraclass correlation coefﬁcient from the random effect ANOVA results
k = the average number of ofﬁcers per workgroup

These ICC(2) coefﬁcients estimate how stable the values for the aggregated attitudinal
measures will be, which is necessary when examining the group level characteristics of
attitudinal constructs. ICC(2) coefﬁcients of 0.70 or above generally indicate acceptable
reliability of the group means, coefﬁcients between 0.50 and 0.70 indicate marginal
reliability, and coefﬁcients below 0.50 generally indicate poor reliability (Klein, Bliese,
Kozlowski, Dansereau, Gavin, Grifﬁn et al., 2000).

ICC(2) estimates alone, however, do not provide a sufﬁcient basis for deciding if
ofﬁcers actually share attitudes or if attitudes can be aggregated to the workgroup level as
contextual measures. The reliability of measures does not necessarily reﬂect agreement
among individuals within groups (James, 1982; Kozlowski & Hattrup, 1992). In order to
empirically test the extent to which ofﬁcers within workgroups share attitudes, it must be
demonstrated that ofﬁcers within workgroups have sufﬁcient levels of agreement on the

attitudinal constructs. This means that ofﬁcers within workgroups should provide similar

112

ratings on each attitudinal measure. Prior research in the organizational psychology

literature has recommended the use of two methods for demonstrating perceptual

agreement among individuals and for validating shared unit properties: rWg values
(James, Demaree, & Wolf, 1984, 1993; Kozlowski & Hattrup, 1992) and the average
deviation index (Burke & Dunlap, 2002; Burke, Finkelstein, & Dusig, 1999).

A popular method for assessing within-group agreement has been the rWg index of

interrater agreement (Cohen, Doveh, & Nahum-Shani, 2009). This index assesses

agreement by estimating the proportion of systematic variance in ofﬁcer attitudes in

relation to the total variance in the attitudes (James et al., 1984: 87). Values of rWg range

from zero to one with higher values indicating greater levels of agreement. For single

item attitudinal measures, agreement is calculated using the following equation (James et

aL,l984y
2
= S X
1..., 1' T
a
E
Where:
S; = the observed variance on variable X

012; = the expected variance on variable X when there is a complete lack of agreement

For the purposes of the current study, the expected variance will be assumed to follow a
uniform distribution or the variance expected if all responses were due solely to random

measurement error (James et al., 1984). As such, the expected variance is derived via the

following formula: 0'12; = (A2 - l )/ 12, where A equals the number of response categories.

113

The above equation has been extended to multi-item indices (i.e., rwg(j)) by taking the

average of each item’s observed variance for the numerator of the above equation (James

et al., 1984; Lindell, Brandt, & Whitney, 1999).

Despite the fact that the rwg index has been the most common method for

estimating agreement, it does have limitations. Most notably, the speciﬁcation of the
expected variance (i.e., the denominator in the above equation) based on the assumption
of a complete lack of agreement has been criticized (Burke & Dunlap, 2002; Burke et al.,
1999; Lindell et al., 1999). For example, Burke et a1. (1999: 51) noted that, “to specify a
random response distribution, the researcher needs to indicate the proportion of
respondents that would respond to each response by chance or with respect to a particular

type of response bias. . .Appropriately specifying a null response distribution is a difﬁcult

task...” This is problematic because rWg values may either be overestimated or

underestimated as a result (Burke et al., 1999).

As such, alternative methods for estimating agreement have been derived. The

average deviation index (e. g., ADM for single items or ADM(J) for multiple items) is a

second type of agreement index used in the current study. The AD index represents “the
extent to which each item rating differs from the mean. . .item rating, summing the

absolute values of these deviations..., and dividing by the number of deviations” (Burke

et al., 1999: 53). ADM values also range from zero to one; however, lower ADM values

indicate greater agreement (Burke & Dunlap, 2002). For example, an ADM value of zero

would indicate perfect agreement among ofﬁcers within a workgroup on attitudinal items.

114

For single items the AD index is calculated using the following equation from Burke et

al. (1999):

N __
Ek=1lxjk—xll
N

 

ADM —

Where:

N = Number of ofﬁcers for the item
xjk = is the kth ofﬁcers rating on the item

Fri: the mean of all ofﬁcers’ scores on the item

For multiple item measures, ADM(J) values are calculated by taking the average of the

item ADM values (Burke & Dunlap, 2002). There are two advantages for using average

deviation indices. First, it does not require the explicit modeling of a null response

distribution. Second, the values can be more easily interpreted because they are in the
units of the attitudinal response scale (Burke & Dunlap, 2002). For example an ADM
value of 0.50 would indicate that ofﬁcers’ responses within a workgroup were on average

half a unit apart on the Likert scale. Thus, it has been recommended that both rWg and

ADM indices be reported in studies (Dunlap, Burke, & Smith-Crowe, 2003). 17

In order to assess whether ofﬁcers within workgroups display sufﬁcient levels of
attitudinal agreement to warrant aggregation, two issues are considered: statistical
signiﬁcance tests and the magnitude of the agreement values. Dunlap et a1. (2003) has

developed a method and program to test whether agreement values are signiﬁcantly

 

l7 . . . . . .
From th1s pomt on rwg and ADM are used to refer to both srngle-1tem and mult1-1tem measures.

115

different from chance agreement using an approximate randomization test.18 Therefore,
the ﬁrst step is to assess whether agreement within workgroups is statistically signiﬁcant.
As Dunlap et a1. (2003) points out, however, statistical signiﬁcance is not

sufﬁcient for justifying aggregation. The magnitude of the agreement values should also

be considered. A rule of thumb for rwg values is as follows: values less than 0.59 indicate

a lack of agreement, 0.60 to 0.69 indicates weak agreement, 0.70 to 0.79 indicates
moderate agreement and 0.80 and above indicates strong agreement (Brown &
Hauenstein, 2005; Cohen et al., 2009). For average deviation indices a standard for
determining practical signiﬁcance has generally been accepted by dividing the number of

response options in the Likert measures by six (Burke & Dunlap, 2002). For the current
study, ADM values equal to or less than 0.67 indicates acceptable levels of agreement
within workgroups (Burke & Dunlap, 2002; Dunlap et al., 2003).19

It should be noted that unlike the calculation of the intraclass correlation

coefﬁcients which represent summary measures of between-group variation and group

mean reliabilities, rwg and ADM estimates are calculated for each individual workgroup.

In the current context, this means there will be 61 rWg and ADM values and signiﬁcance

tests. The decision to determine if there is sufﬁcient agreement across the sample of

workgroups is usually made by assessing the average or median rWg and ADM values

 

18 . . .
The program rs called “agreeexe” and rs avarlable for download at
www.tulane.edu/~dunlap/psylib.html. (Dunlap et al., 2003)

19 See Burke & Dunlap (2002) for the derivation used to identify six as the appropriate denominator. In the
current study, acceptable levels of agreement are decided upon using 0.67 as the cutoff. This cutoff is
calculated by taking the number of Likert responses for the survey items (i.e., 4) and dividing by the
denominator of six: 4/6 =0.67.

116

across workgroups (e.g., Naumann & Bennett, 2000) as well as reporting the percentage
of workgroups that displayed. statistically signiﬁcant levels of agreement (Dunlap et al.,
2003).

The group reliability and within-group agreement results of ofﬁcers’ attitudes
within workgroups are reported in Table 5.4. For each of the eight attitudinal constructs,

group-mean reliabilities were estimated and are based on ICC(2) calculations.

Additionally, within-group agreement results based on rwg and ADM indices are also

reported. Speciﬁcally, the median rWg and ADM values as well as the percentage of

values that were found to be statistically and practically signiﬁcant across workgroups are
reported for each attitude. Within-group agreement results for each of the 61 individual

workgroups from which the results in Table 5.4 were derived from are provided in

 

 

 

 

Appendix D.
Table 5.4: Group Reliability & Within-Group Agreement Results (N=6l)
Attitudinal Measures ICC(2) rwggesults ADM Results
Median °_/o_Sig, Median % Sig. °/o Pgwtical Sig.
Citizen Distrust 0.51 0.46 68.8 0.63 62.2 60.7
Order Maintenance 0.40 0.52 75.4 0.61 77.0 75.4
Direct Supervisor 0.64 0.50 68.8 0.59 73.8 80.3
Top Management 0.73 0.50 72.2 0.62 63.9 67.2
Job Satisfaction 0.58 0.54 80.3 0.58 80.3 88.5
Aggressiveness 0.64 0.61 86.9 0.56 81.9 86.9
Selectivity 0.14 0.62 82.0 0.51 85.2 88.5
Crime ﬁghting 0.46 0.61 75.4 0.53 83.6 83.6

 

 

 

 

 

With respect to the reliability of the workgroup level means for each attitude, the
ICC(2) values in Table 5.4 indicate that ﬁve of the eight constructs met at least the cutoff
criteria for marginal reliability (i.e., ICC(2) > 0.50). Speciﬁcally, the attitudes of citizen

distrust, direct supervisor, job satisfaction, and aggressiveness had ICC(2) values

117

associated with marginal group-mean reliabilities. Only ofﬁcers’ views toward top
management had an ICC(2) value above the 0.70 cutoff for acceptable reliability.
Ofﬁcers’ views toward both order maintenance and crime-ﬁghting roles had ICC(2)

values just below the 0.50 cutoff for marginal reliability. Finally, the results for

selectivity show poor reliability for estimates of workgroup means on this attitude.20
The rWg results in Table 5.4 provide an assessment on whether ofﬁcers within

workgroups share similar attitudinal dispositions. Across all eight attitudes, the majority
of workgroups had statistically signiﬁcant levels of within-group agreement. For
example, citizen distrust and direct supervisor attitudes had statistically signiﬁcant levels
of agreement in over two-thirds (68.8%) of workgroups while order maintenance
(75.4%), top management (72.2%), and crime-ﬁghting attitudes had statistically
signiﬁcant levels of agreement in almost three-quarters of workgroups. Finally, ofﬁcers
in over eighty percent of workgroups had statistically signiﬁcant agreement levels for the
attitudes of job satisfaction, aggressiveness, and selectivity. In terms of statistical

signiﬁcance, there is empirical support for sufﬁcient within group agreement based on

rWg values.

 

20 It should be noted that ICC(2) values are inﬂuenced by group size and the amount of between-group
variability (Bliese, 1998). Recall that these coefﬁcients are calculated using the average size of workgroups
and the ICC(l) values reported in Table 5.2. As such, attitudes with low ICC(l) values and small group
sizes will also have lower ICC(2) values. Since the ICC(l) values in Table 5.2 were within the established
range reported by prior research for attitudinal measures (with the exception of selectivity) a partial
explanation for the lower ICC(2) values is likely due to the fact that workgroups only had an average of
approximately sixteen ofﬁcers. In other words, if workgroups contained more ofﬁcers (which is restricted
given the structure of workgroups in the current study), then more reliable estimates of the group means
would likely result.

118

An examination of the magnitude of the median rWg values in Table 5.4, however,

is not as supportive. Only three attitudes have median rWg values that would meet the

cutoff criteria for weak agreement (i.e., 0.60 to 0.69): aggressiveness, selectivity, and

crime ﬁghting. Based on the magnitude of the median rWg values for the other ﬁve

attitudes, the ﬁndings indicate a lack of agreement. Thus, although within-group

agreement across the 61 workgroups is statistically signiﬁcant, the magnitude of

agreement based on the rWg values appears to be rather weak.

Within-group agreement results based on ADM values shows similar results when

looking at the percentage of workgroups that displayed statistically signiﬁcant levels of
agreement across the eight attitudinal measures. The two exceptions are for top
management and crime-ﬁghting attitudes. Here, fewer workgroups displayed statistically
signiﬁcant levels of agreement for views toward top management while a greater number
of workgroups displayed statistically signiﬁcant levels of agreement for crime-ﬁghting

attitudes.

In terms of practical signiﬁcance, or the percentage of workgroups that had ADM

values equal to or less than the cutoff for practical signiﬁcance (i.e., 0.67), the majority of
workgroups had sufﬁcient levels of agreement. Over eighty percent of workgroups had
practically signiﬁcant agreement values for attitudes associated with direct supervisor,
job satisfaction, aggressiveness, selectivity, and crime ﬁghting. Views toward citizens
had the lowest percentage of practical signiﬁcance; still over sixty percent of workgroups

had practically signiﬁcant levels of agreement on this attitude. Finally, when looking at

119

the median ADM values, all eight attitudes had median values below the 0.67 cutoff for

acceptable levels of agreement. In fact, ﬁve of the eight attitudes had median values in
the 0.50 range. This means that, on average, ofﬁcers within workgroups were

approximately half a unit apart on the Likert measurement scale. Overall, the results from

the ADM analysis are more supportive of within-group agreement than the rwg analysis.

Discussion

This chapter empirically examined whether ofﬁcers’ occupational attitudes were
inﬂuenced by the formal organization of policing. Of particular interest was the extent to
which ofﬁcers’ assigned workgroups inﬂuenced their attitudes. Speciﬁcally, analyses
were conducted to determine if there was signiﬁcant variation in ofﬁcer attitudes between
workgroups, examined this variation in relation to other formal organizational levels, and
accounted for the fact that workgroups were nested within these higher organizational
levels. Additionally, analyses were conducted to determine whether ofﬁcers within the
same workgroup shared similar occupational attitudes.

Although the results were presented and discussed as separate research questions,
it is necessary to examine them collectively as well. This is due to the fact that the results
reported here are important for understanding the approaches taken for the remaining
research objectives of the study. Table 5.5 presents a summary of the results for this
chapter related to the presence and nature of workgroup effects on each of the eight
attitudes in the key areas of between-group variability, group-mean reliability, and
within-group agreement. The following discusses the impact that these results have on

the approaches taken for the two remaining research objectives.

120

With respect to Research Objective #2, there must be signiﬁcant between-group
variability on ofﬁcers’ attitudes in order for workgroup compositional characteristics to
be considered as explanations for attitudinal differences. Simply stated, if ofﬁcers’
attitudes do not differ across workgroups, then examination of workgroup compositional
characteristics is not necessary because there would be no variation to explain. The
results in Table 5.5 show that six of the eight attitudinal measures had signiﬁcant
between-group variance components. Additionally, it was reported in Table 5.2 that
between ﬁve and ten percent of the variation in these attitudes can be explained by
workgroup membership. As a result, workgroup compositional characteristics (i. e.,
proportion male officers, proportion white ofﬁcers, average educational levels, and
average length of experience) are viable factors that could explain diﬂerences in oﬁicers’
attitudes toward citizens, order maintenance roles, direct supervisors, job satisfaction,
aggressiveness, and crime Ighting. The results of these tests are presented and discussed
in Chapter Six.

Table 5.5: Summary of Results for Workgroup Effects & Ofﬁcer Attitudes

 

Attitude Between-group Group-Mean Within-group Within-group
Variability? Reliability? Agreement? Agreement?
(rwg) (ADM)
Citizen Distrust Yes Marginal Insufﬁcient Sufﬁcient
Order Maintenance Yes Poor Insufﬁcient Sufﬁcient
Direct Supervisor Yes Marginal Insufﬁcient Sufﬁcient
Top Management No Acceptable Insufﬁcient Sufﬁcient
Job Satisfaction Yes Marginal Insufﬁcient Sufﬁcient
Aggressiveness Yes Marginal Weak Sufﬁcient
Selectivity No Poor Weak Sufﬁcient
Crime ﬁghting Yes Poor Weak Sufﬁcient

 

With respect to Research Objective #3, a key proposition is that ofﬁcers within
workgroups share similar attitudes, and because of this, attitudes can be aggregated to the

workgroup level to represent valid, reliable measures of the work environment (e.g.,

121

workgroup climates). Furthermore, these aggregate measures are hypothesized to
inﬂuence ofﬁcer behavior. Each of the analyses conducted in this chapter bears
importance as to whether there is sufﬁcient evidence for treating ofﬁcers’ attitudes in this
manner. The evidence used to consider and justify this is discussed below.

First, the summary of within-group agreement results in Table 5.5 indicates
sufﬁcient levels of agreement among workgroups for the measures of aggressiveness,

selectivity, and crime ﬁghting. Discrepancies exist, however, for the remaining ﬁve

attitudes. Generally, the ADM results indicate sufﬁcient agreement while the rwg results

indicate insufﬁcient agreement. When results differ as is the case here, prior research
offers insight for making decisions. Klein et al. (2000: 520) note that “when the indices
lead to differing conclusions regarding the merits of aggregation, research may rely on
theory, prior research, and/or belief in the superior merits of one of the indices in
deciding whether to aggregate their measure(s).” In the current study, theoretical support

for the treatment of attitudes as measures of the workgroup environment was detailed in
Chapter Three. Furthermore, the limitations previously identiﬁed for the rWg indices (e.g.,
Burke & Dunlap, 2002; Burke et al., 1999; Lindell et al., 1999) suggest that giving

preference to the ADM results is arguably warranted. Based upon this, it is argued that

each of the eight attitudes possesses sufﬁcient levels of agreement at the workgroup level.
In other words, each attitudinal measure displays the necessary properties of validity to
be further considered as workgroup constructs.

Second, the group-mean reliability results illustrate whether the group level

constructs can be viewed as reliable aggregate measures. As Bliese (2000: 375) points

122

out, “estimating group-mean reliability — ICC(2) is always important. . . groups need to
have reliably different mean values on the construct of interest if one hopes to detect
emergent relationships.” Based on the summary presented in Table 5.5, only ﬁve of the
eight attitudes possess sufﬁcient reliability levels: citizen distrust, direct supervisor, top
management, job satisfaction, and aggressiveness. Views toward order maintenance,
selectivity, and crime ﬁghting had poor group-mean reliability values. Thus, even though
these three measures represented valid workgroup measures, they do not necessarily
represent reliable ones. Thus, order maintenance, selectivity, and crime ﬁghting were
removed from consideration for aggregation due to low group-mean reliabilities. Finally,
the between-group variability results must also be considered. In order for workgroup
attitudinal measures to be viewed as viable explanations for ofﬁcer behavior, attitudes
must be found to signiﬁcantly vary across workgroups. As such, top management was
further dropped from consideration for aggregation since between-group variability was
found to be non-signiﬁcant.

Overall, Table 5.5 indicates that the attitudinal measures of citizen distrust, direct
supervisor, job satisfaction, and aggressiveness represent valid and reliable measures of
the workgroup environment. These were the four attitudes that sufficiently met the
empirical criteria for within—group agreement, group-mean reliability, and between-
group variability. Thus, these four measures can be aggregated to the workgroup-level
and used as representations of the workgroup environment. These four measures are
hypothesized to inﬂuence ofﬁcer use of force behavior. The results of these tests are

presented and discussed in Chapter Seven.

123

CHAPTER SIX: WORKGROUP COMPOSITION & OFFICER ATTITUDES
This chapter presents the results of the study’s second research objective: whether
workgroup characteristics explain differences in ofﬁcers’ occupational attitudes. The
results from the previous chapter indicated that six of the eight attitudinal measures
signiﬁcantly differed across workgroups. These include ofﬁcers’ attitudes toward
citizens, order maintenance roles, direct supervisors, job satisfaction, aggressiveness, and
the crime-ﬁghting role. As a result, workgroup level factors can be examined as potential
explanations for these differences. Here, the impact that the compositional makeup of
ofﬁcers’ assigned workgroups (i.e., gender composition, racial composition, educational
composition, and experience levels) has on their attitudes is tested. The chapter begins by
discussing the data and measures used for this research objective. Next, the analytic

procedures are discussed. Finally, the results of the analyses are presented.

Data

The data for this second research objective comes from the ofﬁcer survey
described in the previous two chapters and the master rosters obtained from each of the
ﬁve departments. Ofﬁcers’ attitudes were taken from the survey and are the same
measures used in Chapters Four and Five. In addition, ofﬁcer education was also taken
from the survey.21 Demographic characteristics were coded from the master rosters
supplied by each department to assist with survey administration. These rosters included
the gender, race, and date of hire for every sworn ofﬁcer working in each department.

The unique project identiﬁcation number assigned by the research team to each ofﬁcer as

 

21 . .
Recall from Chapter Four that departments d1d not always keep records on ofﬁcers’ education levels. As
a result, an educational question was included on the ofﬁcer survey.

124

a part of the project was added to the roster.22 The use of the identiﬁcation numbers was

necessary to allow ofﬁcers’ demographic characteristics to be linked to survey responses.

Variables
Dependent Variables

The dependent variables of interest are the attitudinal measures of citizen distrust,
order maintenance, direct supervisor, job satisfaction, aggressiveness, and crime ﬁghting.
Based on the results from Chapter Five, these six attitudes displayed signiﬁcant between-
group variability at the workgroup level even after accounting for the precinct and
department where ofﬁcers worked. Because of this, workgroup factors may be used to
explain this variability. The six attitudinal variables are the same measures used and
described in Chapters Four and Five. As such, the indicators, codings, and properties of
validity and reliability are not repeated again here.
Individual Officer Demographic Controls

Four individual ofﬁcer variables are also included in this portion of the study.
These ofﬁcer characteristics have traditionally been the emphasis of prior research that
has taken individual level approaches for explaining attitudinal differences among
ofﬁcers (see Chapter Two). The four variables include ofﬁcer gender, race, education,
and experience. Ofﬁcer gender (l=male) and race (l=white) are both binary variables and
were coded to reﬂect the demographic attributes that have heavily dominated the policing
profession. Ofﬁcer education was based off of the survey question measuring ofﬁcers’
highest level of education completed at the time of the study. This variable was coded so

that higher values indicate a higher level of education (1: less than HS; 2=HS diploma or

 

22 . . . . . . . . . . . .
Agarn, thIS pr0ject 1dent1ﬁcat1on number rs not related to any 1dent1fytng 1nformatron that can be traced
back to ofﬁcers. Furthermore, only the research team has access to these numbers to ensure conﬁdentiality.

125

equivalent; 3=some junior college, no degree; 4=Associates degree; 5=more than two
years of college, no degree; 6=Bachelors degree; 7=some graduate work, no degree; and
8=graduate degree). Finally, ofﬁcer experience refers to years of service based on their
date of hire through the time that the survey was administered. Inclusion of these
variables is important for two reasons. First, they are used as control variables in the
empirical models. Controlling for any individual demographic effects strengthens the
empirical support for any ﬁndings at the workgroup level. Second, the workgroup
compositional measures are derived directly from these four individual ofﬁcer variables.
Workgroup Composition Measures

The current research objective hypothesizes that ofﬁcers’ attitudes can be
explained by the types of ofﬁcers who make up their assigned workgroups. The rationale
for this is similar to the individual level assumptions employed by prior research. The
theoretical framework taken for the study, however, argues that the types of ofﬁcers’ that
individual ofﬁcers are exposed to within their workgroup context will also inﬂuence their
attitudes, a multilevel perspective. For example, in workgroups with higher
concentrations of white or male ofﬁcers, ofﬁcers will more likely hold attitudes depicted
by the occupational perspective (e.g., distrust of citizens, narrow role orientations,
aggressive policing tactics, and negative views toward supervision) regardless of their
own individual characteristics because the gender and racial composition reﬂect that
which has historically dominated the policing profession. Thus, instead of individual
effects for ofﬁcer race and gender, collective, multilevel effects (i.e., contextual effects)
at the workgroup level are what are important for understanding differences in ofﬁcer

attitudes.

126

Four workgroup composition measures are examined. Gender composition is an
aggregate variable that measures the proportion of male ofﬁcers assigned to the
workgroup. Racial composition is an aggregate variable that measures the proportion of
white ofﬁcers assigned to the workgroup. Additionally, the average education level and
years of experience are also used as aggregate, compositional measures. These represent
the same characteristics that empirical research has examined at the individual level. In
the current context, however, these represent aggregate measures reﬂecting the
demographic makeup of workgroups.

Department Controls

In addition to the four workgroup composition variables, the analyses also control
for the department where ofﬁcers worked. Four dummy variables were created and are
treated as workgroup level control variables. These dummy variables measure one of the
following four sites (CSPD, PPB, F WPD, or KPD). APD was chosen as the reference

department because this site contributed the most ofﬁcers to the study sample (29%).

Analyses

Analysis begins with descriptive and bivariate statistics for the dependent, ofﬁcer,
and workgroup variables. Then, multilevel modeling is used for all multivariate analyses.
Since this research objective examines observations that occur at both the individual
ofﬁcer level and the workgroup level, multilevel modeling techniques are appropriate
(Hoffman, Grifﬁn, & Gavin, 2000; Raudenbush & Bryk, 2002). Further support for the
use of multilevel modeling was provided by the between-group variability results from
the previous chapter. Intercepts as outcome models are appropriate for this type of

research question (Hoffman et al., 2000; Raudenbush & Bryk, 2002). Workgroup

127

composition measures are the level 2 variables and individual ofﬁcer characteristics
constitute the level 1 control measures. Additionally, the models control for ofﬁcers’
department. The purpose of these models is to empirically determine whether workgroup
compositional variables predict variation in ofﬁcer attitudes while c0ntrolling for
individual ofﬁcer demographics and the department where ofﬁcers work.

It should be noted that the multilevel models conducted for citizen distrust, order
maintenance, direct supervisor, and job satisfaction represent traditional multilevel
models for continuous variables. The models for aggressiveness and crime ﬁghting, on
the other hand, are forms of hierarchical generalized linear models (HGLM) because each
of these measures is ordinal (i.e., single-item measures based on a four point Likert
scale). For these two attitudes, each multilevel model is an ordinal logistic regression
model estimated using robust maximum likelihood estimation (MLR). Intercept and slope
estimates represent cumulative logit coefﬁcients from which cumulative odds ratios can
be calculated (Muthen & Muthen, 2007; Raudenbush & Bryk, 2002). Since the primary
interest is the effect of level 2 variables (i.e., workgroup composition) on attitudes, the
level 1 ofﬁcer variables remain ﬁxed across models. Furthermore, all level 1 variables are
grand mean centered.

Grand mean centering is necessary for two reasons. First, it is the appropriate
centering option when the research question involves testing for the presence of level 2
effects while controlling for the level 1 variables. Therefore, the level 2 coefﬁcients
reﬂect the relationship between the workgroup measures and attitudes after adjusting for
the effects of the level 1 ofﬁcer variables. Second, grand mean centering allows for the

assessment of any contextual effects among the workgroup measures at level 2. For

128

example, if racial composition is signiﬁcantly associated with ofﬁcers’ attitudes while
controlling for individual ofﬁcer race, then empirical support exists that the composition
of workgroups explains attitudinal differences over and above individual ofﬁcer
characteristics. In other words, the effect of the aggregate measure represents a contextual
effect and grand mean centering allows for the direct determination of these effects
(Enders & Toﬁghi, 2007; Hoffman, 1998; Raudenbush & Bryk, 2002).

Results

Descriptive Statistics

Table 6.1 reports descriptive statistics for the attitudinal, individual ofﬁcer
demographic, and workgroup composition variables. There were 988 total ofﬁcers
surveyed who were nested within the sixty-one workgroups across the ﬁve departments.
On average, there were approximately sixteen ofﬁcers within each workgroup. These
numbers indicate a sufﬁcient number of workgroups and ofﬁcer observations to conduct
multilevel analyses (Maas & Hox, 2005). For the attitudinal measures, the descriptive
results mirror those presented and discussed in the preceding chapter.

With respect to the workgroup composition variables, the majority of workgroups
were staffed by male and white ofﬁcers. This is not surprising since male and white
ofﬁcers still largely dominate the policing profession despite efforts to increase gender
and racial diversity across departments. Furthermore, the study departments were
Primarily comprised of male and white ofﬁcers at the individual ofﬁcer level.
Workgroups were found to differ more extensively in terms of racial composition with
the proportion of white ofﬁcers ranging from 33 percent to 100 percent across
WOrkgroups. The gender composition of workgroups, however, varied considerably less

as the proportion of male ofﬁcers ranged from 66 percent to 100 percent. Average

129

expert:
tong
11011.

- 1
Em;
All“

experience levels ranged from just under two years to over eighteen years. Thus, some
workgroups were composed primarily of relatively inexperienced ofﬁcers while other
workgroups consisted of ofﬁcers with much more years of service to the department.
Finally in terms of average education levels, most workgroups were comprised of college
educated ofﬁcers ranging from some junior college to undergraduate degrees.

Table 6.1: Descriptive Statistics for Ofﬁcer Attitudes, Demographics, & Workgroup
Variables

 

 

Variable N Min. Max. Mean Std.
Deviation
Attitudinal Measures
Citizen Distrust 974 2 8 4.95 1.54
Order Maintenance 975 6 24 13.80 3.22
Direct Supervisor 978 2 8 3.68 1.47
Job Satisfaction 967 3 12 9.11 1.81
Aggressiveness 982 l 3.10 0.75
Crime ﬁghting 983 l 4 2.93 0.74
Workgroup Level
Proportion Male 61 0.66 1.00 0.89 0.08
Proportion White 61 0.33 1.00 0.79 0.17
Average Experience 61 1.83 18.04 7.64 3.74
Average Education 61 3.00 6.22 4.70 0.83
CSPD 61 0 l 0.16 0.37
PPB 61 0 1 0.21 0.41
APD (reference site) 61 0 l 0.29 0.46
F WPD 61 0 l 0.20 0.40
KPD 61 0 l 0.13 0.34
Officer Demographics
Male 984 0 l 0.89 0.30
White 980 0 1 0.79 0.40
Experience (in years) 984 0.17 34.58 7.85 6.43
Education 971 1 8 - 4.76 1.64

 

130

111.1

Th-
5&1

ti;

7’”
:4;

P‘H

«if.

as

Departmental controls are also included as workgroup level variables. Descriptive results
for these variables indicate that the majority of workgroups were from APD (29%)
followed by PPB (21%), F WPD (20%), CSPD (16%), and KPD (13%).

Examination of individual ofﬁcer demographic characteristics showed that the
majority of ofﬁcers were male and white. In terms of ofﬁcer experience, ofﬁcers had
served an average of approximately eight years. There was, however, a wide range of
experience levels as some ofﬁcers had served for less than a year while others had served
their department for over 34 years. Descriptive statistics also revealed that this measure
was positively skewed. As a result, ofﬁcers’ experience was transformed by taking the
natural log to address this issue. The newly transformed variable is used in all subsequent
analyses (Y =1 .71, s.d.=0.9l). Finally, most ofﬁcers had some level of higher education

with the modal ofﬁcer having a bachelor’s degree.

Bivariate Results

Bivariate correlations were estimated among the study variables primarily for two
purposes. First, the correlations were used to assess the relationships among ofﬁcer
attitudes and workgroup composition measures. Second, they were examined to provide
more insight into the compositional makeup of workgroups. A correlation matrix
denoting all bivariate relationships among the study variables is provided in Appendix E.

An examination of the relationships between workgroup compositional measures
and ofﬁcers’ attitudes reveal several interesting ﬁndings, particularly for the effect of
average experience and educational levels. Average experience levels were signiﬁcantly
associated with each of the six occupational attitudes; however, the strongest associations

Were for job satisfaction (r=-.26), aggressiveness (r=-.23), and direct supervisors (r=.21).

131

111

Ofﬁcers in workgroups with higher average experience levels were signiﬁcantly less
likely to be satisﬁed with their jobs, less likely to hold aggressive views toward patrol,
and more likely to hold negative attitudes toward their supervisors. With the exception of
aggressiveness, workgroup average educational levels were also signiﬁcantly associated
with each attitude. The strongest associations were found for job satisfaction (r=-.20),
citizen distrust (r=-. 18) and direct supervisors (r=.18). Ofﬁcers in workgroups with
higher average educational levels were signiﬁcantly less likely to be satisﬁed with their
jobs, less likely to be distrustful of citizens, and more likely to hold negative views
toward supervisors.

Workgroup gender and racial composition were not as widely or strongly related
to ofﬁcers’ attitudes. For example, ofﬁcers in workgroups with higher proportions of
white ofﬁcers were signiﬁcantly less likely to accept order maintenance roles, less likely
to be satisﬁed with their jobs, and less likely to accept the crime-ﬁghting role. However,
the magnitudes of these correlations were smaller than the workgroup experience and
educational variables. The proportion of male ofﬁcers assigned to workgroups was only
positively related to ofﬁcers’ views toward crime ﬁghting (1:25).

With respect to the relationships among the workgroup compositional measures,
there were also several signiﬁcant correlations. Workgroups with higher proportions of
male ofﬁcers also had signiﬁcantly higher proportions of white ofﬁcers (r=.07) but lower
average educational levels (r=-.10). Workgroups with higher proportions of white ofﬁcers
had signiﬁcantly higher average experience (r=.26) and educational levels (r=.36).
Finally, workgroups with higher average experience levels also had signiﬁcantly higher

average educational levels (r=.30)-

132

com

3111}.

Stat):
1111‘
con:
ilt’t’

C1115"

£0111

.l/i'i.‘

€51

 

Although the above results offer empirical support for the notion that workgroup
compositional characteristics are related to ofﬁcers’ occupational attitudes, additional
analysis is necessary. First, the bivariate results do not control for individual ofﬁcer
demographics or for the fact that ofﬁcers’ attitudes cluster by workgroup and department.
Second, workgroup composition measures were examined at the individual ofﬁcer level,
which can lead to problems associated with the ecological fallacy when drawing
conclusions (Raudenbush & Bryk, 2002). As such, multivariate, multilevel analysis is
needed to control for individual ofﬁcer and department variables, to address the
clustering of ofﬁcer attitudes, and to treat the compositional measures as workgroup level

contextual factors.

Multilevel Results

Tables 6.2 and 6.3 present the ﬁndings for the multivariate, multilevel models.
Table 6.2 includes the results for the four continuous attitudinal measures (i.e., citizen
distrust, order maintenance, direct supervisor, and job satisfaction) while Table 6.3
includes the results for the two categorical attitudinal measures (i.e., aggressiveness and
crime ﬁghting). Each of the six models controls for individual ofﬁcer characteristics and
the department where ofﬁcers worked. It should be noted that each of the models
included in Tables 6.2 and 6.3 represent ﬁnal models, meaning that some of the level 2
variables ended up being excluded to ensure stability of the remaining parameter

estimates.23

 

 

The study has 61 workgroups at level 2. Multilevel researchers have suggested estimating one variable
for every ten multilevel units (Maas & Hox, 2005). With the four workgroup composition variables and
four department controls, initial models included eight level 2 variables. To ensure that the level 2
Parameter estimates were stable, two of the workgroup composition variables were excluded from each of
the ﬁnal models. The two excluded variables were nonsigniﬁcant and contributed little explanatory power.
By excluding these variables, the ﬁnal models estimate six level 2 variables for the 61 workgroups.

133

Table 6.2: Multilevel Regression Results Predictin Ofﬁcer Attitudes — Final Models

 

 

 

 

 

Variable Citizen Order Direct Job Satisfaction
Distrust Maintenance Supervisor

b (s.e.) g b (s.e.) Q b (5e) [3 b (sci [3

Level II

Intercept 4.95*** ----- 13.78*** 3.69*** 9.10***
(0.05) (0.10) (0.06) (0.05)

Prop. Male 0.52 0.14 ------------------------------
(0.64)

Prop. White ---------- 1.93* 0.48 --------------------

(0.77)

Avg. -0.01 -0.17 ---------- 0.04* 0.36 -0.08*** -O.88

Experience (0.02) (0.02) (0.02)

Avg. ---------- 0.67* 0.83 028* 0.63 -0.23T -0.53

Education (0.27) (0.12) (0.12)

CSPD -0.11 -0. 14 0.31 0.17 -0.05 -0.05 0.31 0.33
(0.19) (0.41) (0.26) (0.27)

PPB -0.37 ** —0.51 -0.65 -0.40 -0.33 -0.37 0.06 0.07
(0.12) (0.52) (0.30) (0.27)

F WPD -0.08 -0. 10 -0.08 -0.04 -0.02 -0.02 0.22 0.25
(0.17) (0.36) (0.21) (0.22)

KPD 0.53*** 0.60 0.61 0.31 -0.26 -O.24 -0.12 -O.11
(0.13) (0.42) (0.22) (0.18)

LevelI

Male -0.06 -0.01 -0.42 -0.04 0.14 0.03 0.18 0.03
(0.15) (0.33) (0.14) (0.19)

White 0.05 0.01 -0.23 -0.03 -0.10 -0.03 0.03 0.01
(0.13) (0.31) (0.12) (0.14)

Experience -0.23** -O.13 -0.08 -0.02 0.30** 0.18 -0.31** -0.15

(In) (0.07) (0.14) (0.09) (0.10)

Education -0.04 -0.05 0.08 0.04 0.05 0.05 -O.12** -0.11
(0.03) (0.07) (0.03) (0.04)

Model

Results

N 953 954 957 946

1§C(1) 0.06 0.04 0.10 0.08

R LevelI 0.02 0.01 0.04 0.04

we] [is 0.91 0.98 0.56 0.94

 

 

 

 

 

 

 

***p<.001, **p<.01, *p<.05, Tp<.10; aLevel 11Rz values reﬂect the amount of variation explained by the

mtidels in relation to the total amount of variation that is present between workgroups (i.e., the ICC(l)
Va ues).

134

Table 6.3: Multilevel Ordinal Logistic Regression Results Predicting Ofﬁcer
Attitudes — Final Models

 

 

 

Variable Aggressiveness Crime ﬁghting
b (8e) Expibi b (s.e.) Expib)
Level II
Intercept 1 -3.81*** --- -3.60*** ---
(0.19) (0.20)
Intercept 2 -1.63*** --- -l.16*** ---
(0.09) (0.07)
Intercept 3 0.88*** --- 1.37*** ---
(0.09) (0.08)
Prop. Male --------------------
Prop. White ---------- -1.13* 0.32
(0.51)
Avg. Experience -0.07** 0 93 ----------
(0.03)
Avg. Education 0.06 1.06 -0.13 0.87
(0.16) (0.16)
CSPD 0.22 1.25 0.22 1.25
(0-36) (0.28)
PPB -0.01 0.99 -0.09 0.91
(0.43) (0.31)
FWPD -0.88** 0.41 -0.15 0.86
(0.27) (0.21)
KPD 0.36 1.43 0.81** 2.25
(0.24) (0.23)
Level 11
Male 038* 1.48 0.27 1.30
(0.18) (0.19)
White 0.53*** 1.69 0.14 1.15
(0.16) (0.16)
EXperience (In) -0.25* 0.78 -0.14* 0.87
(0.09) (0.07)
Education -0.03 0.97 ~0.07 0.93
(0.04) (0.04)
Model Results
N 961 962
ICC(1) 0.10 0.05
Pseudo R2 Level 1 0.03 0.01
Mo R2 Level IIa 0-89 0'99

 

 

 

 

2 . . .
f**P<.001, **p<.01, *p<.05; aLevel II R values reﬂect the amount of varratron explained by the models
In relation to the total amount of variation that is present between workgroups (i.e., the ICC(1) values).

135

Several interesting ﬁndings emerge when looking at the effects of workgroup

composition on ofﬁcers’ occupational attitudes.24 Again, it is important to note that since
the individual ofﬁcer characteristics are included at level 1 and are grand mean centered,
the coefﬁcients for the workgroup level 2 variables can be viewed as the contextual
effects of workgroup composition over and above individual ofﬁcer demographic effects
(Raudenbush & Bryk, 2002). Workgroup average experience levels had signiﬁcant
effects on three of the six attitudinal variables. In workgroups with higher average
experience levels among ofﬁcers, ofﬁcers were more likely to hold negative views
toward supervisors (B=0.36), less likely to be satisﬁed with their jobs (B=-0.88), and less
likely to hold aggressive attitudes toward patrol (exp(b)=0.93).

Workgroup racial composition also had a signiﬁcant effect on ofﬁcers’ role
orientations. Speciﬁcally, workgroups containing higher proportions of white ofﬁcers
were signiﬁcantly less likely to accept order maintenance roles ([3=0.48), yet also
signiﬁcantly less likely to favor crime-ﬁghting orientations (exp(b)=0.32). Finally,
workgroup average educational level was related to three of the six attitudinal measures.
For example, in workgroups containing ofﬁcers with higher average educational levels,
ofﬁcers were less likely to accept order maintenance roles (0:083), more likely to hold
negative views toward supervisors (0:063), and less likely to be satisﬁed with their jobs
(B=-0.53; p<.10). Overall, these three compositional characteristics were consistently
related to ofﬁcers’ attitudes and exerted some of the strongest effects across the six

models.

 

4 Diagnostic tests for multicollinearity were conducted for all models presented in Tables 6.2 and 6.3 for
each level. At level 1, tolerance levels and variance inﬂation factors (VIFs) were examined and all VlFs
were 1.02 or below. At level 2, this issue was examined using the workgroup mean for each attitude as the
dependent variable. VlF values for the four compositional measures were all less than 1.24. Thus,
multicollinearity was not deemed to be a problem.

136

In terms of the model results for the level 2 portion of the models, the four
composition measures explained between 56 percent (i.e., direct supervisors) to almost 99
percent (i.e., job satisfaction) of the attitudinal variation between workgroups. This is in
part due to the contribution of the compositional characteristics to the model; however,
the high percentages of explained variation are also due to the inclusion of the department
dummy variables as level 2 controls. It should be noted that the results of the
unconditional models presented in the previous chapter indicated that the total amount of
attitudinal variation between workgroups ranged from 4 percent (i.e., ICC(1) for order
maintenance) to 10 percent (ICC(1) for direct supervisor and aggressiveness). Thus, it is
important to keep this in mind when examining the amount level 2 variation explained by
the models in Tables 6.2 and 6.3.

At the individual ofﬁcer level, there were also several signiﬁcant effects on
ofﬁcer attitudes. Individual ofﬁcer experience levels were signiﬁcantly related to ﬁve
attitudes. Speciﬁcally, ofﬁcers with more years of service to the department were less
likely to be distrustful of citizens, more likely to hold negative views toward supervisors,
less likely to be satisﬁed with their jobs, less likely to hold aggressive attitudes toward
patrol, and less likely to believe crime-ﬁghting is the most important role of ofﬁcers.
Ofﬁcers with higher education levels were also less likely to be satisﬁed with their jobs.
Male ofﬁcers were more likely to favor aggressive patrol tactics. Finally, white ofﬁcers
were more likely to hold aggressive attitudes toward patrol. Even though there were a
handful of signiﬁcant effects at the individual ofﬁcer level, examination of the model

results indicate that these variables did not explain much variation in ofﬁcer attitudes.

137

Collectively, these four individual ofﬁcer variables only explained one percent (i.e., order

maintenance) to four percent (i.e., job satisfaction) of the variance in ofﬁcers’ attitudes.

Discussion

This chapter examined the effect that workgroup compositional characteristics
had on ofﬁcers’ occupational attitudes. Speciﬁcally, analyses tested the extent to which
workgroup gender composition, racial composition, average experience levels, and
average educational levels explained ofﬁcers’ views toward citizens, order maintenance
roles, direct supervisors, job satisfaction, aggressive patrol, and crime ﬁghting. The
results of these tests indicated that the composition of ofﬁcers’ workgroups did have an
impact on their attitudes. Of particular importance were the racial composition, average
experience levels, and educational levels of workgroups as these three measures were
consistently and strongly related to ofﬁcer attitudes even after controlling for individual
ofﬁcer characteristics and while controlling for their respective departments.

The results show that the types of ofﬁcers’ that individual ofﬁcers are exposed to
and work with within the workgroup context matter in inﬂuencing attitudes regardless of
ofﬁcers’ own individual characteristics. For example, when looking at ofﬁcers’
acceptance of order maintenance roles, individual ofﬁcer race did not have a signiﬁcant
effect. Yet ofﬁcers who worked in groups with higher proportions of white ofﬁcers were
signiﬁcantly less likely to accept order maintenance roles. Thus, being exposed to and
routinely interacting with ofﬁcers in this type of context inﬂuenced ofﬁcers’ acceptance
of role orientations over and above individual level inﬂuences. Additionally, workgroup
experience levels had both individual and aggregate effects on satisfaction levels. At the

individual level more experienced ofﬁcers were less likely to be satisﬁed with their jobs.

138

However, ofﬁcers exposed to more experienced ofﬁcers within their workgroups were
also less satisﬁed with their jobs even after controlling for their own experience levels.
Finally, average education levels consistently inﬂuenced ofﬁcer attitudes regardless of
ofﬁcers’ own educational level. For example, in workgroups where ofﬁcers were exposed
to other ofﬁcers’ with higher educational levels, they were less likely to accept order
maintenance roles.

One aspect that is important to consider is whether the compositional results
reﬂect socialization effects or selection effects. In other words, are ofﬁcer attitudes
inﬂuenced by their interactions and exposure to their workgroup peers or are ofﬁcers who
are familiar with the practices of a particular workgroup more likely to select into that
environment? For example, an ofﬁcer’s view towards aggressive patrol may be
inﬂuenced as a result of working with other ofﬁcers who have speciﬁc orientations
toward aggressiveness (i.e., a socialization effect). On the other hand, aggressive ofﬁcers
may be attracted to speciﬁc workgroups that they know adhere to aggressive practices
(i.e., a selection effect). Across the study departments, workgroup assignments were
decided by both the administration and individual ofﬁcers through the use of bid systems.
As such, this question cannot be adequately examined due to the cross-sectional nature of
the data. Future research, however, using longitudinal designs could investigate the
nature of these effects. It would be beneﬁcial to know how socialization and/or selection
processes inﬂuence attitudes. In addition, uncovering the processes by which information
regarding the nature of these attitudes is transmitted across ofﬁcers would offer insight as

to how attitudes might become shared by ofﬁcers.

139

Regardless of the nature of the compositional effects, the results do have practical
implications for police administrators. For example, if departments are trying to
implement strategies and practices that require acceptance of broader police role
orientations, then how workgroups are comprised may have an effect on their successful
implementation. It has already been argued that strategies and practices implemented at
the department level are administered and carried out at the street level within the
workgroup context. In order for these to be successful, administrators need to consider
the types of ofﬁcers assigned to workgroups. For example, the current results found that
ofﬁcers in workgroups with higher proportions of white ofﬁcers were signiﬁcantly less
likely to favor broader role orientations even after controlling for ofﬁcer’s individual
race. If departments seek to implement strategies that require acceptance of broader roles,
such as community policing approaches, then administrators would beneﬁt by assessing
the racial composition of workgroups. Perhaps creating greater racial diversity within the
workgroup context might create a work environment that would accept such strategies. In
essence, having an understanding of how the demographic makeup of workgroups can
inﬂuence ofﬁcers’ attitudes may prove beneﬁcial in establishing settings where the ideals
of the department are viewed similarly and carried out at lower levels of the organization.

Although the ﬁndings illustrate the importance of workgroup composition on
ofﬁcer attitudes, the study is not without limitations. First, the focus was solely on the
effect of workgroup composition on ofﬁcer attitudes. Within the workgroup context,
however, there are likely other factors that also inﬂuence attitudes. Two other types of
factors that were identiﬁed by the theoretical framework of the study include ofﬁcers’

supervisors as well as characteristics of their physical environment (e. g., crime rates,

140

workload, or neighborhood structural factors). Incorporation of these other factors could
attenuate some of the results for the workgroup compositional characteristics. Thus, it is
important to consider the current ﬁndings with this limitation in mind. Future research
would beneﬁt by including these additional factors in empirical models.

Second, the individual ofﬁcer characteristics used as controls had little
explanatory power across the six models. The fact that ofﬁcer demographics accounted
for only small amounts of variation in the level 1 models could lead to bias in the
workgroup composition estimates. The inclusion of other individual level variables that
might also explain ofﬁcer attitudes could also attenuate some of the workgroup level
results. Although important to bear in mind, it is unclear what other individual ofﬁcer
characteristics might inﬂuence attitudes as prior research has focused heavily on ofﬁcer
demographics and these results have also shown little explanatory power. Although
ofﬁcer age was not included in the current study, this was due to the fact that age and
experience levels were highly correlated (r=.70) and would likely pose collinearity
problems. Future research would also beneﬁt by identifying and including other
individual level characteristics.

Finally, limitations existed for some of the attitudinal and workgroup measures.
For example, there was little variation across workgroups with respect to gender
composition. The study departments were still heavily comprised of male ofﬁcers. The
lack of diversity could be why gender composition was largely unrelated to ofﬁcers’
attitudes with the exception of views toward order maintenance. Despite these
limitations, the results do offer important insight into how ofﬁcers’ workgroups inﬂuence

their attitudes.

141

CHAPTER SEVEN: WORKGROUP ATTITUDES & FORCE BEHAVIOR

This chapter presents the results testing whether workgroup level attitudinal
measures predict ofﬁcers’ use of force behavior. Speciﬁcally, the results from Chapter
Five revealed that four attitudinal measures displayed valid and reliable group-level
properties: citizen distrust, direct supervisor, job satisfaction, and aggressiveness. As
such, these attitudes can be viewed and treated as contextual measures of ofﬁcers’ work
environments (e. g., workgroup climates) that may explain street-level behaviors, such as
use of force. Here, these measures are examined in relation to three types of force used in
encounters with citizens: soft hand force, hard hand force, and impact force. The chapter
begins by discussing the data used for analysis. Next, the force, workgroup, and control
variables used in the empirical models are described. The analytic techniques used to test
the models are then explained. Finally, the results of the analysis are presented and

discussed.

Data

Data for this portion of the study come from two sources. First, the attitudinal
measures and individual ofﬁcer characteristics are taken from the ofﬁcer survey and
master rosters. These data are identical to the data described and used to test the previous
two research objectives. Second, the force data came from the use of force reports
collected as part of the larger NIJ project. These force reports included data on the level
of force used by ofﬁcers during encounters, the level of resistance suspects presented to
ofﬁcers, and suspect demographic characteristics. From these data, measures of force
behavior (the outcomes of interest) as well as the suspect characteristics used as controls

were derived.

1. 42

Across each site, at least two years of force reports were collected. Department
policy required ofﬁcers to ﬁll out these reports for every encounter in which a reportable
use of force incident occurred. Reporting requirements were similar across all sites.25 The
data collection periods reﬂect a time frame in which each department had a consistent use
of force policy and force reporting method. The exact time frames for each of the ﬁve
study departments were: APD (April, 13 2006-April 12, 2008), PPB (November 5, 2005-
November 4, 2007), CSPD (2006-2007 calendar years), F WPD (December 18, 2004-
December 31, 2007), and KPD (June 1, 2005-May 31, 2007). For purposes of the current
study, however, only force data from the calendar year 2007 was used for analysis as this
coincided with the time frame in which the ofﬁcer survey was administered.

Based on the time frames, two issues had to be addressed. First, the time frame for
KPD only included force data for part of 2007. Because only ﬁve months of force data
was collected, KPD was excluded from this part of the study. Second, only eleven
months of 2007 data could be collected for PPB. This was due to the fact that PPB’s force
reporting requirements changed after this date. To remedy this, force data for PPB was
taken back to November, 2006. Although not ideal, this allowed for the use of twelve
consistent months of force data for the four remaining sites (i.e., APD, PPB, CSPD, and

FWPD).

Variables
Use of Force Measures
The outcomes of interest are the type(s) of force ofﬁcers used during encounters

with citizens. Across each of the four sites, common measures of force usage were

 

25 . . .
Generally, ofﬁcers were requrred to ﬁll out reports for every actron above handcufﬁng or srmple
restraints across each site.

143

constructed from each department’s force reports to ensure that force types were
comparable across departments. Force behavior was coded into two categories:
weaponless tactics and weapon tactics. For weaponless tactics, the following force types
were coded from each department’s reports: simple restraints (e.g., a pat down or escort
maneuver), pain compliance techniques (e.g., pressure point holds), control maneuvers,
takedowns, and strikes. For weapon tactics, the following force types were coded:
chemical spray, baton, EMD (i.e., taser), less lethal weapons, and ﬁrearms.26 Based on
these categorizations, three force measures were constructed to reﬂect the types of force
used by ofﬁcers during encounters. Two of these measures were derived from the
weaponless tactics categorizations and are deﬁned as soft hand force and hard hand
force, while the third measure was derived from the weapon tactic categorizations and is
deﬁned as impact force.

Soft hand force is a binary variable (0=no, l=yes) that captures whether ofﬁcers
reported using any of the following weaponless tactics during a citizen encounter: simple
restraints, pain compliance techniques, control maneuvers, and/or take downs. In essence,
this measure captures lower levels of hands on force that involves guiding or controlling
a suspect. Hard hand force is a binary variable (0=no, l=yes) that captures whether
ofﬁcers reported using striking techniques (e.g. strikes with hands, body, or feet) during a
citizen encounter. This variable captures more aggressive physical force used by ofﬁcers
to gain control of suspects. Impact force is also a binary variable (0=no, l=yes) that

measures whether ofﬁcers reported using any type of weapon tactic during a citizen

 

26 . . . -
For a more detailed descrrptron as to how the common force measures were derrved and coded from
each department’s force reports, see Terrill, Paoline, & Ingram, forthcoming.

144

encounter.27 These three force measures are consistent with the current conceptualization
and operationalization of force usage for the larger NIJ project (Terrill & Paoline, 2009).

In addition to constructing the force measures, each case had to be coded into the
appropriate workgroup. The force data was coded at the ofﬁcer level, meaning that each
ofﬁcer who reported using force during an encounter was a unique observation. As a
result, the force data and survey data could be linked at the ofﬁcer level based on the
unique identiﬁcation numbers assigned to each ofﬁcer for the study. This allowed for the
workgroup level measures and survey data to be merged with the use of force data.
Workgroup Attitudinal Measures

The primary independent variables of interest are the four attitudes found in
Chapter Five to exhibit collective properties that allow for their aggregation to the
workgroup level. Speciﬁcally, the measures of citizen distrust, direct supervisor, job
satisfaction, and aggressiveness were found to have sufﬁcient levels of within-group
agreement, acceptable group mean reliabilities, and signiﬁcant between-group variability.
As a result, these four measures can be treated as contextual measures of ofﬁcers’
workgroup environments. Furthermore, they can be measured by aggregating individual
ofﬁcer responses to the workgroup level by taking the group level means for each attitude
(Bliese, 2000; James, 1982; Kozlowski & Klein, 2000).

For example, the ﬁndings from Chapter Five indicated that ofﬁcers within
workgroups shared similar perceptions of citizens (i.e., within-group agreement).

However, the nature of these shared perceptions varied across workgroups (i.e., between-

 

27 . . .

It should be noted that in order for the observation to be coded as 1mpact force, the ofﬁcer had to have
used the weapon in some capacity during the encounter; mere drawing or pointing of the weapon was not
coded as an instance of impact force.

145

group variability). Ofﬁcers within some workgroups had more distrustful views toward
citizens than ofﬁcers within other workgroups. When viewed from the context of the
workgroup climate literature discussed in Chapter Three, this suggests that a climate of
distrust exists within the formal organization of policing at the workgroup level and that
these climates signiﬁcantly vary across workgroups. Therefore, a workgroup level
variable was constructed by taking the group level mean of the citizen distrust measure.
This measure is deﬁned and referred to as distrust climate and higher values indicate a
greater climate of distrust towards citizens. It is hypothesized that ofﬁcers in workgroups
with distrustful climates will have a greater likelihood of using soft hand, hard hand, and
impact force tactics against suspects than ofﬁcers in workgroups with less distrustful
climates.

A similar approach can be taken for the other three attitudes. The results from
Chapter Five found that ofﬁcers within workgroups shared similar attitudes toward their
direct supervisors in terms of the degree to which supervisors’ management approaches
discouraged extra effort from ofﬁcers and the degree to which supervisors looked out for
the personal welfare of ofﬁcers. The nature of these shared attitudes also varied across
workgroups. This also suggests that a climate of support exists within the formal
organization of policing at the workgroup level and that these climates signiﬁcantly vary
across workgroups. A workgroup level variable was constructed and deﬁned as

unsupportive supervisor to measure this aspect of the workgroup environment. Here,

146

higher values indicate less supportive climates and these climates are hypothesized to

predict ofﬁcers’ use of soft hand, hard hand, and impact force.28

Ofﬁcers within workgroups were also found to share similar levels of job
satisfaction. The nature of these shared attitudes also varied across workgroups: ofﬁcers
in some workgroups had greater levels of satisfaction, while ofﬁcers in other workgroups
shared lower levels of satisfaction. While the previous two measures were described as
workgroup climates, the group-level measure of job satisfaction is perhaps better viewed
as an indicator of group morale. For example, Bliese, Chan, and Ployhart (2007: 553)
have noted that individual ratings of job satisfaction may be inﬂuenced by the shared
experiences of group members, and so “the group mean reﬂects shared group properties
such as. . .group morale.” As a result, the group level measure of job satisfaction is
viewed similarly in the context of the current study and the variable morale was
computed based on the group means of the job satisfaction measure. Higher values reﬂect
greater morale and it is hypothesized that morale will predict soft hand, hard hand, and
impact force because it serves as a way to cope with the stressors associated with that
type of environment.

Finally, aggressive climate is derived from individual ofﬁcers’ responses to the
aggressiveness measure (i.e., a good patrol ofﬁcer is one who patrols aggressively,

stopping cars, checking out people, running license checks and so forth). Similar to the

 

28 Unlike the hypotheses given for the effect of distrust climate on force behavior, directional hypotheses
are not offered for the effect of supervisor support because the results could either be positively or
negatively related to the three force measures. For example, ofﬁcers in less supportive climates may be less
likely to use any of the three force tactics because of their belief that supervisors do not look out for their
well being. Ofﬁcers in these climates might try to refrain from using force during encounters. On the other
hand, ofﬁcers in less supportive climates could also be more likely to rely on impact force tactics, such as
OC spray or the taser because these tactics are less likely to cause injury to the ofﬁcer. Here, reliance on
impact force tactics may be used to make up for the perceived lack of support from supervisors who do not
look out for the personal welfare of ofﬁcers or discourage extra effort from ofﬁcers.

147

distrust and unsupportive supervisor climate measures, workgroup means are used as
measures of aggressive climates for each workgroup. This measure represents a
contextual variable of ofﬁcers’ workgroup environment and it is hypothesized that
ofﬁcers who work in aggressive climates will have a greater likelihood of using soft
hand, hard hand, and impact force than ofﬁcers working in less aggressive climates.
Control Variables

Several individual ofﬁcer and suspect control variables that prior research has
found to be related to use of force behavior are also included in the analysis. Eight
variables are used to control for individual ofﬁcer characteristics. Four of these consist of
the ofﬁcer demographic characteristics used in the second research objective and include
ofﬁcer gender, race, education, and length of service. The other four variables consist of
individual ofﬁcer attitudes of citizen distrust, direct supervisor, job satisfaction, and
aggressiveness. Including these individual ofﬁcer attitudes in the models strengthens
empirical support for any contextual effects of the workgroup attitudinal measures. If the
workgroup attitudinal measures explain variation in force behaviors over and above the
individual level measures, then stronger evidence of contextual effects are present (see
Bursik & Grasmick, 1996; Bliese, et al., 2007; Hauser, 1970; Sampson, Raudenbush, &
Earls, 1997).

In terms of suspect characteristics, the following variables were coded from each
sites’ use of force reports and are included in the analyses: suspect demographics (i.e.,
race, gender, and age), whether the suspect was under the inﬂuence of drugs and/or
alcohol, whether the suspect was mentally or emotionally disturbed, whether the suspect

possessed a weapon, whether the suspect was arrested, and the highest level of resistance

148

presented to the ofﬁcer during the encounter. The operational deﬁnitions and
hypothesized effects for each of these variables as well as the force and workgroup
measures are outlined in Table 7.1.

In addition to controlling for ofﬁcer and suspect characteristics, the study also
controls for the department where each force encounter occurred. This is necessary
because departments selected for the study had different force policies, meaning that each
department had different guidelines for when and how various forms of force could be
used during encounters. It is necessary to account for these factors as it likely produces
differences in the application of force across sites. In subsequent analyses, the department
is controlled for in two manners. First, when examining force usage, the clustering of
force incidents by site is accounted for statistically using the same analytic procedure
used to account for the nesting of workgroups within higher organizational units in
Chapters Five. This procedure is used here by adjusting the standardterrors of parameter
estimates to account for the clustering of force usage within departments. Second, a series
of dummy variables are used as controls for the multilevel models explaining force usage.
Three dummy variables were constructed to measure where the force incident occurred

(CSPD, PPB, or APD) and F WPD serves as the reference site.

149

Table 7.1: Descriptions of Key Force, Workgroup, Officer, & Suspect Variables

 

 

Variable Hypothesized Deﬁnition
Effect
Dependent
Soft Hands Did ofﬁcer use soft hand techniques? (0=No,
1=Yes) Firm grip, pressure point, control
maneuvers, takedowns
Hard Hands Did ofﬁcer use hard hand techniques (0=No,
1=Yes) Striking w/ hands, feet, body
Impact Did ofﬁcer use impact force? (0=No, 1=Yes)
OC spray, Taser, baton, impact munitions,
other weapon
Workgroup
Distrust Climate + Group-level mean of citizen distrust attitude
Unsupportive Supervisor +/- Group-level mean of direct supervisor attitude
Morale +/- Group-level mean of job satisfaction attitude
Aggressive Climate + Group-level mean of aggressiveness attitude
Officer
Citizen Distrust + Distrust of citizens (higher values=greater
distrust)
Direct Supervisor +/- Views toward supervisor (higher
values=negative views)
Job Satisfaction - Job satisfaction (higher values=greater
satisfaction)
Aggressiveness + Aggressive patrol 1(SD) to 4 (SA)
Male + l=male, 0=female
Non-white +/- 1=non-white, 0=white
Experience - Years of experience
Education - Highest level of education ranging from 1 (less
than HS) to 8 (Graduate Degree)
Suspect
Resistance + Highest level of resistance: 1(failure to
comply), 2(defensive physical), 3(defensive
aggressive), or 4 (deadly)
Male + l=male, 0=female
Non-white +/- l=non-white, 0=white
Age - Age in years
Alcohol/Drugs + l=suspect under inﬂuence of alcohol/drugs,
0=no
Mental/Emotion + l=suspect mentally or emotionally disturbed,
0=no
Weapon + l=suspect had weapon, 0=no
Arrest + l=suspect arrested, 0=no

 

150

Analysis

For the ﬁnal research objective, observations occur at two levels of analysis:
ofﬁcers’ attitudes and use of force behavior are nested within workgroups. As such, the
data are hierarchically structured and the research question of interest involves variables
at different levels of analysis. When such factors are present, multilevel modeling
techniques are appropriate analytic strategies for answering the above research objective
(Hoffman etal., 2000; Raudenbush & Bryk, 2002). The multilevel analyses are
conducted in two steps.

After descriptive statistics are reported and discussed, unconditional models are
presented for each of the three force variables. These multilevel models assess the
amount of variation in the force variables between workgroups while accounting for the
clustering of force across sites and are similar to the one-way, random effect ANOVA
analyses conducted for each attitudinal measure in Chapter Five. No other independent
variables are included in the model; the purpose is to determine if sufﬁcient variation
exists on the force measures between workgroups while accounting for departmental
differences in force tactics. There must be empirical evidence that use of force behavior
varies across workgroups. If force behavior does not signiﬁcantly differ across
workgroups, then multilevel analyses are unnecessary. Evidence of signiﬁcant variation
on the force measures exists if the between-group variance components estimated by the
models are signiﬁcant even after accounting for clustering by department. Furthermore,
similar to the ANOVA models, an ICC(1) value can be calculated to determine the
amount of variation in force behavior that can be explained at the workgroup level

(Hoffman et al., 2000; Raudenbush & Bryk, 2002).

151

Second, two-level, random intercept models are conducted to test whether
workgroup-level measures explain force behavior while controlling for individual and
suspect characteristics. If the results of the unconditional models indicate sufﬁcient
variation in soft hand, hard hand, and impact force between workgroups, then a two-level
multilevel model can be speciﬁed to explain this variation for each force measure.
Speciﬁcally, intercepts as outcome models are appropriate for this type of research
question (Hoffman et al., 2000; Raudenbush & Bryk, 2002). The workgroup measures are
used in the level 2 portion of the model, and all individual ofﬁcer and suspect variables
are used as level 1 control variables. Additionally, the models will control for the
department at the workgroup level using the site dummy variables. In essence, this model
tests whether workgroup perceptual measures explain force behavior while controlling
for individual, suspect, and department variables.

Finally, it should be noted that the multilevel models represent hierarchical
generalized linear models (HGLM) because each of the force outcomes are binary
variables. Each multilevel model is a binary logistic regression model estimated using
robust maximum likelihood estimation (MLR). Intercept and slope estimates represent
logit coefﬁcients from which odds ratios and predicted probabilities can be calculated
(Muthen & Muthen, 2007; Raudenbush & Bryk, 2002). Since the primary interest is the
effect of level 2 variables (i.e., workgroup measures) on force behavior, the level 1
ofﬁcer and suspect control variables remain ﬁxed across models. Furthermore, all level 1
variables are grand mean centered.

As with the multilevel analyses conducted in the previous chapter, grand mean

centering is necessary to estimate level 2 effects while controlling for level 1 variables.

152

Therefore, the level 2 logit coefﬁcients reﬂect the relationship between the workgroup
measures and force behavior after adjusting for the effects of the level 1 control variables.
Furthermore, grand mean centering allows for the assessment of any contextual effects
among the workgroup attitudinal measures at level 2. For example, if distrust climate is
signiﬁcantly associated with force behavior while controlling for the individual ofﬁcer
measure of citizen distrust, then empirical support exists that ofﬁcers’ social
environments predict behavior over and above individual ofﬁcers’ own views toward
citizens. Again, the effect of the aggregate measure is greater than the effect of the
individual responses that make it up. Grand mean centering allows for the direct
determination of any potential contextual effects (Enders & Toﬁghi, 2007; Hoffman,
1998; Raudenbush & Bryk, 2002).
Results
Descriptive Statistics

Table 7.2 presents descriptive statistics for each variable used in subsequent
analyses. Beginning with the force measures, there were 1,648 instances where some type
of force was used by an ofﬁcer across the four sites in 2007. The use of soft hand force
tactics by ofﬁcers was present in 52 percent of those 1,648 instances, while the use of
hard hand force tactics by ofﬁcers was present in 23 percent of the 1,648 instances. With
respect to impact force, ofﬁcers used some form of weapon tactic in 42 percent of the
1,648 instances where force was used. Two aspects should be noted about these results.

First, the application of force appears to be much higher than is reported by
previous research, particularly research that has used observational data to examine force

behavior (e.g., see Chapter Two). This is due to the fact that force behavior in the current

153

study comes from ofﬁcial records and therefore only includes cases where reportable
force was actually used. Thus, unlike observational studies that allow for the use of a
wider conceptualization of force behavior (i.e., no force, verbal force, physical force,
impact force), the current study can only examine instances where reportable force was
used during an encounter. Second, the 1,648 cases used for analysis represent instances
where ofﬁcers used some type of reportable force against a suspect. This means that in
cases where multiple ofﬁcers used force or a single ofﬁcer used force against multiple
suspects, each separate application of force was coded as a line in the data. On the other
hand ofﬁcers may have also used multiple force types against the same suspect (e.g.,
takedown, strike, and OC spray). Each force application against the same suspect was
coded on the same line of data. Therefore, the 1,648 force cases do not necessarily reﬂect
unique encounters with citizens. This is the reason the percentages associated with the
means for the force measures are greater than 100 percent.

The use of force cases were nested within the ﬁfty-three workgroups across the
four sites. On average there were just over thirty-one force instances within each
workgroup which suggest that there are a sufﬁcient number of workgroup units as well as
observations within units to provide for accurate tests of the proposed multilevel models
(Maas & Hox, 2005). With respect to the four workgroup attitudinal measures,

workgroup means and standard deviations indicated that the measures were normally
. . . . . . 29
drstrrbuted and that there should be sufﬁc1entvar1atron 1n measures across workgroups.

On average, workgroup climates were not overly distrustful ('2? = 4.86), were generally

 

9 There may be some concern that the validity and reliability properties of the workgroup attitudinal
measures changed as a result of having to exclude the eight KPD workgroups from analysis. To assess this,
the analyses from Chapter Five were conducted without these eight workgroups. The results did not show
any signiﬁcant deviations from the ﬁndings discussed in Chapter Five in terms of within-group agreement,
group-mean reliabilities, and between-group variability.

154

supportive (E = 3.75), and aggressive (E = 3.06). Workgroups also displayed rather high
levels of morale (if = 9.06). Finally, ten workgroups were from CSPD (19%), thirteen
workgroups were from PPB (24%), eighteen workgroups were from APD (34%), and
twelve workgroups were from FWPD (23%).

Table 7.2: Descriptive Statistics for Force, Workgroup, Ofﬁcer, & Suspect Variables

 

 

Variable N Min. Max. Mean Std.
Deviation
Dependent
Soft Hand Force 1,648 0 l 0.52 0.50
Hard Hand Force 1,648 0 1 0.23 0.42
Impact Force 1,648 0 1 0.42 0.49
Workgroup
Distrust Climate 53 3.71 6.27 4.86 0.50
Unsupportive Supervisor 53 2.72 5.00 3.75 0.56
Morale 53 7.00 10.63 9.06 0.69
Aggressiveness Climate 53 2.36 3.67 3.06 0.30
CSPD 53 0 l 0.19 0.33
PPB 53 0 l 0.24 0.43
APD 53 0 l 0.34 0.48
FWPD (reference site) 53 0 l 0.23 0.42
Officer
Male 1,640 0 l 0.94 0.23
Non-white 1,637 0 l 0.15 0.36
Experience 1,643 0.50 30 5.22 4.42
Education 1,628 1 8 5.18 1.55
Citizen Distrust 1,632 2 8 4.90 1.44
Direct Supervisor 1,620 2 8 3 .45 1.39
Job Satisfaction 1,616 3 12 9.35 1.68
Aggressiveness 1,637 1 4 3.24 0.70
Suspect
Resistance 1,582 1 4 2.31 0.67
Male 1,635 0 l 0.87 0.33
Non-white 1,622 0 l 0.53 0.50
Age 1,620 11 84 31.54 11.16
Alcohol/Drugs 1,646 0 l 0.55 0.50
Mental/Emotion 1,646 0 1 0. 13 0.33
Weapon 1,643 0 l 0.12 0.33
Arrest 1,645 0 1 0.84 0.37

 

155

With respect to ofﬁcer variables, the majority of ofﬁcers who used force were
white, male, had approximately ﬁve years of experience, and had some college education.
Individual ofﬁcer attitudes toward citizens, supervisors, job satisfaction, and aggressive
patrol were similar to what was reported in Chapter Five. Ofﬁcer experience was found
to be positively skewed. This variable was transformed by taking the natural log and the
transformed variable is used in all subsequent analyses G = 1.33, s.d.=0.82). With respect
to suspect characteristics, the majority of suspects who had force used against them were
minority, male, and had been impaired by alcohol or drugs at the time of the encounter.
The modal level of resistance involved defensive physical resistance. Few cases involved
mentally or emotionally unstable suspects or suspects with weapons, while approximately

eighty-four percent of cases involved an arrest.

Soft Hand Force Results

Discussion of the multilevel results begins with the models for soft hand force.
The results for the unconditional model are used to determine the extent to which
ofﬁcers’ use of soft hand force varied across workgroups. Initial results indicated the
presence of signiﬁcant variation based upon the estimated between-groups variance
component (I = 0.39, s.e. = 0.14, p<.01). Additionally, the approximate amount of
variation between workgroups was calculated to be eleven percent (ICC(1)=0.11).30 This
initial model, however, did not account for the department in which the force incidents
occurred. After controlling for department, the between-group variance component

became non-signiﬁcant (t = 0.39, s.e. =0.24, p>.05).

 

30 . . . . .
The ICC(1) was calculated usmg the same equation employed In Chapter Five for the categorrcal

attitudinal variables. ICC(1) = .39/(.39+ 738) = 0.1 l.

156

Even though a portion of the variation in soft hand force was found to exist
between workgroups (i.e., 11%), this variation is likely to be accounted for by
departmental factors rather than workgroup factors. Although not empirically tested,
differences in ofﬁcers’ use of soft hand tactics may be attributed to the types of less-
lethal force policies implemented across departments. Ofﬁcers within the same
department operate under the same policies which guide their decisions on when and how
to use lower levels of force such as grips, holds, control maneuvers, and takedowns. As a
result, ofﬁcers working in the same department are likely to behave similarly with respect
to this type of force usage. For the current inquiry, the results show that workgroup level
measures are not important for explaining multilevel differences in soft hand force
behavior. As such, multilevel modeling at the workgroup level is not warranted. Since the
primary research question focuses on the extent to which workgroup attitudinal measures
predict soft hand force behavior, no further analyses are conducted. Attention is now

tumed to the two remaining force measures.

Hard Hand Force Results

The results of the unconditional model for hard hand force also initially indicated
the presence of signiﬁcant variation based upon the between-groups variance component
(I = 0.62, s.e. =0.15, p<.001). The approximate amount of variation in hard hand force
between workgroups was calculated to be sixteen percent (ICC(1)=0.16). Furthermore,
signiﬁcant between-group variation remained after accounting for the clustering of hard
hand force by department (I =0.62, s.e. = 0.27, p<.05). These results show that
workgroup-level factors can be used to explain multilevel differences in hard hand force

behavior. In other words, the application of hard hand force techniques by ofﬁcers

157

differed across workgroups even after taking into account similarities in hard hand force
usage within departments. As such, further tests can be conducted to explain this
variation.

Table 7.3 presents the results for two multilevel models. First, an initial model
was tested to examine the impact of the four workgroup climate measures on hard hand
force while controlling for ofﬁcer and suspect characteristics. The purpose of this model
is to gain a preliminary understanding of the effect of the workgroup measures. The
results indicate that three of the four measures signiﬁcantly predicted hard hand force.
Hard hand force usage was signiﬁcantly higher in workgroups with distrustful and
unsupportive climates but signiﬁcantly lower in aggressive climates. Workgroup morale
was not associated with hard hand force. A second model was conducted which included
the three department dummy variables as level 2 controls. Additionally, workgroup
morale was dropped from this model since it contributed little explanatory power to the
ﬁrst model.3 ' After controlling for ofﬁcer and suspect variables as well as the department
in which the encounter occurred, distrust climate, unsupportive supervisor, and
aggressive climate were still found to be signiﬁcantly associated with the hard hand force
measure. In addition, when compared to workgroups from F WPD, ofﬁcers in workgroups
from PPB and APD were signiﬁcantly less likely to use hard hand force techniques.
Collectively, these measures explained approximately 75 percent of the level 2 variation

(i.e., ICC(1)=0.16) in hard hand force across workgroups.

 

31 Multilevel researchers have stressed the importance of sample size when considering the number of
variables to include at level 2 to ensure statistical power. An accepted rule of thumb is to estimate one
variable for every ten level two units (Maas & Hox, 2005). Thus, morale was excluded to keep the model
parsimonious.

158

Table 7.3: Multilevel Model Results for Hard Hand Force (N=1,457)

 

 

Variable Initial Model Full Model
b (SE) Exp(b) b (SE) Expib)

Workgroup Level
Intercept 1.60 (0.12)*** 4.95 0.80 (0.20)*** 2.22
Force Variance Comp. 0.32 (0.11)** ---- 0.15 (0.07)* ----
Distrust Climate 0.62 (0.29)* 1.86 0.69 (0.28)* 1.99
Unsupportive Supervisor -0.49 (0.25)* 0.61 -0.68 (0.23)** 0.51
Morale 0.01 (0.21) 1.01 -------- ---
Aggressive Climate -1.64 (0.40)*** 0.19 -0.70 (0.33)* 0.49
CSPD ------- —0.46 (0.46) 0.63
PPB ------- -0.74 (0.27)** 0.47
APD ------- -1.49 (0.28)*“ 0.23

Control Variables

Officer
Male 0.92 (0.42)* 2.52 0.91 (0.41)* 2.49
Non-white -0.21 (0.23) 0.81 -0.11 (0.23) 0.89
Experience (In) 0.09 (0.14) 1.10 0.07 (0.14) 1.07
Education -0.03 (0.03) 0.97 -0.04 (0.03) 0.96
Citizen Distrust -0.02 (0.07) 0.98 -0.04 (0.07) 0.96
Direct Supervisor -0.08 (0.06) 0.93 -0.08 (0.06) 0.93
Job Satisfaction 0.02 (0.01) 1.02 0.01 (0.06) 1.01
Aggressiveness -0.02 (0.09) 0.98 —0.01 (0.09) 0.99

Suspect
Resistance 0.45 (0.14)** 1.57 0.47 (0.14)** 1.60
Male 1.84 (0.34)*** 6.28 1.84 (0.33)*** 6.31
Non-white 0.27 (0.11)* 1.31 0.31 (0.11)** 1.37
Age 0.00 (0.01) 1.00 0.00 (0.01) 1.00
Alcohol/Drugs -0.23 (0.16) 0.79 -0.16 (0.16) 0.85
Mental/Emotion -0.43 (0.26) 0.65 -0.43 (0.26) 0.65
Weapon 0.14 (0.20) 1.16 0.15 (0.20) 1.17
Arrest 0.65 (0.22)** 1.91 0.61 (0.23)“ 1.85

Model Results

Pseudo R2 Level 1 0.18 0.18

Pseudo R2 Level IIa 0-44 0-75

Log Likelihood(MLR Factor) -698.06 (0.97) -690.75 (1.01)

 

***p<.001, **p<.01, *p<.0; 8Level II R2 values reﬂect the amount of variation explained by the models
in relation to the total amount of variation that is present between workgroups (i.e., ICC(1)=0.16).

159

With respect to distrust climate, results of the full model show that the probability
of using hard hand force techniques during encounters is elevated in workgroups with
distrustful climates. In fact, the odds of using hard hand force techniques are 99 percent
greater for every unit increase in the distrust climate measure based on the odds ratio
estimate. In addition to the estimates reported in Table 7.3, predicted probabilities can
also be calculated to better illustrate the nature of the effect of distrust climate on hard
hand force usage. The predicted probabilities were calculated using the intercept estimate
adjusted for the level 1 control variables (i.e., 0.80) and while holding the other level 2
variables at their sample means.

Based on the predicted probability calculations, the probability of using hard hand
force in encounters in workgroups with average climates of distrust (i.e., the sample
mean) was 0.21. On the other hand, the probability of using hard hand force in
workgroups with low climates of distrust (i.e., the minimum sample value) or highly
distrustful climates (i.e., the maximum sample value) were 0.11 and 0.42 respectively.
Thus, Officers working in environments highly distrustful of citizens had a much higher
likelihood of using hard hand force than Officers working in less distrustful workgroup
environments. Overall, these ﬁndings indicate that ofﬁcers are more likely to use
aggressive forms of physical force in distrustful climates even after controlling for ofﬁcer
characteristics, suspect characteristics, suspect resistance, and while accounting for the
department where the workgroup was located.

With respect to the unsupportive supervisor measure, a negative relationship was
found. The probability of using hard hand force techniques during encounters was lower

in workgroups with unsupportive supervisor climates. Furthermore, the odds of using

160

hard hand force are 46 percent less for every unit increase in this measure. The associated
predicted probabilities also illustrate this negative association. In workgroups with
average values of support (i.e., the sample mean), the probability of using hard hand force
is 0.21. In more supportive climates (i.e., the minimum sample value) and in highly
unsupportive climates (i.e., the maximum sample value), the probabilities of using hard
hand force were calculated to be 0.35 and 0.10 respectively. Here, Officers working in
highly unsupportive supervisor environments had a significantly lower likelihood of using
hard hand force than Officers working in more supportive supervisor environments.
Overall, these ﬁndings indicate that ofﬁcers are less likely to use aggressive forms of
physical force in environments where supervisors are perceived to be unsupportive of
ofﬁcers’ even after controlling for ofﬁcer characteristics, suspect characteristics, suspect
resistance, and while accounting for the department where the .workgroup was located.
With respect to the aggressive climate measure, a negative relationship was also
found. The probability of using hard hand force techniques during encounters decreased
in aggressive workgroup environments. Furthermore, the odds of using hard hand force
are 77 percent less for every unit increase in this measure. The associated predicted
probabilities also illustrate this negative association. In workgroups with average values
of aggressiveness (i.e., the sample mean), the probability of using hard hand force is 0.21.
In less aggressive climates (i.e., the minimum sample value) and in highly aggressive
climates (i.e., the maximum sample value), the probabilities of using hard hand force
were calculated to be 0.31 and 0.15 respectively. Here, oﬂicers working in highly
aggressive workgroup environments had a significantly lower likelihood of using hard

hand force than Officers working in less aggressive environments. Overall, these ﬁndings

161

indicate that ofﬁcers are more likely to use aggressive forms of physical force in less
aggressive environments even after controlling for ofﬁcer characteristics, suspect

characteristics, suspect resistance, and while accounting for the department where the

encounter occurred.32

Finally, it should be noted that there were several signiﬁcant ﬁndings among the
level 1 control variables. In the model, hard hand force techniques were more likely to be
used against male and non-white suspects as well as suspects who were not under the
inﬂuence of alcohol or drugs at the time of the encounter. Hard hand force was also more
likely to be used against suspects who exerted higher levels of resistance and in
encounters that resulted in arrest. In terms of ofﬁcer characteristics, hard hand force was
more likely to be used by male ofﬁcers. Overall, the ofﬁcer and suspect variables

explained approximately eighteen percent of the variation in hard hand force usage within

workgroups (pseudo R2 = 0.18). 33

Impact Force Results
The results of the unconditional model for impact force also initially indicated the

presence of signiﬁcant variation based upon the between-groups variance component (I =-

 

32 The relationship is in the opposite direction of what was hypothesized as well as what should be
expected based on the literature. Diagnostics were conducted to examine this. First, the coding of the
individual-level measure was checked and was found to be coded correctly. Second, bivariate associations
for the workgroup measures were examined. For example, aggressive climates were positively associated
with distrust climate and morale while negatively correlated with unsupportive supervisors, all of which are
in the expected direction. Third, the full model was re-run with only the aggressive climate measure to
determine if multicollinearity with the other workgroup measures was an issue. The effect was also
negative. Finally, different centering options were used and all produced the same negative relationship.

3 Diagnostic tests for multicollinearity were conducted for the models presented in Tables 7.3 at each
level using OLS regression models. At level 1, variance inﬂation factors (VIFs) were examined and all
VlFs were 1.27 or below. At level 2, this issue was examined using the workgroup mean of hard hand force
as the dependent variable. VIF values were all below 2.0. Thus, multicollinearity does not appear to be an
issue.

162

0.48, s.e. = .15, p<.01). The approximate amount of variation in impact force between
workgroups was calculated to be thirteen percent (ICC(1) = 0.13). After accounting for
department, however, the between-group variance component was found to be only
marginally signiﬁcant (t = 0.48, s.e. = 0.29, p<.10). Although a substantial amount of
variation in impact force exists between workgroups (i.e., 13%), it is likely to be
accounted for by departmental factors.

One such factor that might account for this ﬁnding involves the types of impact
weapons the study departments allowed their ofﬁcers to use, speciﬁcally regarding the
use of the taser. For example, at the time of the study all ofﬁcers were trained on and
required to carry the taser while on duty in PPB and CSPD. On the other hand, APD and
FWPD did not require all ofﬁcers to carry the taser. Compared to PPB and CSPD, fewer
APD ofﬁcers carried tasers while in F WPD only about twenty-ﬁve ofﬁcers in the entire
department were trained on and allowed to carry the taser. Thus, one reason for the non-
signiﬁcant workgroup effect on impact force after accounting for department may be due
to the differences in the types of impact weapons available to ofﬁcers across sites. Unlike
in the case of soft hand force where the tactics were at least available for ofﬁcers to use
across each department, different weapon restrictions, particularly with the taser, may
have made the impact force measure in its’ current form less comparable across
departments.34

In order to investigate this possibility, the unconditional model for the impact

force measure was conducted again. This time all instances involving the taser were

 

34 . . . . . . . . .

Unlike policy restrictions placed on impact force there were very few policy restrictions applied across
sites in the case of soft hand force. Hence all soft hand force options are available to all ofﬁcers across all
sites; this is not the case, however, for impact force options.

163

removed from the analyses which left 1,266 observations across the 53 workgroups. After
accounting for the department where the encounter occurred, the between-group variance
component was found to be signiﬁcant (1 = 0.70, s.e. = 0.28, p<.05). The approximate
amount of variation in impact force between workgroups was calculated to be eighteen
percent (ICC(1)=0.18). As such, the application of impact force by ofﬁcers differed
across workgroups upon excluding taser cases and further tests can be conducted to
explain this variation.

Table 7.4 presents the results for two multilevel models.35 First, an initial model
was tested to examine the impact of the four workgroup climate measures on impact
force while controlling for ofﬁcer and suspect characteristics. This initial model does not
account for department. The results indicate that none of the four measures signiﬁcantly
predicted impact force usage. Collectively, these measures explained only eleven percent
of the level 2 variation in hard hand force across workgroups. Despite the non-signiﬁcant
ﬁndings, a second model was still conducted and included the three department dummy
variables as level 2 controls. Additionally, distrust climate and aggressive climate were
dropped from this model since they provided the weakest results in the initial model.

After controlling for ofﬁcer and suspect variables as well as the department in
which the encounter occurred, unsupportive supervisor and morale remained non-
signiﬁcant. In fact only the PPB dummy variable was signiﬁcant at level 2: compared to
workgroups in FWPD, ofﬁcers in workgroups from PPB were signiﬁcantly less likely to

use impact force techniques. Collectively, these measures explained approximately 45

 

5 Diagnostic tests for multicollinearity were conducted for the models presented in Tables 7.4 at each
level using OLS regression models. At level 1, all VIFs were 1.26 or below. At level 2, this issue was
examined using the workgroup mean of impact force as the dependent variable. VIF values were all below
1.67. Thus, multicollinearity does not appear to be an issue.

164

Table 7.4: Multilevel Model Results for Impact Force without Taser Cases
Initial Model

 

Variable
134%)

Workgroup Level
Threshold 1.41 (0.16)***
Force Variance Comp. 0.59 (0.24)*
Distrust Climate 0.23 (0.43)
Unsupportive Supervisor 0.38 (0.37)
Morale 0.45 (0.28)
Aggressive Climate -0.49 (0.51)
CSPD --------
PPB --------
APD --------

Control Variables

Officer
Male -0.22 (0.42)
Non-white 0.22 (0.23)
Experience (In) 0.23 (0.11)*
Education -0.01 (0.05)
Citizen Distrust 0.08 (0.07)
Direct Supervisor 0.05 (0.09)
Job Satisfaction 0.09 (0.07)
Aggressiveness -0.17 (0.13)

Suspect
Resistance 0.53 (0.14)***
Male 0.08 (0.32)
Non-white -0.02 (0.17)
Age -0.01 (0.01)
Alcohol/Drugs -0.27 (0.18)
Mental/Emotion 0.20 (0.24)
Weapon 0.51 (0.27)'l'
Arrest 0.58 (0.31)T

Model Results

Pseudo R2 Level I 0.09

Pseudo R2 Level Ila 0-11

Log Likelihood (MLR Factor)

-535.36 (1.09)

Full Model

 

Expi b i

b (SE)

1.54 (0.13)***
0.37(0.17)*
0.24 (0.34)

-0.09 (0.29)
0.04 (0.51)

-1.10 (0.36)**
0.39 (0.37)

-023 (0.42)
0.17 (0.23)
0.24 (0.10)*

-0.01 (0.05)
0.07 (0.07)
0.04 (0.09)
0.10 (0.07)

-019 (0.13)

0.49 (0.14)**
0.06 (0.32)
-0.05 (0.17)
-001 (0.01)
-0.27 (0.18)
0.16 (0.25)
0.54 (0.27)*
0.48 (0.33)

0.09
0.45

-529.98 (1.09)

Expi b i

4.66

1.27

0.79
1.18
1.27
0.99
1.07
1.04
1.11
0.82

1.63
1.06
0.95
0.99
0.76
1.17
1.72
1.62

 

***p<.001, **p<.01, *p<.05,

p<.06; N=l ,l 16; aLevel II R2 values reﬂect the amount of variation

explained by the models in relation to the total amount of variation that is present between workgroups

(i.e., ICC(1)=0.18).

165

percent of the level 2 variation in hard hand force across workgroups. Again, although the
level 2 pseudo R2 is 0.45 only 18% of the variation in impact force is between-
workgroups. As such, the results from both models indicate that workgroup attitudinal
measures are not important predictors of Officers ’ impact force behavior.

When examining ofﬁcer and suspect characteristics, it should be noted that there
were only three signiﬁcant ﬁndings among the level 1 control variables. In the model,
impact force was more likely to be used against suspects who exerted higher levels of
resistance during the encounter and against suspects who possessed a weapon. Impact
force was also more likely to be used by more experienced ofﬁcers. Overall, the ofﬁcer

and suspect variables explained only approximately nine percent of the variation in hard

hand force usage within workgroups (pseudo R2 = 0.09).

Discussion

The ﬁnal research objective examined whether ofﬁcers’ collective perceptions of
the workgroup environment inﬂuenced use of force behavior. Speciﬁcally, analyses
sought to determine if workgroup measures of distrust climate, supervisor support,
morale, and aggressive climate explained differences in ofﬁcers’ use of soft hand, hard
hand, and impact force techniques during encounters with citizens. Although the use of
soft hand force and impact force was not found to be inﬂuenced by workgroup level
factors, several noteworthy results were found for hard hand force. Ofﬁcers’ use of hard
hand force techniques (i.e., striking with the hands, body, or feet) was signiﬁcantly
elevated in workgroup environments that had distrustful climates toward citizens. On the
other hand, hard liand force usage was signiﬁcantly lower in workgroups with

unsupportive supervisors and in less aggressive workgroup climates. Overall, these

166

ﬁndings indicate support for the notion that collective, perceptual measures of ofﬁcers
work environments inﬂuence use of force behavior even after controlling for suspect
resistance, suspect characteristics, individual ofﬁcer characteristics, and while controlling
for site.

The current ﬁndings offer insight into the study of ofﬁcer behavior and attitudes.
First, the ﬁndings indicate that the formal organization of policing inﬂuences ofﬁcers’
use of force behavior, particularly at the workgroup level. This illustrates the importance
of taking into account ofﬁcers’ organizational environments when examining
discretionary behaviors. Additionally, it questions approaches that solely emphasize
individual level assumptions. For example, unconditional model results showed that
between 1 1 (soft hand force) and 18 (impact force) percent of the variation in ofﬁcers’
use of force tactics can be explained by the formal organization of policing. This implies
that force usage among ofﬁcers are not independent, but rather cluster across
organizational levels (i.e., workgroups). The results reported throughout this chapter
illustrate the need to heed calls made by policing scholars to incorporate the formal
organization of policing into the study of ofﬁcer behavior (e.g., Klinger, 1997, 2004).

Second, speciﬁc factors associated with the organizational environment should
begin to be incorporated into studies of force behavior. Although, research has focused
on how neighborhood characteristics inﬂuence use of force behavior (Smith, 1986; Terrill
& Reisig, 2003), the importance of the organizational environment has not been
adequately examined. The current results found that the application of hard hand and
impact force signiﬁcantly differed across workgroups. Furthermore, workgroup factors

were examined and found to predict hard hand force behaviors. The ﬁndings illustrate the

167

importance of incorporating organizational approaches into future studies of ofﬁcer
behaviors.

Finally, the results build upon the study of the police attitude-behavior
relationship. Ofﬁcers’ collective perceptions of their workgroup environments were
related to differences in hard hand force usage. Historically, individual level approaches
to the police attitude-behavior relationship have failed to uncover such relationships (See
Chapter Two). More recently, studies that view ofﬁcer attitudes as shared attributes (i.e.
police culture) but treat them as individual attributes have begun to uncover signiﬁcant
effects (McCluskey et al., 2005; Paoline & Terrill, 2005; Terrill et al., 2003). The current
study has shown that ofﬁcers’ occupational attitudes represent valid and reliable
measures of ofﬁcers’ work environments at the workgroup level. Furthermore, these
workgroup measures inﬂuence street-level behaviors of ofﬁcers. As such, the study has
extended this line of research by taking a multilevel approach that more accurately
captures the shared and collective properties of the occupational attitudes of police.

Although, the results offer insight into the study of ofﬁcer attitudes and behavior,
they are not without limitations. First, the three separate force measures used differs from
prior research. For example, recent studies place the use of force on a continuum and
have used the highest level of force used in the encounter as the outcome indicator
(McCluskey et al., 2005; Terrill, 2001; Terrill & Mastrofski, 2002; Terrill et al., 2003).
Such a measure was difﬁcult to construct in the context of the current study because the
progression of force tactics varied across sites; similar types of force were placed at
different levels of the continuum. For example, in PPB OC spray (impact force tactic)

was regarded as a lower level of force than hard hand force, while in APD OC spray and

168

hard hand force were regarded as the same level of force. Thus, a common measure of
force that captures such an ordinal progression could not be deﬁned across sites.
Although the use of three separate measures allows insight into how each different type
of force was inﬂuenced by workgroup measures, the approach does differ from recent
research.

Second, upon excluding the taser cases from the impact force measure, impact
force is driven largely by the use of OC Spray. For example after excluding the taser,
impact force was used in 282 of the 1,266 remaining force incidents. Approximately 239
(85%) of these cases involved the use of OC spray. When assessing the results for impact
force, then, it is important to consider this because the use of OC spray generally
represents a lower level of impact force when compared to other tactics such as the use of
the baton or impact munitions. As such, it is important to note that different
conﬁgurations for measuring force may impact the results.

Finally, the multilevel models associated with hard hand and impact force can be
improved upon. For example, the collective contribution of the level 1 control variables
explained only eighteen and nine percent of the variation in hard hand and impact force,
respectively. Because the models test the impact of workgroup measures while
controlling for ofﬁcer and suspect variables, the presence of unexplained variation at
level 1 means that there are other individual level variables missing from the models.
This is a limitation of using ofﬁcial police data. Even though all relevant control
variables that could be incorporated from the force report data were included in the
models, inclusion of these unaccounted variables could alter the workgroup results.

Therefore, the ﬁndings should be interpreted with this in mind.

169

CHAPTER EIGHT: CONCLUSION

The purpose of this study was to offer and empirically test a multilevel framework
for examining police ofﬁcers’ occupational attitudes. Early observational studies of the
police identiﬁed a set of attitudes ofﬁcers’ developed in response to the challenges they
faced in their work environment, which include distrustful views toward citizens, narrow
role orientations geared toward crime-ﬁghting, negative views toward supervision,
aggressive policing tactics, and job satisfaction (Bittner, 1970; Manning, 1997; Skolnick,
1967; Westley, 1970). Not all ofﬁcers hold similar dispositions, however, as research on
the police has documented attitudinal differences among ofﬁcers (Brown, 1988; Worden,
1995). As a result, a considerable amount of empirical research has sought to explain
ofﬁcers’ attitudinal differences as well as examine how ofﬁcers’ attitudes impact their
street level discretionary behaviors. Although these studies have contributed greatly to
our understanding of policing, they are not without limitations. The extant literature on
ofﬁcers’ attitudes has been heavily rooted within individual approaches that neglect the
environments in which ofﬁcers work (Klinger, 2004). This inhibits conceptual and
methodological development of research in this area.

The current study has expanded this line of research by moving beyond individual
approaches and taking a multilevel approach to the study of ofﬁcers’ occupational
attitudes. Speciﬁcally, the current work incorporates ofﬁcers’ organizational context and
ideas associated with police culture into a conceptual and methodological framework for
understanding environmental inﬂuences on ofﬁcer attitudes. Central to this framework is
the importance of the formal organization of policing as it represents structural

boundaries of policing (i.e., ofﬁcers work in shifts, shifts are embedded within precincts,

170

precincts are embedded in departments). The current study identiﬁed ofﬁcers’ shifts as
the most salient boundary that inﬂuence ofﬁcers’ attitudes. Deﬁned as workgroups (i.e.,
shifts within precincts), ofﬁcers assigned to the same workgroups are exposed to similar
environmental characteristics, supervisors, and interact routinely with fellow ofﬁcers. In
other words, workgroups are where the daily routine of police work is carried out, and as
a result, represent the most immediate and proximal level of the organization that shapes
ofﬁcers’ occupational attitudes. In addition, based on ideas central to police culture the
current framework also argued that ofﬁcers within the same workgroup setting share
occupational attitudes. As a result of their daily interactions and exposure to similar
environments and supervisors, ofﬁcers develop shared attitudes and these should be
examined as collective, contextual measures of ofﬁcers’ workgroup environments.

Two key ideas emerge from the conceptual framework of the study. First, features
of ofﬁcers’ organizational environment should be examined as explanations for
attitudinal differences among ofﬁcers. Second, if ofﬁcers share attitudes then collective
attitudinal measures should be examined in relation to ofﬁcers’ street level discretionary
behaviors. In order to examine these ideas, the current study tested three research
objectives. First, the study tested the inﬂuence of ofﬁcers’ workgroups on their
occupational attitudes. It also empirically examined whether ofﬁcers within workgroups
shared attitudes. Second, the study tested the effect of workgroup compositional
characteristics as explanations of ofﬁcer attitudes. Finally, the study tested the effect of
workgroup level attitudinal measures as predictors of ofﬁcer use of force behavior. Thus,
the study not only offered a theoretical framework but also assessed the usefulness of the

framework in explaining attitudinal differences as well as extending research on the

171

police attitude-behavior relationship. The remainder of this chapter provides a summary
of the key ﬁndings from the study, identiﬁes the study limitations, and ofﬁcers practical
and future research implications based upon the results.
Summary of Key Findings
Views toward Citizens

Ofﬁcers’ views toward citizens have been a key occupational attitude since the
early observational studies of police work were conducted. In order to examine the nature
of ofﬁcers’ attitudes toward citizens, the current study asked ofﬁcers the extent to which
they believed that ofﬁcers should be distrustful and suspicious of citizens. Responses
were examined in relation to the formal organization of policing, particularly at the
workgroup level. Based on the results, attitudes toward citizens provided some of the
most interesting ﬁndings.

0 Ofﬁcers within workgroups shared attitudes toward citizens. Further,
ofﬁcers’ attitudes toward citizens differed across workgroups,
meaning that ofﬁcers within some workgroups had more distrustful
views toward citizens than ofﬁcers within other workgroups.

0 Although ofﬁcers’ attitudes toward citizens signiﬁcantly differed
across workgroups, workgroup compositional characteristics offered
little insight as to why these differences existed.

0 Ofﬁcers’ views toward citizens represented valid, contextual measures
of ofﬁcers’ workgroup environments. In other words, a climate of
distrust was present at the workgroup level. These climates of distrust
were important explanations of the use of hard hand force during
ofﬁcers’ encounters with citizens.

These results offer insight into the collective, multilevel nature of the police

attitude-behavior relationship. Ofﬁcers were found to share attitudes toward citizens and

this collective perception within the workgroup setting inﬂuenced ofﬁcers’ behavior.

172

Thus, even though attitudes are held by individual ofﬁcers, the current results suggest
that their effects are ampliﬁed and manifest at higher organizational levels. Here, ofﬁcers
working in more distrustful workgroups were signiﬁcantly more likely to use hard hand
force tactics than ofﬁcers who worked in less distrustful workgroups. Ofﬁcers in
workgroups with more distrustful climates were more likely to use aggressive forms of
physical force (e.g., strikes) even after controlling for resistance levels, suspect
characteristics, ofﬁcer characteristics, and departments.

It is important to note that the purpose of the study was not to examine improper
or excessive use of force. The results only suggest that ofﬁcers in more distrustful
climates rely on more aggressive forms of physical force at greater rates than ofﬁcers in
less distrustful climates. As such, whether this ﬁnding is viewed as a positive or negative
one would depend on how one views the nature of this attitude. For example, if ofﬁcers
within workgroups stress a distrustful attitude toward citizens as a safety precaution, then
these results may be viewed in a positive light. On the other hand, if distrustful attitudes
result from negative byproducts of police work, such as prejudices (e.g., how citizen
distrust is often described in the policing literature), then these results may not be ideal
for police administrators. The current study did not uncover speciﬁc workgroup factors to
explain why ofﬁcers within some workgroups were more distrustful of citizens than
ofﬁcers within other workgroups. Additional inquiry is necessary to shed light on the

underlying reasons behind the multilevel nature of this attitudinal measure.

Role Orientations
Traditionally, ofﬁcers are said to hold narrow attitudes toward the police role that

emphasize crime-ﬁghting rather than accepting broader roles that encompass order

173

maintenance type activities (e.g., Bittner, 1990; Manning, 1997). The current study asked
ofﬁcers about their acceptance of different job roles and examined the extent to which
their attitudes were inﬂuenced by workgroups. First, ofﬁcers were asked whether they
believed law enforcement (i.e., crime-ﬁghting) was their most important responsibility.
Second, they were asked the extent to which ofﬁcers should be required to do anything
about six order maintenance activities, including public nuisances, neighbor disputes,
family disputes, parents who do not control their children, litter and trash, and nuisance
businesses. Examining ofﬁcer role orientations in relation to their workgroups yielded
several interesting results:

0 Workgroups did have an inﬂuence on both crime-ﬁghting and order
maintenance role orientations. In some workgroups, ofﬁcers held
more narrow orientations geared towards crime ﬁghting while in
other workgroups ofﬁcers held broader role orientations. Also,
ofﬁcers within workgroups were found to share role orientations.

o Workgroup racial composition and educational levels were strong
inﬂuences on role orientations. Controlling for individual and
department factors, workgroups with higher proportions of white
ofﬁcers and higher educational levels contained ofﬁcers who were less
likely to accept order maintenance roles. On the other hand,

workgroups with higher proportions of white ofﬁcers were less likely
to believe that law enforcement was the most important police role.

0 Although ofﬁcers within workgroups shared role orientations,
reliability tests did not offer empirical support for treating role
orientations as contextual measures of the workgroup environment.

These results offer support for the notion that higher level, organizational factors
inﬂuence ofﬁcers’ acceptance of j ob roles. At the workgroup level, racial composition
and educational levels had an impact; however the nature of the results is conﬂicting. For
example, workgroups with higher proportions of white ofﬁcers may reﬂect ideas

underlying the early observational studies of the police that took an occupational

174

perspective. This offers a potential explanation for why ofﬁcers within these workgroups
were less likely to accept order maintenance activities, but does not necessarily explain
the negative crime ﬁghting effect. Thus, the current results offer mixed results and future
research could examine the effect of racial composition on role orientations with this in
mind.

The ﬁnding for workgroup educational levels is also concerning. One of the
arguments for higher education for ofﬁcers is to expose them to viewpoints that would
make them more accepting of broader police roles (e.g., Worden, I990). The current
results, however, indicate the opposite effect at an aggregate level. At the collective level,
ofﬁcers who are exposed to more highly educated ofﬁcers within their workgroups and
interact on a routine basis may conclude that the performance of order maintenance
activities (particularly the types included in survey) may perceive these activities as
rather menial tasks over time. Ofﬁcer education as a compositional effect appears to
amplify more narrow views toward order maintenance within workgroup settings.

Finally, future research may beneﬁt by further examining the collective properties
of ofﬁcer role orientations. The results indicated that ofﬁcers within workgroups shared
similar orientations, yet the reliability for treating them as measures of the workgroup
environment was poor. Even though the ﬁndings suggested that workgroup level
orientations would have been valid constructs, they would not represent reliable ones.
This is likely due to the fact that these two measures had low ICC(1) values across
workgroups. Based on these values the current study indicates that role orientations best
represent individual level constructs; however these constructs are inﬂuenced by the

broader organizational environment.

175

Views toward Supervision

Within the policing literature, ofﬁcers have been described as holding negative
views toward two levels of supervision: direct supervisors and top management (e.g.,
Reuss-Ianni, 1983; Van Maanen, 1974). In order to determine the effect of the formal
organization of policing on ofﬁcers’ attitudes toward supervision, ofﬁcers were asked the
extent to which they felt that direct supervisors supported their efforts as well as the
extent to which they felt top management recognized their efforts and treated ofﬁcers
fairly. Overall, the results revealed that ofﬁcers held much more positive outlooks toward
direct supervisors than top management. When viewed within the context of ofﬁcers’
assigned workgroups several interesting ﬁndings emerged:

With respect to ofﬁcers’ views toward direct supervisors, key ﬁndings included:

0 Ofﬁcers within workgroups shared attitudes toward direct
supervisors. In addition, ofﬁcers’ attitudes toward supervisors
differed across workgroups, meaning that ofﬁcers within some
workgroups had more negative views toward supervisors than ofﬁcers
within other workgroups.

- Differences in ofﬁcers’ attitudes toward direct supervisors were
explained by workgroup experience and educational levels.
Workgroups with higher experience and educational levels had more
negative views toward supervisors.

0 Ofﬁcers’ attitudes toward supervisors were found to reﬂect a climate
of support at the workgroup level. Furthermore, these climates
signiﬁcantly inﬂuenced ofﬁcers’ use of force behavior. Ofﬁcers
working in unsupportive climates were signiﬁcantly less likely to
engage in hard hand force tactics.

Collectively, these ﬁndings highlight two multilevel inﬂuences that are important for
understanding the nature of ofﬁcers’ views toward supervisors. First, the results suggest

that within workgroup settings, aggregate experience and educational levels lead to an

176

ampliﬁed effect on attitudes toward supervisors. Perhaps, the daily interactions of
ofﬁcers within workgroups that contain ofﬁcers who have longer years of service or
higher educational levels yield increased levels of frustration toward supervisors whereby
ofﬁcers feel that their own ideas and contributions are not supported by their supervisors.

Second, the collective, negative effect of unsupportive supervisor climates also
provides important insight into ofﬁcers’ force behavior. In these unsupportive climates,
ofﬁcers were more likely to report that supervisors do not look out for their personal
welfare and that their supervisors’ approach discourages them from giving extra effort.
The current results indicate that because of this, ofﬁcers may fail to rely on hard hand
force tactics. Much like the results for citizen distrust, whether this ﬁnding is viewed
positively or negatively depends on how one views the nature of this attitude.

On one hand, ofﬁcers may refrain from using aggressive forms of physical force
because these tactics may be less likely to be approved by supervisory review measures.
Although it should be noted here again that the study did not examine excessive or
improper force, the use of hard hand tactics, such as strikes, are likely the types of force
that lead to such allegations. Supervisors may be less inclined to support these types of
tactics as a result, especially when other types of tactics are available to ofﬁcers.
Furthermore, since hard hand tactics may lead to potential injuries of citizens and
ofﬁcers, supervisors may be less inclined to support this type of behavior. In other words,
this negative effect could be viewed positively through the lens of affective supervision.

On the other hand, this ﬁnding could also be viewed as a negative one. Whether
ofﬁcers in unsupportive climates consciously or unconsciously refrain from the use of

hard hand force cannot be determined; however, this could raise potential problems if

177

ofﬁcers feel hesitant to rely on certain force tactics that are available to them simply
because they feel that supervisors will not back them up on it. Ofﬁcers’ may feel overly
restricted in the types of force they can readily use, even when it is an appropriate option
based upon the circumstances of the incident. For example, if it is the case that ofﬁcers
feel that they must rely on lower or alternative forms of force in encounters with higher
levels of resistance, this could pose safety issues to ofﬁcers.
With respect to ofﬁcers’ views toward top management, the key ﬁndings

included:

0 Ofﬁcers within workgroups shared views toward top management.

However, after controlling for differences across departments, this

measure was not found to signiﬁcantly differ across workgroups.

«- Ofﬁcers’ views toward top management did not reﬂect contextual
measures of the workgroup environment.

Initial analyses for this attitude seemed to indicate that top management would be
the most strongly inﬂuenced by workgroups. However, after taking into account the
clustering of these attitudes by department, the inﬂuence of workgroups became non-
signiﬁcant. This result does suggest that higher-level organizational factors do inﬂuence
ofﬁcers’ views toward top management. It is just that these views are likely inﬂuenced by
department level factors, such as leadership and administrative factors rather than
workgroup related factors. These results are important for future studies of ofﬁcers’
occupational attitudes that seek to use measures of top management. The current ﬁndings
suggest that future studies would need to employ a sample of ofﬁcers from a larger
number of different departments in order to detect and explain any attitudinal differences

among ofﬁcers on this measure.

178

Policing Tactics

Two key occupational attitudes that are expected to inﬂuence ofﬁcers’
discretionary behaviors include two views toward policing tactics: aggressiveness and
selectivity. Ofﬁcer aggressiveness measures the importance of proactive patrol tactics
(Brown, 1988). Ofﬁcers were asked the extent to which they believed conducting stops
and checking citizens out were important tactics that characterize a “good” police ofﬁcer.
The inﬂuence of the formal organization of policing, particularly at the workgroup level
on ofﬁcers’ aggressiveness was tested. Several important ﬁndings emerged.

0 Ofﬁcers within workgroups shared attitudes on aggressiveness.
Aggressive attitudes also differed across workgroups, meaning that
ofﬁcers within some workgroups held more aggressive attitudes then
ofﬁcers in other workgroups.

0 Differences in aggressive attitudes were explained by workgroup
experience. Workgroups containing ofﬁcers with more experience
held less aggressive attitudes.

0 Ofﬁcer attitudes were found to reﬂect a workgroup climate of
aggressiveness. Furthermore, this climate of aggressiveness explained
ofﬁcers’ use of hard hand force; however, the ﬁnding was in the
opposite direction of what was expected.

Once again the results offer support for examining the inﬂuence of higher level,
organizational factors on ofﬁcers’ occupational attitudes. Here, workgroup experience
levels were related to aggressive attitudes even after controlling for individual ofﬁcer
experience levels. Thus, further evidence exists to support the idea that features of
ofﬁcers’ organizational environment inﬂuence ofﬁcer attitudes.

It should be noted that the negative relationship between workgroup aggressive

climates and the use of hard hand force is in the opposite direction of what might be

179

expected. For example at the individual level, prior studies have found that ofﬁcers in
aggressive peer groups were more likely to use higher levels of force during citizen
encounters (McCluskey et al., 2005). The multilevel result in the current study, however,
indicated a negative effect. Methodologically, this could be due to measurement error
since only a single item was used to measure aggressiveness. On the other hand, the
negative relationship could be due to the fact that aggressiveness reﬂects ofﬁcers’
proactive views toward patrol practices rather than reﬂecting a more authoritative
measure. Traditional depictions of aggressiveness imply that ofﬁcers are aggressive as a
way to minimize potential threats of violence. Perhaps, this is an important distinction to
be made when conceptualizing “aggressiveness.” Thus, the current results may suggest
that even though aggressive environments favor proactive patrol practices, ofﬁcers within
them will not necessarily resort to more aggressive forms of physical force during
encounters that require force to be used.

Selective enforcement measures how ofﬁcers view the use of the law to perform
their duties. Speciﬁcally, selectivity provides an indication as to how ofﬁcers view their
discretionary powers to control crime problems (Brown, 1988). Ofﬁcers were asked
whether they believed it was more effective to focus on felony offenses rather than
spending time on more minor law violations such as trafﬁc and misdemeanor offenses.
Several ﬁndings are worth highlighting.

First, ofﬁcers within workgroup settings were found to share this occupational
attitude. In fact, views toward selective enforcement had the strongest within group
agreement results across workgroups. Although there was evidence that ofﬁcers within

workgroups did share this attitude, selectivity was not inﬂuenced by the formal

180

organization of policing. No signiﬁcant differences were found among ofﬁcers across
workgroups, precincts, or departments for this attitude. When examined in relation to the
objectives of the current study two conclusions can be drawn:
0 Organizational factors (e.g., workgroup composition) are not viable
explanations of ofﬁcers’ views toward selective enforcement of the

law.

- Selectivity does not reﬂect a contextual measure of ofﬁcers’
workgroup environment.

Across the entire sample, over eighty percent of ofﬁcers disagreed with the idea that it is
more effective to focus on serious offenses rather than more minor offenses. The fact that
the majority of ofﬁcers disagreed with selective enforcement practices explains why
ofﬁcers within workgroups shared this attitude, yet there were no signiﬁcant differences
in this attitude across workgroups (or precincts and departments). From a methodological
standpoint, this could be due to the fact that only a single item was used to measure
selectivity. Although, the item used was the same measure adopted by Brown (1988) who
developed the concept, perhaps a better conceptualizationof this construct is necessary in
order to detect differences in ofﬁcers’ views toward enforcement practices.

From a practical standpoint, however, these ﬁndings could be the result of a
broader trend in urban departments that emphasize enforcement of minor crime and
disorder issues. For example, policing strategies related to broken windows theory have
been implemented in departments since the 19803. These strategies emphasize addressing
minor crime problems and disorder as an effective way of preventing more serious crime
problems from occurring (Wilson & Kelling, 1982). Additionally, emphasis has also been
placed on efforts to enforce minor trafﬁc offenses because they can lead ofﬁcers to
uncover more serious felony offenses as a result (e. g., pretextual trafﬁc stops) (Harris,

181

2002). For these reasons, it may also be the case that ofﬁcers’ preferences for selective
enforcement of the law have changed so that most patrol ofﬁcers now believe that

focusing on minor offenses is just as important as focusing on more serious offenses.

Job Satisfaction

One ﬁnal attitude concerns ofﬁcers’ job satisfaction. Ofﬁcers were asked
questions about their satisfaction in terms of whether they enjoyed their job and whether
they would prefer having another job. These responses were then examined in relation to
ofﬁcers’ workgroup environment. The following highlights some of the key ﬁndings
from the study for this attitudinal measure.

0 Ofﬁcers within workgroups shared levels of satisfaction. Further,
ofﬁcers’ job satisfaction levels differed across workgroups, meaning
that ofﬁcers within some workgroups were more satisﬁed with their
jobs than ofﬁcers in other workgroups.

0 Differences in satisfaction levels were explained by workgroup
experience and educational levels. Workgroups containing ofﬁcers
with more experience and higher educational levels had lower levels of
job satisfaction.

0 Ofﬁcers’ satisfaction levels were found to reﬂect a workgroup
measure of morale. Workgroup morale, however, was not found to
explain ofﬁcers’ use of force behavior.

These results provide support for the notion of examining how features of
ofﬁcers’ organizational environment inﬂuence job satisfaction. Controlling for individual
ofﬁcer characteristics, it was found that within the workgroup setting, ofﬁcers who were
exposed to other ofﬁcers with more experience and higher educational levels were less
likely to be satisﬁed with their jobs. Prior research has suggested that ofﬁcers with longer

years of service may become more cynical about their job (e.g., Barker, 1999); however

182

empirical research has not found a strong effect at the individual level linking experience
to satisfaction (see Chapter Two). The current results suggest that being exposed to more
experienced ofﬁcers and interacting with them on a routine basis may lead to increased
levels of dissatisfaction. In other words, experience levels exert a stronger aggregate
effect on satisfaction levels than individual ofﬁcer experience levels.

Although workgroup morale was not found to be an important explanatory
variable for use of force behavior, the fact that morale reﬂected a contextual measure of
ofﬁcers’ workgroup environment warrants further examination in relation to other types
of street level discretionary behaviors. For example, Moskos (2008) observed that
ofﬁcers who worked together in environments with low morale were less likely to make
discretionary arrests. Thus, examining workgroup morale as a contextual inﬂuence may

offer insight into the decision to arrest.

Study Limitations

Although there are a number of strengths of the study, it is not without limitations.
One limitation concerns the generalizability of the results to other departments. While the
research design enhances generalizability of the results to departments of similar size and
structure, the multilevel theoretical framework and methodology may not extend to
smaller departments. Workgroups may not serve as important contexts in smaller
departments because smaller agencies are more highly centralized and have fewer
ofﬁcers. There will likely be fewer attitudinal differences among ofﬁcers within smaller
agencies.

A second limitation is the cross-sectional nature of the survey design. A key

component of the theoretical framework is that ofﬁcers within workgroups share

183

occupational attitudes. This assumes that ofﬁcers in workgroups have worked together
long enough to have sufﬁcient time to develop shared meanings. The cross-sectional
nature of the study, however, does not allow for the assessment of the amount of time
ofﬁcers worked together in each workgroup. Two characteristics may serve to lessen the
nature of this limitation. First, the bid systems for shift allocations across each site
indicated that shift assignments should remain relatively stable over the’course of the
year. Second, surveys were administered in the fall of 2007 allowing most workgroups to
have been formed for at least nine months prior to data collection. Despite these points,
the amount of time ofﬁcers were a part of each workgroup cannot be explicitly
incorporated into the models. When examining the within group agreement results from
Chapter Five, this might offer a partial explanation as to why some workgroups had
strong levels of attitudinal agreement while others had insufﬁcient levels of agreement.

A third limitation is also related to the cross-sectional nature of the study design.
Because the study was cross-sectional, ofﬁcers’ prior assignments could not be
incorporated. This is unfortunate because ofﬁcers’ prior experiences within their
department could also serve as attitudinal inﬂuences. Similarly, research has examined
the effect that certain assignments such as being a ﬁeld training ofﬁcer (Sun, 2002) or
community policing ofﬁcer (Paoline, 2001) has on their attitudes. Unfortunately, only
patrol ofﬁcers were included in the study. It would be interesting to see how the results of
the study would compare to specialized units that have been described as being more
homogenous with respect to their views.

A ﬁnal limitation is that the study focused solely on how the formal organization

of policing affected ofﬁcer attitudes. This excludes examination of other informal units.

184

Prior research has stressed the importance of informal groupings of ofﬁcers based on
typology studies (e.g., Cochran & Bromley, 2003; Jermier et al., 1991; Paoline, 2001) in
shaping ofﬁcer attitudes. Although a limitation, focusing solely on formal organizational
units was deemed necessary in the current work because the multilevel approach taken
has not been applied to the study of ofﬁcer attitudes. Kozlowski and Klein (2000: 13)
note that when such is the case, researchers “may ﬁnd it helpful to initially act as if the
phenomena occur at only one level of theory and analysis,” and then extend the approach

to other units when necessary. This was the approach taken here.

Research & Practical Implications
The study results have important implications for future police research,

particularly for research on explaining ofﬁcer attitudes, police culture, and police

behavior.36 First, research that seeks to explain differences in ofﬁcers' occupational
attitudes should move beyond individual level factors and focus efforts on contextual
factors. Prior studies have suggested that features of ofﬁcers‘ organizational contexts be
incorporated into research on ofﬁcer attitudes (e.g., Brooks et al., 1993; DeJong, 2004;
Paoline et al., 2000; Sun, 2002, 2003). The current study found ofﬁcers' assigned shifts
within precincts (i.e., workgroups) to inﬂuence their views toward citizens, role
orientations, supervision, job satisfaction, and aggressive patrol tactics. The current
inquiry also went beyond determining if workgroups inﬂuenced attitudes by examining
what it is speciﬁcally about workgroups that affect attitudes by exploring the impact of

workgroup compositional characteristics.

 

This section addresses research and practical implications from a broad perspective. For more specrﬁc
discussion as to how this study extends research with respect to the speciﬁc research objectives of the
study, refer to the discussion sections at the end of Chapters Six and Seven.

185

Future research on this topic should incorporate other contextual factors
associated with ofﬁcers' workgroup environments. For example at the workgroup level,
ofﬁcers' distrust of citizens was not explained by compositional characteristics. Levels of
distrust may more likely be inﬂuenced by the type of environments ofﬁcers work in
within their workgroup setting, such as area crime rates or the types of neighborhoods in
which they patrol. These characteristics may better explain citizen distrust since they are
more closely tied to this construct. In essence, future research should begin to explore
other types of multilevel inﬂuences on ofﬁcer occupational attitudes.

Second, current research that takes an attitudinal approach to the study of police
culture is based on the notion that ofﬁcers share occupational attitudes (Paoline, 2003).
As Paoline (2003: 210) has noted, however, "a critical missing component of [police
culture] is establishing a threshold of ofﬁcer agreement in assessing cultural and
subcultural attitudes, values, and norms." Based on multilevel theoretical and
methodological frameworks, this study has offered an empirical approach for assessing
ofﬁcers' attitudinal agreement. Future research on police culture may beneﬁt by
incorporating such approaches to determine the extent to which ofﬁcers do share
attitudes, whether it be across various ofﬁcer typologies or other ofﬁcer groupings.

Furthermore, studies of police culture have combined ofﬁcer attitudes into a
single attribute (e. g., pro-culture, mid-culture, con-culture) and then assigned this
attribute to individual ofﬁcers (e.g., Paoline & Terrill, 2005; Terrill et al., 2003). Results
from the current study, however, indicate that cultural attitudes may be better viewed as
collective attributes, or multilevel constructs rather than individual attributes. Cultural

researchers have noted that if a concept is considered to be shared by individuals, then it

186

is collective (or multilevel) in nature (e.g., Ostroff et al., 2003). This implies that the
conceptualization and operationalization of culture should be taken from a multilevel
rather than individual perspective. The current work has begun to explore the viability of
such multilevel approaches by integrating ofﬁcer attitudes with the organizational
environment of policing (see also Hassell, 2006). This approach readily extends to the
area of police culture as the two topics are closely related.

Finally, the ﬁndings from this dissertation offer insight into the police attitude-
behavior relationship. Within workgroups ofﬁcers shared views toward citizens,
supervisors, job satisfaction, and aggressiveness indicating that multilevel climates of
distrust, support, aggressiveness, and morale existed at the workgroup level. These
climates were related to ofﬁcers' use of hard hand force. As a result, the multilevel
approach taken in this study should be examined in relation to other types of street level
discretionary behaviors. For example, workgroup climates of distrust may also explain
behaviors that result in complaints against ofﬁcers. Ofﬁcers who work in distrustful and
suspicious climates may be more likely to have citizen complaints brought against them.
F urthennore, ofﬁcers who work in aggressive climates may be more likely to make
discretionary arrests and conduct trafﬁc stops than ofﬁcers who work in less aggressive
climates. Thus, one of the major strengths of this study is that is has offered an approach
that supports further inquiry into research on the police attitude-behavior relationship.

The purpose of this chapter was to summarize and highlight the key ﬁndings
across each occupational attitude examined in the study. When appropriate, practical
implications were offered based on the results for each attitude. However, there are two

overarching themes that consistently were found and serve as important practical

187

implications for police administrators. First, the results show that groups inﬂuence ofﬁcer
attitudes, particularly at the workgroup (i.e., shift level). Ofﬁcers working the same shift
in the same precinct share similar occupational attitudes. Furthermore, the nature of
ofﬁcers’ attitudes is also inﬂuenced by characteristics of other ofﬁcers working the same
shift. Thus, police administrators may need to examine how ofﬁcers are assigned to
patrol in order to ensure that ofﬁcer attitudes align with the department’s values. If
departments make patrol assignments based exclusively on bid systems that favor ofﬁcer
preferences or seniority, then patrol shifts may lack the diversity that would facilitate the
development of different occupational outlooks among ofﬁcers. For example, one of the
key ﬁndings of the current study was that ofﬁcers who work in shifts composed of higher
proportions of white ofﬁcers were less likely to accept broader job roles associated with
order maintenance activities. Perhaps this lack of diversity within the shift context creates
a “group think” mentality among ofﬁcers. Increasing the racial diversity of shifts may
facilitate broader acceptance of job roles. In essence, administrators should consider the
inﬂuence that group dynamics has on ofﬁcer attitudes.

Second, the ﬁndings related to the multilevel nature of the police attitude-
behavior relationship have implications for police administrators. For example, police
scholars have contended that features of the police organization serve to attenuate the link
between individual ofﬁcer attitudes and their behavior. As the National Research Council
(2004: 136) surmised, “the failure to ﬁnd substantial links between personal attitudes and
behaviors is not necessarily bad news. . ., if that means that ofﬁcers are behaving
consistent with the organizations’ goals instead of their own beliefs.” However, the

current study found ofﬁcers to share attitudes across the workgroup context, which

188

suggests that characteristics of the organization serve to amplify rather than attenuate the
attitude-behavior link. Thus, departmental policies, such as training and disciplinary
policies, may need to be examined with this in mind.

Using the results for workgroup climates of distrust on hard hand force usage may
better illustrate this point. Shifts with distrustful climates toward citizens had ofﬁcers
who used higher rates of hard hand force during encounters with citizens. In some
instances, this may not be viewed as a favorable outcome as it could plausibly lead to
allegations of improper force or excessive force complaints by citizens. When such
instances occur, traditional departmental policies have focused on re-training individual
ofﬁcers or disciplinary procedures aimed at changing individual ofﬁcer attitudes (e. g.,
NRC, 2004). If ofﬁcers’ attitudes toward citizens, however, are shared by other ofﬁcers
within the same shift (e. g., there is a climate of distrust), then the effectiveness of such
practices are called into question because ofﬁcers are placed back into a context that
perpetuates a distrustful view towards citizens.

Organizational scholars have argued that practices aimed at the individual often
fail to consider how group level factors impact behavior and “tend to rely-on a model that
assumes that attitudes and behavior would be driven by individual level considerations”
(Alutto, 2002: 322). Police administrators should begin to explore how features of the
organizational context undermine current policies and practices. The proposed inquiry
begins to shed light on how group dynamics inﬂuence the occupational attitudes of
ofﬁcers, as well as how these attitudinal inﬂuences might impact ofﬁcers’ street-level

behaviors.

189

APPENDIX A

Table A.1: Attitudinal Survey Items

 

Variables*

 

Direct Supervisors (2 items)

My supervisor’s approach tends to discourage me from giving extra effort. (S1;
recoded)

My supervisor looks out for the personal welfare of his/her subordinates. (S2)

 

Top Management (3 items)

When an ofﬁcer does a particularly good job, TM will publicly recognize his/her
performance? (Tl)

When an ofﬁcer gets written up for minor violations of the rules, he/she will be treated
fairly by TM? (T2)

When an ofﬁcer contributes to a team effort rather than look good individually, TM
will recognize it? (T3)

 

Crime-ﬁghting

Enforcing the law is by far a patrol ofﬁcer’s most important responsibility. (R1;

recoded)

Order Maintenance (6 items)

Law enforcement ofﬁcers should be required to do something about...
Public nuisances (e.g., loud parties, barking dogs, etc.). (R2)
Neighbor disputes. (R3)

Family disputes. (R4)

Parents who don’t control their kids. (R5)
Litter and trash. (R6)

Nuisance businesses. (R7)

 

Selective Enforcement (1 item; recoded)
An ofﬁcer is more effective when she/he patrols for serious felony violations rather
than stopping people for minor trafﬁc violations and misdemeanors. (SE)

 

Aggressiveness (1 item; recoded)
A good police ofﬁcer is one who patrols aggressively by stopping cars, checking out
people, running license checks, and so forth. @GL

 

Citizen Distrust (2 items; recoded)
Police ofﬁcers have reason to be distrustful of most citizens. (C 1)
Police ofﬁcers have reason to be suspicious of most citizens. (C2)

 

 

Job Satisfaction (3 items; recoded)

I would not consider taking another job. (J 81)

I like my job better than the average police ofﬁcer does. (J 82)
I ﬁnd real enjoyment in my job. (JS3)

 

 

*Measures based on 4 point Likert scale ranging from 1 (Strongly Agree) to 4 (Strongly
Disagree).

190

Ammo“ cozﬂésﬁc _o>c_ cc. 2: 2w ammo. E EmocEwi Em Eon E 356580 ”202

 

 

_ hm. cv. vcr vc. mm. cur 3.- :6 :6 c—r :6 c—r cm. 2.. a: 2.- nmr 2“.- mm_.
_ cm. vc. cc. VN. mcr .3.- vcr vcr cc.- cc: cc.- Mn. 2: 3: cc: 3. :6 Nma
_ cc. cc. cN. mo: n—r :6 vor 2: cc.- c—.- c_. v~.- cm: c~.- —N.- _n.- _m_.
_ ch. cN. cm. cc. cm. cc. cc. c—. cc. c—. Ncr cc. .8.- cc. 8.. NU
_ mN. 3. cc. hc. vc. c—. cc. cc. cN. 2.. no; hcr mc. :6 _U
_ No. :.- mcr «or 3.- .8. cc.- 3.. cc: cc.- mo: 3: MN: O<
_ mo: 5. cc. \ic. cc. bc. c_. No. no. mcr 3. cc. mm
_ cm. cm. cm. mm. mm. mcr cc. NH. bc. cc. no. S—
c hm. cN. mm. cN. cc.- cc. 3. co. m—. 2. cm
_ cv. Nv. Ne. mcr N—. mm. cc. cc. c_. mm
c cc. mm. vcr co: co. cor cﬁ. c_. em
_ Nm. 3.- cc. 2. 8. no. cc. mm
_ cc.- 2. 2. cc. 2. 2. mm
_ 3: n—r :6 no: 2: :—
_ cc. cc. cn. cm. 2.
c em. cm. NV. .2-
_ cm. cm. C.
_ vc. mm
_ cm
mmm mmm 5.. ND _0 O< mm 5— cc mm em my— Nd :— 2. NH. :- Nm 5

 

2:»: metric .5535; nga 22:22.50 org—obem :5 ~35.

 

m X—QZHmA—<

191

Table C.1: Citizen Distrust

APPENDIX C

 

 

 

 

 

 

 

Index Score Frequency Percentage

2 71 7.0

3 76 7.4

4 295 28.9

5 147 14.4

6 291 28.5

7 65 6.4

8 63 6.2
Missing 14 1.4
Total 1,022 100%

Table C.2: Order Maintenance
Index Score Frequency Percengge

6 10 1.0

7 7 0.7

8 32 3.1

9 40 3.9

10 54 5.3

11 76 7.4

12 117 11.4

13 150 14.7

14 136 13.3

15 112 11.0

16 85 8.3

17 67 6.6

18 47 4.6

19 28 2.7
20 15 1.5
21 11 1.1
22 10 1.0
23 5 0.5
24 7 0.7
27 4 0.4
Missing 13 1.3
Total 1 ,022 100%

 

192

C-3: Direct Supervisors

 

 

 

 

 

 

 

 

 

 

 

Index Seere Frequency Percentage
2 287 28.1
3 192 18.8
4 275 26.9
5 145 14.2
6 69 6.8
7 22 2.2
8 22 2.2
Missing 10 1.0
Total 1,022 100%
C-4: Top Management
Index Score Frequency Percentage
3 15 1.5
4 16 1.6
5 25 2.5
6 133 13.0
7 132 12.9
8 145 14.2
9 190 18.6
10 135 13.2
11 102 10.0
12 106 10.4
Missing 23 2.3
Total 1,022 100%
C-5: Job Satisfaction
Index Score Frequency Percentage
3 6 0.6
4 9 0.9
5 14 1.4
6 51 5.0
7 86 8.4
8 176 17.2
9 237 23.2
10 173 16.9
11 161 15.8
12 88 8.6
Missing 21 2.1
Total 1 ,022 100%

 

193

C-6: Aggressiveness

 

 

 

 

 

 

 

 

 

 

 

Agreement Frequency Percentgg
[l] Disagree Strongly 28 2.7
[2] Disagree Somewhat 156 15.3
[3] Agree Somewhat 511 50.0
[4] Agree Strongly 321 31.4
Missing 6 0.6
Total 1,022 100%
C-7: Selective Enforcement

Agreement Frequency Percentage
[1] Disagree Strongly 291 28.4
[2] Disagree Somewhat 549 53.7
[3] Agree Somewhat 139 13.6
[4] Agree Strongly 30 2.9
Missing 13 1.3
Total 1,022 100%
C-8: Crime-ﬁghting

Agreement Frequency Percentage
[l] Disagree Strongly 31 3.0
[2] Disagree Somewhat 225 22.0
[3] Agree Somewhat 544 53.2
[4] Agree Strongly 217 21.2
Missing 5 0.5
Total 1,022 100%

 

194

Table D.l: rwg and r*wg0) Values by Workgroup

APPENDIX D

 

 

Group N Cit. Order Dir. Top Job Aggress. Select. Crime-
ID Dis. Maint. Sup. Man. Sat. ﬁght
1 19 0.49 0.61 0.58 0.54 0.62 0.64 0.34 0, 39
2 10 0.25 0.41 0.60 0.39 0.56 0.64 0.81 0.64
5 N/A N/A N/A N/A N/A N/A N/A N/A N/A
5 23 0.37 0.60 0.62 0.46 0.40 0.35 0.57 0,50
5 23 0.50 0.42 0.52 0.35 0.45 0.54 0.77 0,64
5 16 0.36 0.41 0.42 0.42 0.47 0.76 0.82 0,52
2 20 0.38 0.42 0.19 0.56 0.28 0.33 0.70 0,66
8 23 0.33 0.45 0.42 0.34 0.51 0.30 0.42 0,47
2 14 0.3 5 0.54 0.36 0.52 0.62 0.7 0 0.53 0,68
m N/A N/A N/A N/A N/A N/A N/A N/A N/A
_l_1 13 0.35 0.29 0.21 0.54 0.45 0.82 0.81 0,58
[2 15 0.35 0.28 0.05 0.42 0.57 0.66 0.83 0,56
Q 24 0.34 0.49 0.42 0.48 0.44 0.56 0.49 0,42
[3 23 0.56 0.40 0.60 0.51 0.75 0.73 0.61 0,59
l_5 l 5 0.36 0.23 0.50 0.45 0.48 0.56 0.29 0.95
m 19 0.36 0.61 0.44 0.34 0.28 0.65 0.66 0,79
l_7 1 7 0.65 0.52 0.63 0.47 0.58 0.51 0.71 0.29
18 l 3 0.29 0.57 0.72 0.53 0.46 0.68 0.82 0,67
12 9 0.62 0.73 0.27 0.58 0.52 0.83 0.77 0,10
20 N/A N/A N/A N/A N/A N/A N/A N/A N/ A
2_1 8 0.72 0.68 0.54 0.58 0.69 0.37 0.54 0,31
22 1 9 0.56 0.39 0.49 0.61 0.62 0.38 0.7 2 0,53
2 25 0.57 0.45 0.36 0.51 0.50 0.52 0.75 0.62
2_4 l 6 0.50 0. l 7 0.43 0.47 0.44 0.59 0.68 0,59
25 13 0.56 0.33 0.08 0.48 0.40 0.75 0.00 0,71
2_6 20 0.63 0.54 0.63 0.47 0.44 0.44 0.74 0,42
2_7 N/A N/A N/A N/A N/A N/A N/A N/A N/A
2_8 1 3 0.45 0.68 0.00 0.49 0.49 0.27 0.71 0,40
2_9 1 7 0.56 0.43 0.47 0.34 0.56 0.59 0.52 0.31
50 N/A N/A N/A N/A N/A N/A N/A N/A N/A
3_1 N/A N/A N/A N/A N/A N/A N/A N/A N/A
3 16 0.37 0.32 0.27 0.51 0.23 0.63 0.42 0.68
55 2 1 0.32 0.35 0.70 0.30 0.66 0.72 0.59 0,50
35 8 0.21 0.61 0.55 0.50 0.57 0.56 0.54 0,67
55 23 0.25 0.55 0.62 0.47 0.59 0.59 0.49 0,60
56 2 l 0.62 0.61 0.55 0.47 0.49 0.85 0.56 0.61
51 l 7 0.42 0.58 0.62 0.14 0.57 0.39 0.29 0,61
ﬂ 7 0.66 0.65 0.66 0.57 0.70 0.50 0.62 0,62
52 19 0.54 0.52 0.73 0.48 0.49 0.81 0.03 0,52
510 1 7 0.35 0.51 0.53 0.66 0.50 0.85 0.45 0,68
55 15 0.42 0.54 0.47 0.28 0.64 0.81 0.43 0.60
g 5 0.56 0.65 0.72 0.73 0.47 0.44 0.44 0.76

 

195

Table D.1: rwg and r*wg0) Values (cont.)

 

 

 

 

Group N Cit. Order Dir. Top Job A ggress. Select. Crime-
ID Dis. Maint. Sgp. Man. Sat. ﬁght
35 24 0.49 0.56 0.54 0.28 0.64 0.66 0.54 0. 17
313 24 0.61 0.37 0.46 0.30 0.39 0.45 0.41 0.32
_4_5 14 0.33 0.70 0.55 0.62 0.62 0.49 0.82 0.51
E 5 0.08 0.29 0.44 0.60 0.63 0.76 0.76 0.60
4_11 25 0.48 0.44 0.63 0.42 0.69 0.47 0.50 0.25
4_8_ 15 0.70 0.60 0.48 0.67 0.69 0.49 0.88 0.68
Q 14 0.47 0.41 0.22 0.52 0.55 0.43 0.70 0.67
E l 7 0.44 0.60 0.25 0.73 0.45 0.35 0.62 0.79
5_1 1 7 0.43 0.49 0.24 0.27 0.48 0.49 0.51 0.49
52 20 0.49 0.54 0.44 0.70 0.63 0.64 0.67 0.79
_55 20 0.58 0.59 0.50 0.67 0.50 0.67 0.88 0.71
_11 1 1 0.32 0.65 0.35 0.64 0.49 0.16 0.45 0.78
5_ 1 5 0.48 0.74 0.56 0.50 0.69 0.47 0.71 0.72
56 10 0.45 0.45 0.64 0.30 0.49 0.63 0.74 0.60
ﬂ 1 2 0.56 0.39 0.32 0.70 0.50 0.50 0.66 0.81
58 l 7 0.80 0.45 0.25 0.53 0.47 0.55 0.50 0.25
_9 l 4 0.71 0.46 0.63 0.68 0.60 0.68 0.73 0.79
6_ 24 0.60 0.45 0.31 0.48 0.29 0.68 0.40 0.46
61 22 0.43 0.55 0.64 0.51 0.58 0.72 0.32 0.47
62 l 7 0.41 0.48 0.71 0.40 0.61 0.79 0.58 0.19
65 l l 0.58 0.70 0.56 0.78 0.60 0.68 0.82 0.64
641 l 6 0.51 0.59 0.43 0.51 0.68 0.69 0.52 0.68
Q l 5 0.27 0.55 0.88 0.49 0.64 0.67 0.72 0.44
66 8 0.23 0.32 0.78 0.50 0.63 0.56 0.77 0.56
61 7 0.00 0.57 0.85 0.62 0.24 0.77 0.79 0.55

Mean 16.2 0.45 0.50 0.49 0.49 0.53 0.59 0.60 0.58

Median 0.46 0.52 0.50 0.50 0.54 0.61 0.62 0.61

 

Values in bold indicate statistical signiﬁcance at the .05 level or greater.

196

Table D.2: ADM and ADM”) Values

 

 

Group N Cit. Order Dir. Top Job A ggress. Select. Crime-
ID Dis. Maint. Sup. Man. Sat. ﬁght
1 19 0.56 0.53 0.53 0.63 0,53 0,53 0.70 0.70
2 10 0.82 0.67 0-49 0.64 0.52 0.56 0.42 0.56
3 N/A N/A N/A N/A N/A N/A N/A N/A N/A
4. 23 0.73 0.54 0.53 0.70 0.69 0.71 0.56 0.56
2 23 0.67 0.70 0.62 0.72 0.64 0.53 0.45 0.48
g 16 0.71 0-67 0.65 0.69 0.66 0.41 0.43 0.53
2 20 0.74 0.66 0.77 0.56 0.75 0.73 0.48 0.52
3 23 0.77 0.64 0.62 0.77 0.60 0.72 0.65 0.62
2 14 0.74 0.58 0.64 0.63 0.49 0.51 0.61 0.55
19 N/A N/A N/A N/A N/A N/A N/A N/A N/A
H 13 0.74 0.77 0.78 0.62 0.65 0.43 0.28 0.59
D 15 0.68 0.78 0.90 0.69 0.55 0.40 0.25 0.46
D 24 0.72 0.65 0.63 0.66 0.68 0.58 0.52 0.67
L4. 23 0.63 0.70 0.45 0.62 0.44 0.43 0.45 0.57
D 15 0-71 0-80 0.63 0.66 0.64 0.64 0.75 0.12
is 19 0.64 0.54 0.64 0.75 0.74 0.47 0.54 0.37
11 17 0.54 0.63 0.59 0.69 0.51 0.62 0.53 0.79
is 13 0.76 0.55 0.54 0.62 0.55 0.53 0.43 0.43
B 9 0.49 0-42 0.75 0.58 0.58 0.37 0.50 0.87
29 N/A N/A N/A N/A N/A N/A N/A N/A N/A
2t 8 0.42 0.49 0.50 0.54 0.44 0.62 0.50 0.75
2 19 0.62 0.71 0.59 0.55 0.54 0.63 0.49 0.65
a 25 0.59 0-66 0.60 0.58 0.67 0.54 0.42 0.59
2_4 16 0.63 0.84 0.66 0.64 0.67 0.62 0.56 0.62
2_5_ 13 0.62 0.73 0.92 0.68 0.68 0.39 0.99 0.47
26 20 0.42 0.63 0.44 0.70 0.66 0.62 0.49 0.64
2_7 N/A N/A N/A N/A N/A N/A N/A N/A N/A
E 13 0.67 0.51 0.99 0.63 0.51 0.72 0.39 0.72
2 17 0.61 0.66 0.62 0.74 0.58 0.55 0.59 0.70
3_0 N/A N/A N/A N/A N/A N/A N/A N/A N/A
3_l N/A N/A N/A N/A N/A N/A N/A N/A N/A
3_2 16 0.74 0-73 0.81 0.65 0.78 0.47 0.59 0.37
3_3 21 0.70 0.73 0.51 0.76 0.54 0.51 0.61 0.57
33 8 0.81 0.53 0.59 0.54 0.58 0.62 0.50 0.44
3_5 23 0.77 0.61 0.59 0.65 0.59 0.50 0.61 0.50
3Q 21 0.59 0-53 0.62 0.68 0.64 0.26 0.59 0.44
3_7 17 0.70 0.56 0.59 0.90 0.59 0.75 0.79 0.52
3_8 7 0.55 0.49 0.55 0.58 0.52 0.65 0.49 0.49
3_9 19 0.64 0.60 0.46 0.62 0.62 0.44 0.96 0.65
E 17 0.75 0.59 0.45 0.45 0.59 0.36 0.56 0.48
4_1 15 0.70 0-62 0.64 0.78 0.50 0.44 0.53 0.50
Q 5 0.60 0.47 0.32 0.40 0.53 0.64 0.64 0.48

 

197

Table D.2: ADM and ADM”) Values (cont.)

 

 

 

 

Group N Cit. Order Dir. Top Job Aggress. Select. Crime-
ID Dis. Maint. Sup. Man. Sat. ﬁght
4_3 24 0.68 0.57 0.60 0.75 0.52 0.58 0.56 0.56
4_4 24 0.58 0.69 0.67 0.76 0.71 0.68 0.64 0.75
£5 14 0.73 0.46 0.65 0.52 0.58 0.56 0.26 0.29
g 5 0.84 0.70 0.64 0.51 0.47 0.48 0.48 0.72
12 25 0.67 0.67 0.60 0.73 0.48 0.59 0.69 0.64
g 1 5 0.53 0.56 0.58 0.54 0.45 0.62 0.14 0.62
_4_9_ 14 0.62 0.68 0.71 0.61 0.58 0.69 0.51 0.56
59 l 7 0.71 0.54 0.73 0.53 0.57 0.69 0.58 0.50
51_ l 7 0.69 0.61 0.80 0.80 0.64 0.67 0.52 0.48
g 20 0.68 0.61 0.66 0.49 0.53 0.51 0.45 0.28
55 20 0.63 0.60 0.58 0.53 0.64 0.45 0.19 0.48
5_4 l 1 0.76 0.49 0.67 0.58 0.61 0.88 0.51 0.33
55 l 5 0.63 0.43 0.51 0.60 0.48 0.67 0.49 0.25
56 10 0.68 0.65 0.56 0.73 0.5 8 0.56 0.36 0.48
ﬂ 1 2 0.50 0.64 0.72 0.44 0.61 0.47 0.56 0.75
58 1 7 0.48 0.68 0.77 0.60 0.61 0.45 0.47 0.48
2 l 4 0.42 0.63 0.53 0.52 0.52 0.55 0.45 0.49
@ 24 0.57 0.67 0.64 0.67 0.74 0.56 0.65 0.69
6_1 22 0.65 0.58 0.59 0.62 0.57 0.54 0.73 0.47
6_2 1 7 0.66 0.60 0.52 0.72 0.59 0.48 0.50 0.55
Q l l 0.48 0.37 0.58 0.41 0.56 0.36 0.40 0.45
65 l 6 0.62 0.52 0.56 0.61 0.47 0.55 0.66 0.55
Q 1 5 0.75 0.56 0.28 0.54 0.57 0.57 0.37 0.43
66 8 0.75 0.72 0.48 0.62 0.53 0.62 0.50 0.62
6_2 7 0.83 0.53 0.33 0.54 0.62 0.49 0.44 0.49

Mean 16.2 0.65 0.61 0.61 0.63 0.58 0.56 0.52 0.54

Median 0.63 0.61 0.59 0.62 0.58 0.56 0.51 0.53

 

Values in bold indicate statistical signiﬁcance at the .05 level or greater; Even if values are statistically
signiﬁcant, values that are above 0.67 are not considered to be of practical signiﬁcance. Also, non-bold

values that are 0.67 or below are considered to be practically signiﬁcant even though they are not

statistically signiﬁcant.

198

.wEEcu-oEtU H m0 .mmo:o>_82cw< H O< dozen-«88m no.- u
m.- .uoﬂtoczm 525 n ma 6285582 520 H 20 .85me 5850 H CO .38: 3:353: 65. mc. 65 8 “82 8 cgomﬁwmm v.8 Eon E 355580 “652

 

 

_ 8.- 8.- 8.- 8.- 8.- 8.- 8. 8. 8.- 8.- 8. 8. 8. 8. 8. 8.- 8. 8. PE
_ 8.- 8.- 8.- 2.- 8. 2. 8. 8.- 8. 8. 8. 8.- 8.- 8.- 8. 8.- 8.- 9::
_ 8.- 8.- 8.- 8.- 8.- 2.- 8.- 8.- 8.- 8.- 8. 8. 8. 8.- 8.- 8. 92
_ 8.- 8. 8. 8. 8.- _m. :. 2. 8.- 8.- 8. 8.- 8. 8. 8.- m8
_ 8. 8.. 8. 8. 8. 8. 8. 8. 8.- 8.- 8.- :. :. 8.- 980
.85
_ 8.... 8. 8.- 8. 8. 8. 8.- 5.- 8.- 8.- 8. 8. 8.- MN
_ 8. 8. 8. 8. 8. 8. 2.. 8.- 8.- 8. 8. 8.- .w><
833
_ 8. 8. 8. 8. 8. 8.- 8.- 2.- 8. 8. 8.- .28
2 8.- 8. 8. 8. 8. 8 8. 8.- 8.- 8. 6:82.68
_ 8. 8. 8.- 8.- 8.- 8.- 2. 8. 8.- .83
_ 8. 8.- 8.- 8.- 8.- 8. 8. 8.- €me
_ 8. 8. 8. 8.- 8. 8. 8. 253
_ 8. 8 8. 8. 8.- 8.- 2e:
_ 8. 8. 8.- 8.- 8. 5
_ 8. 8.- 8.- 8. o<
_ 8.- 2.- 8. 8
_ S. 8.- mo
_ 8. 20
_ no
.88 .98 823 6.32 ca
Ex ESE 9:. mad Emu .m>< .82 .68 .68 2.3 8m 8:; 2e: .5 3 8 mo 20 no

 

555950 circa—.53 4c .mzscuuwcagn £2553. hoe—EC w5=5 85:22.50 "—.m «Ea-_-

 

m XuQZm—mmd.

199

REFERENCES

Ajzen, I. & F ishbein, M. (1977). Attitude-behavior relations: A theoretical analysis and
review of empirical research. Psychological Bulletin, 84(5), 888-918.

Alutto, J. (2002). Culture, levels of analysis, and cultural transition. In F. Yammarino
and F. Dansereau (Eds.). Research in multi-level issues volume 1: The many faces
of multilevel issues (pp. 321-325). New York: JAI.

Barker, J. (1999). Danger, duty, and disillusion: The worldview of Los Angeles police
officers. Prospect Heights, IL: Waveland Press.

Bayley, D. & Bittner, E. (1984). Learning the skills of policing. Law and Contemporary '
Problems, 47(4), 35-59.

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

Bittner, E. (1970). T he functions of police in modern society: A review of background
factors, current practices, and possible role models. Chevy Chase, MD:
National Institute of Mental Health Center for Studies of Crime and Delinquency.

Bliese, P. (2000). Within-group agreement, non-independence, and reliability:
Implications for data aggregation and analysis. In K. Klein and S. Kozlowski
(Eds) Multilevel theory, research, and methods in organizations: Foundations,
extensions, and new directions (pp. 349-381). San Francisco: Jossey-Bass.

Bliese, P. (1998). Group size, ICC values, and group-level correlations: A simulation.
Organizational Research Methods, 1(4), 355-373.

Bliese, P., Chan, D., & Ployhart, R. (2007). Multilevel models: Future directions in
measurement, longitudinal analyses, and nonnonnal outcomes. Organizational
Research Methods, 10(4), 551-563.

Bliese, P. & Hanges, P. (2004). Being both to liberal and too conservative: The perils of
treating grouped data as though they were independent. Organizational Research
Methods, 7(4), 400-417.

Brehm, l. & Gates, S. (1993). Donut shops and speed traps: Evaluating models of
supervision on police behavior. American Journal of Political Science, 37(2),

555-581.

Broderick, J. (1977). Police in a time of change. Morristown, NJ: General Learning
Press.

200

Brooks, L., Piquero, A., & Cronin, J. (1994). Work-load rates and police ofﬁcer attitudes:
An examination of busy and slow precincts. Journal of Criminal Justice, 22(3),
277-286.

Brooks, L., Piquero, A., & Cronin, J. (1993). Police ofﬁcer attitudes concerning their
communities and their roles: A comparison of two suburban police departments.
American Journal of Police, 12(3), 115-140.

Brown, M. (1988). Working the street: Police discretion and the dilemmas of reform.
New York: Russell Sage Foundation.

Brown, R. & Hauenstein, N. (2005). Interrater agreement reconsidered: An alternative to
the rwg indices. Organizational Research Methods, 8(2), 165-184.

Burke, M., & Dunlap, W. (2002). Estimating interrater agreement with the average
deviation index: A user’s guide. Organizational Research Methods, 5(2), 159
172.

Burke, M., F inkelstein, L., & Dusig, M. (1999). On average deviation indices for
estimating interrater agreement. Organizational Research Methods, 2(1), 49-68.

Bursik, R., & Grasmick, H. (1996). The use of contextual analysis in models of criminal
behavior. In J .D. Hawkins (Ed.) Delinquency and crime: Current theories (pp.
237-267). New York: Cambridge University Press.

Carlan, P. (2007). The search for job satisfaction: A survey of Alabama policing.
American Journal of Criminal Justice, 32, 74-86.

Cochran, J. & Bromley, M. (2003). The myth(?) of the police subculture. Policing.
26(1), 88-117.

Cohen, A., Doveh, 13., & Nahum-Shani, I. (2009). Testing agreement for multi-item
scales with the indices rwg(j) and ADmG). Organizational Research Methods,
12(1),148-164.

Crank, J. (2004). Understanding police culture 2'"1 ed. Cincinnati: Anderson Publishing.

Dantzker, M. & Kubin, B. (1998). Job satisfaction: The gender perspective among police
ofﬁcers. American Journal of Criminal Justice, 23(1), 19-31.

Davey, J ., Obst, P., & Sheehan, M. (2001). Demographic and workplace characteristics
which add to the prediction of stress and job satisfaction within the police

workplace. Journal of Police and Criminal Psychology, 16(1), 29-39.

DeJong, C. (2004). Gender differences in ofﬁcer attitudes and behavior: Providing
comfort to citizens. Women and Criminal Justice, 15(3/4), 1-32.

201

DeJong, C., Mastrofski, S., & Parks, R. (2001). Patrol ofﬁcers and problem solving: An
application of expectancy theory. Justice Quarterly, 18(1), 31-61.

Drummond, D.S. (1976). Police culture. Beverly Hills: Sage.

Dunlap, W., Burke, M., & Smith-Crowe, K. (2003). Accurate tests of statistical
signiﬁcance for rwg and average deviation interrater agreement indexes. Journal of
Applied Psychology, 88(2), 356-362.

Enders, C. & Toﬁghi, D. (2007). Centering predictor variables in cross-sectional
multilevel models: A new look at an old issue. Psychological Methods, 12(2),
121-138.

Engel, R. (2000). Supervisory styles of patrol sergeants and lieutenants. Journal of
Criminal Justice, 29, 341-355.

Engel, R. & Worden, R. (2003). Police ofﬁcers’ attitudes, behavior, and supervisor
inﬂuences: An analysis of problem solving. Criminology, 41(1), 131-166.

Fielding, N. (1988). Joining forces: Police training, socialization, & occupational
competence. New York: Rutledge.

Frank, J. & Brandl, S. (1991). The police attitude-behavior relationship methodological
and conceptual considerations. American Journal of Police, 10(4), 83-103.

Friedrich, R]. (1980). Police use of force: Individuals, situations, and organizations.
Annals of the American Academy of Political and Social Science, 452, 82-97.

Goldstein, H. (1977). Policing a free society. Cambridge, MA: Ballinger.

Gould, J. & Mastrofski, S. (2004). Suspect searches: Assessing police behavior under the
US. Constitution. Criminology and Public Policy, 3(3), 315-361.

Guo, G. & Zhao, H. (2000). Multilevel modeling for binary data. Annual Review of
Sociology, 26, 441-462.

Haarr, R. (2001). The making of a community policing ofﬁcer: The impact of basic
training and occupational socialization on police recruits. Police Quarterly, 4(4),
402-433.

Haarr, R. (1997). Patterns of interaction in a police patrol bureau: Race and gender
barriers to integration. Justice Quarterly, 14(1), 53—85.

202

Hackman, J .R. (1992). Group inﬂuences on individuals in organizations. In M.D.
Dunnette & L.M. Hough (Eds.) Handbook of industrial and organizational
psychology (V0. 3, pp. 199-267). Palo Alto, CA: Consulting Psychologist Press.

Harris, DA. (2002). Proﬁles in injustice: Why racial proﬁling cannot work. New York:
The New Press.

Hassell, K.D. (2006). Police organizational cultures and patrol practices. New York:
LFB Scholarly Publishing.

Hauser, R. (1970). Context and consex: A cautionary tale. The American Journal of
Sociology, 75(4), 645-664.

Hedeker, D. & Gibbons, R. (2006). Longitudinal data analysis. Hoboken, NJ: John
Wiley.

Herbert, S. (1998). Police subculture reconsidered. Criminology, 36(2), 343-369.

Hochstedler, E. (1981). Testing types: A review and test of police types. Journal of
Criminal Justice, 9, 451-466.

Hoffman, D.A. & Gavin, M.B. ( I998). Centering decisions in hierarchical linear models:
Implications for research in organizations. Journal of Management, 24(5), 623
641)

Hoffman, D.A., Grifﬁn, M.A., & Gavin, MB. (2000). The application of hierarchical
linear modeling to organizational research. In K. Klein & S. Kozlowski (eds.)
Multilevel theory, research, and methods in organization: Foundations,
extensions, and new directions (pp. 467-511). San Francisco: Jossey-Bass.

Hoffman, D.A. & Stetzer, A. (1996). A cross-level investigation of factors inﬂuencing
unsafe behaviors and accidents. Personnel Psychology, 49, 307-339.

Hunt, R. & McCadden, K. (1985). A survey of work attitudes of police ofﬁcers:
Commitment and satisfaction. Police Studies, 8, 17-25.

James, L. (1982). Aggregation bias in estimates of perceptual agreement. Journal of
Applied Psychology. 67(2), 219-229.

James, L.R., Demaree, R.G. & Wolf, G. (1993). ng: An assessment of within-group
interrater agreement. Journal of Applied Psychology, 78(2), 306-309.

James, L.R., Demaree, R.G., & Wolf, G. (1984). Estimating within-group interrater
reliability with and without response bias. Journal of Applied Psychology, 69(1),
85-98.

Jermier, J ., Slocum, J ., Fry, L. & Gaines, J. (1991). Organizational subcultures in a

203

soft bureaucracy: Resistance behind the myth and facade of an ofﬁcial culture.
Organization Science. 2(2), 170-194.

Katz, D. & Kahn, R. (1978). The social psychology of organizations (2nd ed.). New York:
John Wiley.

Kenny, D. & Judd, C. (1986). Consequences for violating the independence assumption
in analysis of variance. Psychological Bulletin, 99(3), 422-431.

Klein, K.J., Bliese, P., Kozlowski, S.W.J., Dansereau, F., Gavin, M., Grifﬁn, M. & et a1.
(2000). Multilevel analytical techniques: Commonalities, differences, and
continuing questions. In K. Klein & S. Kozlowski (eds.) Multilevel theory,
research, and methods in organization: Foundations, extensions, and new
directions (pp. 512-553). San Francisco: Jossey-Bass.

Klein, K.J., Dansereau, F ., & Hall, R.J. (1994). Levels issues in theory development, data
collection, and analysis. Academy of Management Review, 19(2), 195-229.

Klinger, D. (2004). Environment and organization: Reviving a perspective on the police.
Annals of the American Academy of Political and Social Science, 593,119-136.

Klinger, D. (1997). Negotiating order in patrol work: An ecological theory of police
response to deviance. Criminology. 35(2). 277-307.

Kline, R. (1998). Principles and practice of structural equation modeling. New York:
Guilford Press.

Kozlowski, S.W.J. & Bell, BS. (2003). Work groups and teams in organizations. In W.C.
Borman, D.R. Ilgen, & R.J. Kilmoski (Eds.) Handbook of psychology: Industrial
and organizational psychology (Vol. 12, pp. 333-375). London: Wiley.

Kozlowski, S. & Hattrup, K. (1992). A disagreement about within-group agreement:
Disentangling issues of consistency versus consensus. Journal of Applied
Psychology, 77(2), 161-167.

Kozlowski, S. & Klein, K]. (2000). A multilevel approach to theory and research in
organizations: Contextual, temporal, and emergent processes. In K. Klein & S.
Kozlowski (eds.) Multilevel theory, research, and methods in organization:
Foundations, extensions, and new directions (pp. 3-90). San Francisco: Jossey-
Bass.

Lindell, M., Brandt, C., & Whitney, D. (1999). A revised index of interrater agreement

for multi-item ratings of a single target. Applied Psychological Measurement, 23,
127-135.

204

Lord, V. & Friday, P. (2008). What really inﬂuences ofﬁcer attitudes toward COP?: The
importance of context. Police Quarterly, 11(2), 220-238.

Louis, MR. (1985). An investigator’s guide to workplace culture. In P. Frost, L. Moore,
M.R. Louis, C. Lundberg, & J. Martin (Eds). Organizational culture (pp. 73-93).
Beverly Hills: Sage.

Maas, C. & Hox, J. (2005). Sufﬁcient sample sizes for multilevel modeling.
Methodology, 1(3), 86-92.

Manning, PK. (2005). The study of policing. Police Quarterly. 8(1), 23-44.

Manning, PK. (1997). Police work: The social organization of policing (2"d Ed.).
Prospect Heights, IL: Waveland Press.

Manning, PK. (1995). Occupational culture. In W. Bailey (Ed.), The encyclopedia of
police science (pp. 472-475). New York: Garland Publishing.

Mastrofski, S. (2004). Controlling street-level police discretion. Annals of the American
Academy of Political and Social Science, 593, 100-118.

Mastrofski, S., Snipes, J., Parks, R., & Maxwell, C. (2002). The helping hand of the law:
Police control of citizens on request. Criminology, 38(2), 307-342.

Mastrofski, S., Ritti, R., & Snipes, J. (1994). Expectancy theory and police productivity
in DUI enforcement. Law & Society Review, 28(1), 113-148.

Mastrofski, S., Worden, R., & Snipes, J. (1995). Law enforcement in a time of
community policing. Criminology, 33(4), 539-563.

McCluskey, J ., Terrill, W., & Paoline, E. (2005). Peer group aggressiveness and the use
of coercion in police-suspect encounters. Police practice and research, 6(1), 19-
37.

Moon, B. & Zager, L. (2007). Police ofﬁcers’ attitudes toward citizen support: Focus on
individual, organizational, and neighborhood characteristic factors. Policing,
30(3), 484-497.

Moskos, P. (2008). Cop in the hood. Princeton, NJ: Princeton University Press.

Muir, W. K. (1977). Police: Streetcorner politicians. Chicago: University of Chicago
Press.

Muthen, L. & Muthen, B0. (2007). Mplus user ’s guide. 5th ed. Los Angeles, CA:
Muthén & Muthén.

205

National Research Council. (2004). Fairness and eﬂectiveness in policing: The evidence.
Committee to Review Research on Police Policy and Practices. W. Skogan and K.
Frydl (Eds) Washington, DC: The National Academies Press.

Nauman, S. & Bennett, N. (2000). A case for procedural justice climate: Development
and test of a multilevel model. Academy of Management Journal, 43(5), 881-889.

Novak, K., Smith, B., & Frank, J. (2003). Strange bedfellows: Civil liability and
aggressive policing. Policing, 26(2), 352-368.

Ostroff, C., Kinicki, A., & Tamkins, M. (2003). Organizational culture and climate. In
W.C. Borgman, D.R. Ilgen, & R.J. Klimoski (Eds) Comprehensive Handbook of
Psychology. I/O Psychology (Vol. 12, pp. 565-593). New York: Wiley.

Paoline, E. (2004). Shedding light on police culture: An examination of ofﬁcers’
occupational attitudes. Police Quarterly. 7(2), 205—236.

Paoline, E. (2003). Taking stock: Toward a richer understanding of police culture.
Journal of Criminal Justice, 31, 199-214.

Paoline, E. (2001). Rethinking police culture: Officers ’ occupational attitudes. New
York: LFB Scholarly Publishing.

Paoline, E., Myers, 8., & Worden, R. (2000). Police culture, individualism, and
community policing: Evidence from two police departments. Justice Quarterly,
17(3), 575-605.

Paoline, E., & Terrill, W. (2005). The impact of police culture on trafﬁc stop searches:
An analysis of attitudes and behavior. Policing, 28(3), 455-472.

Pogrebin, M.R., & Poole, ED. (1988). Humor in the brieﬁng room: A study of the
strategic uses of humor among police. Journal of Contemporary Ethnography.
17(2), 183-210.

Raudenbush, S. & Bryk, A. (2002). Hierarchical linear models: Applications and data
analysis methods. 2nd ed. Thousand Oaks: Sage.

Reiss, A. (1992). Police organization in the twentieth century. Crime and Justice, 15, 51
97.

Reiss, A. (1967). Career orientations, job satisfaction, and the assessment of law
enforcement problems by police ofﬁcers. Field Surveys III: Studies in Crime and

Law Enforcement in Major Metropolitan Areas, 2, 1-123.

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

206

Riksheim, E. & Chermak, S. (1993). Causes of police behavior revisited. Journal of
Criminal Justice, 21, 353-382.

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

Sampson, R.J., Raudenbush, S., & Earls, F. (1997). Neighborhoods and violent crime: A
multilevel study of collective efﬁcacy. Science, 277, 918-924.

Schafer, J. (2002). “I’m not against it in theory. . .”: Global and speciﬁc community
policing attitudes. Policing, 25(4), 669-686.

Schneider, B. (1981). Work climates: An interactionist perspective. In N. Feimer & E.
Geller (eds) Environmental psychology: Directions and perspectives (pp. 401
436. New York: Praeger.

Schuman, H. & Johnson, M. (1976). Attitudes and behavior. Annual Review of Sociology,
2, 161-207.

Sharp, E. (1982). Street-level discretion in policing: Attitudes and behaviors in the
deprofessionalization syndrome. Law and Policy Quarterly, 4(2), 167-189.

Sherman, L. (1980). Causes of police behavior: The current state of quantitative research.
Journal of Research in Crime and Delinquency, 69-100.

Shrout, P. & F leiss, J. (1979). Intraclass correlations: Uses in assessing rater reliability.
Psychological Bulletin, 86, 420-428.

Skogan, W. (2008). Why reforms fail. Policing & Society, 18(1), 23-34.

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

Skolnick, J. (1967). Justice without trial: Law enforcement in democratic society. New
York: John Wiley & Sons.

Smith, B, Novak, K., Frank, J ., & Lownkamp, C. (2005). Explaining police ofﬁcer
discretionary activity. Criminal Justice Review, 30(3), 325-346.

Smith, D. (1986). The neighborhood context of police behavior. In Albert J. Reiss
and Michael Tonry (Eds) Communities and crime (pp. 313-341). Chicago:

University of Chicago Press.

Smith, D. & Klein, J. (1984). Police control of interpersonal disputes. Social Problems,
31(4), 468-481.

207

Smith, D. & Klein, J. (1983). Police agency characteristics and arrest decisions. In G.P.
Whitaker and CD. Phillips (Eds) Evaluating performance of criminal justice
agencies (Vol. 19, pp. 63-96). Beverly Hills, CA: Sage.

Snipes, J. & Mastrofski, S. (1990). An empirical test of Muir’s typology of police
ofﬁcers American Journal of Criminal Justice, 9(2), 268-296.

Sun, I. (2003). Police ofﬁcers’ attitudes toward their role and work: A comparison of
black and white ofﬁcers. American Journal of Criminal Justice, 28(1), 90-108.

Sun, I. (2002). Police ofﬁcers attitudes toward peers, supervisors, and citizens: A
comparison between ﬁeld training ofﬁcers and regular ofﬁcers. American Journal
of Criminal Justice, 27(1), 69-83.

Terrill, W. (2001). Police coercion: Application of the force continuum. New York:
LF B Scholarly Publishing.

Terrill, W. & Mastrofski, S. (2002). Situational and ofﬁcer-based determinants of police
coercion. Justice Quarterly, 19(2), 215-248.

Terrill, W. & Paoline, E. (2009). Police use of force practices and policies: A multi-city
examination. Paper presented at the annual meeting of the American Society of
Criminology, Philadelphia.

Terrill, W., Paoline, E., & Manning, P. (2003). Police culture and coercion. Criminology,
41(4), 1003-1034.

Terrill, W. & Reisig, M. (2003). Neighborhood context and police use of force. Journal
of Research in Crime and Delinquency. 40(3), 291-321.

Van Maanen, J. (1975). Police socialization: A longitudinal examination of job attitudes
in an urban police department. Administrative Science Quarterly, 20(2), 207-228.

Van Maanen, J. (1974). Working the street: A developmental view of police behavior.
In H. Jacob (Ed.) T he potential for reform of criminal justice (pp. 83-130).
Beverly Hills: Sage.

Varriale, J ., Gibbs, J., Ahlin, E., Gugino, M., & Na, C. (2007). Trends in police research:
A cross-sectional analysis of the 2004 literature. Police Practice and Research,
8(5), 461-485.

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

White, S. (1972). A perspective on police professionalization. Law and Society Review, 7,
61-85.

208

Wilson, J .Q. (1968). Varieties of police behavior: The management of law and order in
eight communities. Cambridge: Harvard University Press.

Wilson, J .Q. & Kelling, G. (1982). The police and neighborhood safety. The Atlantic, 29
38.

Worden, AP. (1993). The attitudes of women and men in policing: Testing conventional
and contemporary wisdom. Criminology, 31(2), 203-241.

Worden, R. (1996). The causes of police brutality: Theory and evidence on police use of
force. In W. Geller and H. Toch (Eds) Police violence: Understanding and
controlling police abuse of force (pp. 23-48). New Haven: Yale University Press.

Worden, R. (1995). Police ofﬁcers’ belief systems: A framework for analysis.
American Journal of Police, 14(1), 49-81.

Worden, R. (1990). A badge and a baccalaureate: Policies, hypotheses, and further
evidence. Justice Quarterly, 7(3), 565-592.

Worden, R. (1989). Situational and attitudinal explanations of police behavior: A
theoretical reappraisal and empirical assessment. Law & Society Review, 23(4),
667-711.

Yu, C .Y. (2002). Evaluating cut-off criteria of model ﬁt indices for latent variable
models with binary and continuous outcomes. Unpublished dissertation. Los
Angeles, CA: University of California, Los Angeles.

Zhao, J ., Thurman, Q., & He, N. (1999). Sources of job satisfaction among police
ofﬁcers: A test of demographic and work environment models. Justice Quarterly,
16(1),153-173.

Zohar, D. (2000). A group-level model of safety climate: Testing the effect of group
climate on microaccidents in manufacturing jobs Journal of Applied Psychology,
85(4), 587-596).

Zohar, D. & Luria, G. (2005). A multilevel model of safety climate: Cross-level

relationships between organization and group-level climates. Journal of Applied
Psychology, 90(4), 616-628.

209

      

           

ll“1111111111lllﬂlllﬂllllllll

03063

11111

3

11111