DATE DOWNLOADED: Sat Jul 11 15:30:16 2020

SOURCE: Content Downloaded from HeinOnline

Citations:

Bluebook 20th ed.

William Terrill, Eugene A. III Paoline & Peter K. Manning, Police Culture and
Coercion, 41 Criminology 1003 (2003).

ALWD 6th ed.

William Terrill, Eugene A. III Paoline & Peter K. Manning, Police Culture and
Coercion, 41 Criminology 1003 (2003).

APA 7th ed.

Terrill, W., Paoline, E., & Manning, P. K. (2003). Police culture and coercion.
Criminology, 41(4), 1003-1034.

Chicago 7th ed.

William Terrill; Eugene A. III Paoline; Peter K. Manning, "Police Culture and
Coercion," Criminology 41, no. 4 (November 2003): 1003-1034

McGill Guide 9th ed.

William Terrill, Eugene A III Paoline & Peter K Manning, "Police Culture and
Coercion" (2003) 41:4 Criminology 1003.

MLA 8th ed.

Terrill, William, et al. "Police Culture and Coercion." Criminology, vol. 41, no. 4,
November 2003, p. 1003-1034. HeinOnline.

OSCOLA 4th ed.

William Terrill and Eugene A III Paoline and Peter K Manning, 'Police Culture and
Coercion' (2003) 41 Criminology 1003

-- Your use of this HeinOnline PDF indicates your acceptance of HeinOnline's Terms and
Conditions of the license agreement available at
https://heinonline.org/HOL/License
-- The search text of this PDF is generated from uncorrected OCR text.
-- To obtain permission to use this article beyond the scope of your license, please use:
Copyright Information
POLICE CULTURE AND COERCION*

WILLIAM TERRILL
Northeastern University

EUGENE A. PAOLINE III

University of Central Florida

PETER K. MANNING
Northeastern University

Researchers have long noted the link between police culture and
coercion. To date, however, there have been no empirical studies of
this relationship. Using data collected as part of a systematic social
observation study of the police in Indianapolis,Indiana, and St. Peters-
burg, Florida, this research examines the relationship between tradi-
tional views of police culture-from an attitudinal perspective-and
coercion-from a behavioralperspective. After developing a classifi-
cation scheme of officers' outlooks in the context of police culture, we
examine the extent to which officers' alignment with cultural attitudes
translates into differences in coercive behavior. The findings indicate
that those officers who closely embody the values of the police culture
are more coercive compared with those that differentially align with the
culture, suggesting that police use of force is a function of officers'
varying attitudinal commitments to the traditionalview of police cul-
ture. The implications of these findings for policy and future research
are considered.

KEYWORDS: Police, culture, coercion, force, attitudes and behaviors

Since Westley's study of policing in Gary, Indiana in the 1950s, police
scholars have studied the existence, formation, and boundaries of police
culture. This research has generally focused on ways in which officers
cope with the strains of their occupational and organizational environment
(Brown, 1988; Fielding, 1988; Herbert, 1998; Kappeler et al., 1998; Man-
ning, 1995; Paoline, 2001; Reiner, 1985; Reuss-lanni, 1983; Skolnick, 1994;
* This research was supported by Grant 95-IJ-CX-0071 from the National Institute
of Justice. Opinions or points of view expressed in the paper do not necessarily
represent the official positions or policies of the National Institute of Justice or the U.S.
Department of Justice. We would like to thank Rob Worden, David McDowall, Editor
Robert J. Bursik, Jr., and the anonymous reviewers for their helpful comments in the
preparation of this article. Please address all correspondence to William Terrill,
Northeastern University, College of Criminal Justice, 421 Churchill Hall, Boston, MA
02115.

CRIMINOLOGY VOLUME 41 NUMBER 4 2003 1003

1004 TERRILL ET AL.

Van Maanen, 1974). Studies of police coercion have also generated con-
siderable interest. Much of this attention has centered on why officers use
force (Fyfe, 1988; Garner et al., 1995; Klinger, 1995; Muir, 1977; Reiss,
1972; Terrill, 2001; Toch, 1969; Westley, 1970; Worden, 1996). Curiously
absent from these two bodies of literature are inquires that attempt to
quantitatively examine the relationship between alignments with police
culture and acts of coercion. Given that coercive behavior is implicitly,
and perhaps even explicitly, viewed as a salient correlate of police culture,
an empirical examination of the connection between culture and coercion
is warranted.
Using data collected as part of a systematic social observation study of
the police in Indianapolis, Indiana, and St. Petersburg, Florida, this
research examines the relationship between culture-from an attitudinal
perspective-and coercion-from a behavioral perspective. More specifi-
cally, after developing a classification scheme of officers' outlooks in the
context of several culture measures collectively, we examine the extent to
which officers' alignment with cultural attitudes translates into differences
in coercive behavior. Our objective is to determine whether officers who
closely embody the attitudes of the traditional view of police culture are
more likely to use coercion in day-to-day encounters, compared with
officers whose attitudes diverge from this view.
We begin by considering traditional views of police culture as a theoreti-
cal framework. 1 Next, we consider the relationship between attitudinal
dimensions of culture and the use of coercion. Finally, we examine the
extent to which officers vary according to their views of traditional culture,
and the extent to which variation in cultural attitudes translates into differ-
ences in coercive behavior.

POLICE CULTURE
As Chan (1996:111) appropriately notes, "the concept of police culture
in the criminological literature is loosely defined." For example, Westley's
(1970) characterization of culture stressed the secrecy and loyalty aspects
among officers working in a dangerous and hostile work environment.
Skolnick's (1994) depiction of culture described a "police personality"
that, similar to Westley's characterization, was a function of the dangers of

1. We use the term theoretical framework over literature review because, in

many respects, the traditional view of police culture contains all of the components of
theory (i.e., concepts and interrelated propositions), although there is variation in terms
of the phenomena in which it seeks to explain (i.e., usually some form of police attitude,
behavior, or both). In this respect, the environments, coping mechanisms, and out-
comes that are outlined here as part of the police culture are organized in such a way to
explain, as it has been posited by previous police scholars, the use of coercion. We
thank an anonymous reviewer for bringing this distinction to our attention.
 POLICE CULTURE AND COERCION 1005

policing, but also noted that police use their coercive authority over citi-
zens and seek to appear efficient in the eyes of administrators. Others,
like Reuss-lanni (1983), delineate a culture among "street cops" that is a
product of 20 codes, highlighting several different dimensions, including
aspects of both the internal and external environments of policing. In a
similar vein, Sparrow et al. (1990) deduce six "building blocks" or beliefs
of culture that shape the world view of officers. More recently, Herbert
(1998) argued a more structural view of police culture that is built around
six "normative orders" that constrain the choices and social world of
police officers. Although differences in the conceptual nuances of police
culture exist, one can deduce a number of common themes present within
the literature on police culture-primarily in the context of how officers
view and respond to their occupational and organizational environment.
Traditional accounts of police culture describe the coping mechanisms
that officers use to deal with the strains created in their two working envi-
ronments. The occupational environment, which comprises interactions
with citizens, includes the physical danger of police work and the unique
coercive authority that officers wield. The organizational environment,
which comprises interactions with superiors, includes the unpredictability
of supervisory oversight and the ambiguity of the police role. These two
environments are said to cause much stress and anxiety for officers that is
relieved through the prescriptive coping mechanisms of the police culture.
Brown (1988:9) summarizes the relationship between police officers and
their environments:
What must be recognized is that patrolmen lead something of a schiz-
ophrenic existence: they must cope not only with the terror of an
often hostile and unpredictable citizenry, but also with a hostile-even
tyrannical-and unpredictable bureaucracy.
Within their occupational environment, officers cope with danger and
coercive authority by being suspicious (Skolnick, 1994; Westley, 1970) and
maintaining the edge or being one up on citizens at all times (Rubinstein,
1973; Van Maanen, 1974). Within the organizational environment, officers
cope with the strains of punitive supervisors, who endorse an ambiguous
role orientation, by adopting a lay-low/cover-your-ass orientation to police
work (Herbert, 1996; Van Maanen, 1974), and by embracing a crime-
fighter role orientation (Brown, 1988).2 The cumulative effects of the

2. A lay-low approach may be manifested in two different ways. One may be for
officers to avoid virtually all police work, choosing to handle the least amount of work
possible for organizational survival. A second way, and more of an active approach to
laying-low, would be for officers to focus on less ambiguous situations (i.e., more seri-
ous offenses), whereby procedures might be more clearly defined within the organiza-
tion. As such, officers lay-low within the organization, while adhering to the mandate
1006 TERRILL ET AL.

strains that officers confront in their work environment, and the prescrip-
tive coping mechanisms to deal with these strains, produce two defining
outcomes of police culture-social isolation and group loyalty (Westley,
1970; Brown, 1988).
The dangers associated with the occupational environment often
prompt officers to distance themselves from the chief source of danger-
citizens. The coercive authority that officers possess also separates them
from the public. The cultural prescriptions of suspiciousness and main-
taining the edge over citizens in creating, displaying, and maintaining their
authority (Manning, 1995) further divides police and their clientele.
Officers who are socially isolated from citizens, and who rely on one
another for mutual support from a dangerous and hostile work environ-
ment, are said to develop a "we versus they" attitude toward citizens and
strong norms of loyalty to fellow officers. The collectiveness of culture
among officers, and the mechanisms used to cope with the strains of the
occupation, are related to the use of coercion over citizens. That is,
officers, as culture carriers, are expected to "show balls" (Reuss-lanni,
1983:14) on the street during encounters with citizens.

ATTITUDINAL DIMENSIONS OF CULTURE

Previous research holds that particular valences on a number of dimen-
sions are a part of the traditional view of police culture. Given that the
traditional view of police culture is said to consist of a set of attitudes and
values that are shared by officers who collectively cope with the strains of
their work environment (Paoline et al., 2000), what follows are the most
prominent attitudinal features of the police culture.
Research on police officers has noted the negative attitudes that police
hold toward citizens (Reuss-lanni, 1983; Skolnick, 1994; Westley, 1970).
As part of a "we versus they" outlook, officers have generally not trusted
and have been suspicious of the citizens they police. In addition, officers
have historically not believed that those outside the policing profession
would assist them in performing their duties, and even if "outsiders" did
try to assist they would not be of any real help (Sparrow et al., 1990).
Just as research on police culture has noted the negative attitudes that
police officers hold toward citizens, so too have patrol officers' attitudes
been characterized with regard to their primary supervisors and upper
level managers (Crank, 1997; Crank and Caldero, 1991). Feelings of
uncertainty with regard to supervisory expectations and scrutiny have
marked these negative attitudes.
The police culture literature also notes that a major source of tension

of aggressive crime fighting. The latter type of laying low is more of the cultural norm
to minimize scrutiny from police bosses.
 POLICE CULTURE AND COERCION 1007

between officers and supervisors has centered on supervisors' focus of pro-

cedural rules and regulations (Skolnick, 1994). Accounts of culture have
noted that officers' attitudes toward these restrictions have been unfavora-
ble. The cultural reaction has been a lay-low/cover-your-ass attitude,
whereby officers choose to avoid ambiguous situations where negative
evaluation might follow from supervisors.
In addition, police culture research suggests that one of the ways that
officers cope with the ambiguities of their role in society is by focusing
exclusively on crime fighting activities, as service, order maintenance, and
community policing efforts have historically not been regarded as "real"
police work (Kelling and Kliesmet, 1996; Reuss-lanni, 1983).
Related to the ways in which officers identify with their role is their
general beliefs about tactics of law enforcement or how the role should be
performed-aggressiveness and selectivity. Police culture research asserts
that officers hold positive attitudes toward aggressive stops of cars and
"checking out people," as well as favorable attitudes toward selective
enforcement of laws (i.e., assigning felonies a higher priority) (Brown,
1988).
In sum, the traditional view of police culture posits that officers should,
almost uniformly, hold strongly unfavorable views of both citizens and
supervisors, show disdain and resentment toward procedural guidelines,
reject all roles except that which involves fighting crime, and value aggres-
sive patrolling tactics and selectivity in performing their law enforcement
duties.

CULTURE AND COERCION-LINKING ATTITUDES

AND BEHAVIOR

A focal point of nearly all police culture studies is the way in which
officers view and behave toward their main clientele-citizens. Since the
beginning of police culture research, explanations of the police-citizen
relationship have centered around the coercive authority that officers pos-
sess over citizens. As Brown (1988:37) notes, "the routine use of coercion
sets policemen off from society." More specifically, police officers are
granted the right to use legitimate force, which is accompanied by legal
protection (via the courts). The use of coercive authority is, at times,
problematic for officers, especially those who are not comfortable using
such powers. As a way of coping with this dilemma, the police culture
mandates that officers maintain the edge at all times by being prepared or
"one-up" on citizens (Rubinstein, 1973), and never backing down from
citizen resistance to authority (Reiss, 1972; Reuss-lanni, 1983). The failure
to do so is said to have drastic consequences in terms of police-citizen
relationships, as it may allow citizens to gain the upper hand. As Reiss
1008 TERRILL ET AL.

(1972:150) points out, "[tihere are strong subcultural beliefs that the
officer who ignores challenges from citizens loses the respect of citizens
and makes it difficult for other officers to work in the precinct." In this
sense, the relationship between police and citizens is one in which coercive
force is reciprocal, and the culture demands that police officers maintain
control (Brown, 1988).
Despite the inherent link between culture and coercion, findings from
some studies suggest that not all officers equally share the attitudes, val-
ues, and norms of the traditional police culture. More specifically, typol-
ogy research suggests that officers might cope in different ways with the
strains created by their work environment (Broderick, 1977; Brown, 1988;
Muir, 1977; White, 1972). If officers differ in the their cultural commit-
ments, one might reasonably expect differences in the use of coercion. For
example, in examining Muir's (1977:50) qualitative typology study, he
finds that the enforcer embodies many of the attitudes of police culture,
using high levels of coercion, while the reciprocator fails to "exemplify a
capacity to integrate coercion into morals."
Taken collectively, typology studies have included several attitudinal
dimensions. However, individually, each typology has classified officers
based on only two or three dimensions (Worden, 1995). Although this
body of research is informative in identifying attitudinal variation among
officers along several dimensions, it fails to tell us the extent to which dif-
ferences in one's attitudinal style translates into differences in behavior.
Most often, typologies are created based on officer interviews (either
structured, unstructured, or some combination of both); and the connec-
tion between attitudes and behavior is either assumed congruence, or
when tests are done, they rely on a few illustrative examples that reify the
"type" or "style" of officers.
Although some researchers have examined the attitude-behavior rela-
tionship-no study involving a patrol officer classification scheme has
empirically linked cultural attitudes to some form of behavior (Worden,
1995). Snipes and Mastrofski (1990) attempted to replicate Muir's (1977)
research (in another department-"Euphoria"), failing to find a connec-
tion between officers' attitudinal and behavioral style. To the authors'
credit, they addressed a worthy research question, but their sample relied
on extended interviews of only nine officers, and as the authors noted, "a
more compelling methodological limitation in the present study is the
small number of observations per officer" (1990:290). In addition, in
assessing Muir's behavioral "critical incidents," the authors reported that
the frequency of these incidents occurred just over once per shift. This
leads the authors to conclude that future attitude-behavior studies should
include both more officers and observations.
Jermier et al. (1991) used a cluster analysis technique to classify officers
 POLICE CULTURE AND COERCION 1009

with respect to their differences from the "official" (crime fighting com-
mand and bureaucracy) police culture. This work sheds light on cultural
fragmentation in police departments, as the authors found five groups of
officers based on their occupational attitudes and characteristics. How-
ever, like typology studies before them, any links to behavior were based
on single illustrative observations of officers from each of the five groups.
Herbert's (1998) recent ethnographic research in the LAPD highlighted
connections between police culture, as a function of six normative orders
of policing (i.e., law, bureaucratic control, adventure/machismo, safety,
competence, and morality), and differential behavior among officers. Her-
bert (1998:361) explains that each of the normative orders revolve around
a principle value that "provide different sets of rules and practices that
officers use to define situations and to determine their response." Her-
bert's dynamic view of culture has important implications for understand-
ing behavioral variation (i.e., coercion), based on differences in how
officers interpret and define their occupational world, although systematic
examinations of this relationship have yet to be conducted.

RESEARCH PROPOSITIONS
To date, studies of police culture have failed to empirically address the
extent to which differences in coercion are a result of differences in cul-
tural based attitudes. Moreover, police researchers have often failed to
uncover an attitude-behavior relationship. This failure has been traced to
problems in the measurement of police attitudes, behaviors, or both
(Frank and Brandl, 1991; Worden, 1995). As Worden (1995:74) explains,
,'more and better research is needed before these hypotheses can be
rejected, however, and better measures of a broader range of police out-
looks would improve the quality of such research." In this article, we
attempt such an endeavor, by first developing a classification scheme of
officers cultural outlooks, and then by examining the ways in which these
outlooks relate to their behavior. We argue that the use of coercion with
suspects will vary depending on the ways in which officers adhere to the
attitudes associated with the traditional view of the police culture. More
specifically, based on prior work that suggests an underlying coercive ele-
ment within the police culture, we posit that those officers who closely
embody the values of the traditional views of police culture should be
more coercive compared with those that do not equally align with the
culture.

DATA
The present analyses draw on data from the Project on Policing Neigh-
borhoods (POPN), which examined policing in Indianapolis, Indiana and
1010 TERRILL ET AL.

St. Petersburg, Florida during the summers of 1996 and 1997, respectively.
Each city was diverse in social, economic, and demographic terms.
Although many features were similar across the two departments (i.e.,
attempts to civilianize numerous positions, upgrading technological capa-
bilities, emphasis on improving community relations), they differed in one
key respect that is particularly relevant to the current study-the style of
policing promoted by management. Indianapolis officials promoted an
aggressive order maintenance approach (i.e., the suppression of public dis-
orders, drug crime, gang activity, and illegal weapons by intrusive law
enforcement methods). Conversely, St. Petersburg emphasized a prob-
lem-solving/community partnership approach (i.e., identify problems using
crime and calls-for-service data and by working closely with community
groups) (see Terrill and Mastrofski, 2003 for a more detailed description of
both departments' occupational and organizational environments). As
such, we would expect, all else being equal, that Indianapolis officers, who
work in an organizational environment that supports and endorses aggres-
sive crime fighting, will resort more readily to the use of coercion.
Two sources of data from POPN are used in the present analyses: in-
person interviews of patrol officers and systematic observation of officers.
Data on officer attitudes are taken from in-person interviews conducted in
a private room by trained researchers who did not conduct field observa-
tions. Officers were assured by each interviewer that responses would be
kept confidential by the project staff. The interview consisted of a mix of
questions posed by interviewers and checklists completed by respondents
in the interviewers' presence, the latter to minimize interviewer effects.
Of the 426 officers assigned to patrol in Indianapolis, a total of 398 were
surveyed, producing a completion rate of 93%. In St. Petersburg, 240 out
of a possible 246 patrol officers were interviewed, a completion rate of
98%.
Patrol observation was conducted in 12 beats in each city, with the sam-
ple of beats matched as closely as possible across the two sites according to
the degree of socioeconomic distress. Socioeconomic distress was mea-
sured as the sum of the percentages of families with children headed by a
single female, the adult population that is unemployed, and the population
below 50% of the poverty level-an index similar to one used by Sampson
et al. (1997). The sample excluded those beats with the lowest socioeco-
nomic distress; observations concentrate in areas where police-citizen
interactions are most frequent (see Parks et al., 1999:492-493 for further
detail concerning POPN sampling strategy).
Observations were conducted according to systematics social observa-
tion (SSO) methodology by trained researchers, who accompanied officers
throughout a matched sample of work shifts in each of the selected beats,
for a total of approximately 240 hours per beat (see Mastrofski et al., 1998
 POLICE CULTURE AND COERCION 1011

for further detail). Observers noted officers' encounters with the public.
An encounter was a face-to-face communication between officers and citi-
zens that was more than a passing greeting. Observers recorded
encounters with approximately 6,500 citizens in Indianapolis and 5,500 cit-
izens in St. Petersburg, with events ranging from less than a minute to
several hours. Among the citizens encountered were crime victims, wit-
nesses, a variety of service recipients, and criminal suspects. The selection
criteria used for the behavioral analysis is based on those interactions with
people whom police or other citizens present placed in the role of suspect
(wrongdoers, peace disturbers, or persons for whom complaints were
received). The data file consists of 3,223 police-suspect encounters.

ANALYSIS AND FINDINGS

In testing the proposition that differences in coercion are a result of
variation in cultural alignments, we begin by developing a classification
scheme of officers based on their attitudes toward citizens, supervisors,
procedural guidelines, role orientation, and policing tactics. 3 As previ-
ously noted, these attitudes are all rooted in the theoretical framework of
police culture. Our expectation is that officers who attitudinally align with
the traditional views of police culture will use more coercion than those
who do not. Table 1 provides an overview of each of the attitudinal
dimensions of culture. A total of ten measures were constructed, based on
4
22 survey items.
Two measures of officers' attitudes toward citizens are examined. The
first measure is based on a single Likert item examining officer distrust of
citizens. The second measure is a three-item additive index concerning the

3. Although we measure some of the more widely regarded aspects of police cul-
ture, the data do not permit us to measure all components of culture-for example,
peer group loyalty and the extent to which officers perceive danger in their environ-
ment. Hence, our culture measure offers the opportunity for a partial test of the cul-
ture-coercion link, as opposed to a complete test. Moreover, among the measures of
culture that we include generally as occupational "attitudes," there are a few that might
be regarded more as perceptions or assessments (e.g., citizen cooperation and district
management). Although some social psychologists distinguish more finely among
these, and other subjective outlooks (see for example Rokeach, 1972), like other social
scientists, we include them together as they all are geared toward more concrete objects
and situations (i.e., the environments of policing) and not a higher level of abstraction
like beliefs.
4. Five of the measures are multiple-item additive indices. The results of prelimi-
nary and confirmatory factor analysis indicated that the indices tap a single underlying
dimension. Each of the indices correlates with the corresponding factor scale at or
above .95. The reliabilities of the indices are acceptable: The alpha coefficients are .75,
.89, .71, .73, and .64, respectively, for the citizen cooperation, sergeants, district manage-
ment, order maintenance, and community policing indices.
1012 TERRILL ET AL.

Table 1. Attitudinal Measures and Descriptive Statistics

(N = 585)
Dimension Survey Item(s) (mean, standard deviation, and range in parentheses)
Citizens
Distrust' 1. Police officers have reason to be distrustful of most citizens. (2.01,
.79, 4-12)
Cooperation' 2. How many citizens in [the respondent's beat/CPA] would call the
police if they saw something suspicious?
3. How many citizens in [the respondent's beat/CPA] would provide
information about a crime if they knew something and were asked
about it by the police?
4. How many citizens [the respondent's beat/CPA] are willing to
work with the polic to try to solve neighborhood problems? (9.46,
1.81, 3-12)
Supervisors a
Sergeants 5. My supervisor's approach tends to discourage me from giving
extra effort (reverse coded).
6. My supervisor is NOT the type of person I enjoy working with
(reverse coded).
7. My supervisor lets officers know what is expected of them.
8. My supervisor looks out for the personal welfare of his/her
subordinates.
9. My supervisor will support me when I am right even if it makes
things difficult for him or her. (16.95, 3.64, 5-20)
Management' 10. When an officer does a particularly good job, how likely is it
that top management will publicly recognize his or her
performance?
11. When an officer gets written up for a minor violation of the
rules, how likely is it that he or she will be treated fairly?
12. When an officer contributes to a team effort rather than look
good individually how likely is it that top management here will
recognize it? (7.59, 2.13, 3-12)
ProceduralGuidelines' 13. In order to do their jobs, patrol officers must sometimes
overlook search and seizure laws and other legal guidelines. (1.64,
.88, 1-4)
Role Orientation a
Law Enforcement 14. Enforcing the law is by far a patrol officer's most important
d responsibility. (3.13, .75, 1-4)
Order Maintenance 15. How often they think that patrol officers should be expected to
do something about neighbor disputes.
16. How often they think that patrol officers should be expected to
do something about family disputes.
17. How often they think that patrol officers should be expected to
do something about public nuisances. (9.08, 1.95, 3-12)
Community Policing' 18. How often they think that patrol officers should be expected to
do something about nuisance businesses.
19. How often they think that patrol officers should be expected to
do something about parents who don't control their kids.
20. How often they think that patrol officers should be expected to
do something about litter and trash. (7.05, 1.88, 3-12)
Policing Tactics
Aggressive Patrol" 21. A good patrol officer is one who patrols aggressively by
stopping cars, checking out people, running license checks, and so
forth. (2.92, .85, 1-4)
Selective Enforcement' 22. How frequently would you say there are good reasons for not
arresting someone who has committed a minor criminal offense.
(3.00, .59, 1-4)
"Coded: 1 = disagree strongly, 2 = disagree somewhat, 3 = agree somewhat, 4 = agree strongly.
Coded: 1 = none, 2 = few, 3 = some, 4 = most.
Coded: 1 = very unlikely, 2 = somewhat unlikely. 3 = somewhat likely, 4 = very likely.
d Coded: 1 = never, 2 = sometimes, 3 = much of the time. 4 = always.
 POLICE CULTURE AND COERCION 1013

degree to which officers perceive citizens as cooperative. Officers' atti-

tudes toward supervisors are measured with two separate additive indices.
The first measure reflects the degree to which officers hold favorable opin-
ions of their most immediate or front-line supervisors-sergeants. The
second measure focuses on more senior personnel, reflecting a general ori-
entation toward top district management.
Officers' attitudes toward procedural guidelines are measured in terms
of a single Likert item that assesses the degree to which officers believe
they should sometimes overlook search and seizure laws and other legal
guidelines. Role orientation is measured with three separate items. The
first measure examines officers' law enforcement role orientation, and it is
based on a single Likert questionnaire item. The second measure, role
orientation toward the order maintenance function, is concerned with the
degree to which order maintenance situations are included in their role
definitions, which is measured with an additive index of three survey
items. The third measure, patrol officers' orientation toward community
policing, measures the degree to which officers consider disorderly condi-
tions to be police matters, which is measured with an additive index, com-
prising three survey items.
Finally, patrol officers' attitudes toward policing tactics are measured in
terms of two separate items. The first measure examines officers' views of
aggressive patrolling tactics, and it is based on a single Likert question-
naire item (identical to one that Brown [1988] used). A second measure
assesses officers' views of selective enforcement, and it is also measured in
terms of a single Likert questionnaire item.

CLUSTER ANALYSIS AND PATROL OFFICER GROUPS

A cluster analysis procedure was used to construct our culture measure.
Cluster analysis is the most appropriate analytical technique because we
are interested in assessing officer similarity based on a set of theoretically
relevant variables simultaneously, as opposed to the independent or addi-
tive effects of these variables. Cluster analysis allows the user to classify
units of analysis (i.e., officers) based on their similarity with respect to a
set of variables (i.e., attitudes). Although there are several types of clus-
tering techniques, the one chosen for this research is K-Means Cluster
Analysis, which is the most ideal clustering method for large data sets (i.e.,
more than 200 cases) (Nourusis, 1990). The K-Means procedure, based on
the nearest centroid sorting method, makes a preliminary pass through the
data to determine the initial cluster centers. Once the centers are deter-
mined, the procedure assigns cases to each cluster based on an estimation
of the smallest or closest distance between that case and the center of a
given cluster-the cluster's centroid (Nourusis, 1990). A group's centroid
is a compilation of a combined mean score across all variables clustered.
1014 TERRILL ET AL.

Cases are then compared with one another based on their cluster member-
ship or how much alike they are to one another, as determined by their
nearness to their cluster centroid. For this study, each officer's combined
responses to ten attitudinal dimensions are compared with every other
officer, and those officers who are most similar are placed in the same
cluster or group. Officers in each cluster do not have to be totally identi-
cal to one another, but they are more like the officers in their cluster com-
pared with the other clusters.
Because no attitude in this analysis should be weighted more heavily
than another, the variables were standardized into z scores (Everitt, 1980).
Using a LISTWISE selection criteria, which excludes cases where a
response to a given attitude is missing, the working data set includes 585 of
the total 638 patrol officers, or 92% of the sample population. In deter-
mining the number of clusters, an iterative selection process was used,
which established a range of clusters to assess the best fit to the data (i.e.,
the ratio of the collective distance between officers and their cluster cent-
roid to the number of clusters). In examining the mean of officers' dis-
tance from their cluster centers at different cluster solutions (i.e., a range
of clusters), one is able to identify the most efficient solution, as mean
scores flatten out or fail to exhibit differences from one cluster specifica-
tion to another. Higher mean scores suggest that officers are more dis-
persed or further away from the cluster centroids. Table 2 displays the
range of specified clusters, the squared distances (to accentuate differ-
ences) from the cluster centers of officers for each number of clusters
specified, and the difference in mean scores as one moves iteratively
across solutions.
Results of the K-Means Cluster Analysis reveal that seven distinguisha-
ble groups of officers emerge, as the last substantial decrease in means
occurs from a six-cluster model to a seven-cluster model, with few fluctua-
tions through further cluster enumerations. These groups reflect degrees
of variation in terms of the most prominent attitudinal features of culture
(for a detailed description of the seven clusters, please see Paoline, 2001).
Although these findings suggest that officers' attitudes are less represen-
tative of a universally shared culture (as we find more than one group of
officers), there are no a priori or ex post facto ways to rank order each of
the seven groups (e.g., a continuum of cultural commitment). Interest-
ingly, there are groups that are more "positively" disposed toward culture
than others (i.e., their attitudes are consistent with many of the expecta-
tions of the police culture), whereas others appear to represent the antith-
esis of what is expected from cultural members. For example, three
groups hold attitudes congruent with many of the postulates of the police
culture, but ranking one over the other in terms of "more" alignment is
not readily possible. The police culture literature only suggests that strong
 POLICE CULTURE AND COERCION 1015

Table 2. K-Means Centroid Cluster Formations

Number of Clusters Squared Means Mean Difference
2 8.93 *
3 8.30 .63
4 7.75 .55
5 7.29 .46
6 6.97 .32
7 6.61 .36
8 6.46 .15
9 6.39 .07
10 6.16 .23
11 6.01 .15
12 5.95 .06
13 5.84 .11
14 5.67 .17
15 5.56 .11

cultural alignment will result in high levels of coercion, but variation in

coercion from groups other than the extremes (i.e., high or low) remains
much more speculative. For these reasons, the groups were trichotomized
to represent those who were positively oriented toward the traditional
views of police culture (pro-culture), those who were negatively oriented
(con-group), and those falling somewhere in the middle (mid-range). This
will also make the analysis of behavior more readily interpretable, as
groups can be compared with one another in more meaningful ways.
Table 3 provides descriptive statistics of the clusters that comprise each
of the three groups of varying cultural commitment. Also reported here
are the results of a one-way analysis of variance (ANOVA), which was
used to assess statistical differences in attitudinal means across the seven
clusters of officers. In order to minimize Type-I errors with multiple com-
parisons (i.e., overestimating statistical significance), the most conserva-
tive ANOVA statistical criteria (i.e., Bonferroni) was used (Nourusis,
1990).5 It is evident, based on Table 3, that some of the attitudes typically
associated with the police culture work to separate the clusters (e.g., role
orientation, procedural guidelines, and citizen cooperation) more than
others (e.g., selective enforcement and sergeants). These findings have

5. In order to reject the null hypothesis that the means for each of the attitudes
were equal across the seven groups, an F-ratio of 2.12 was needed to be significant at
the .05 level. The null hypothesis was rejected for all of the attitudes, as F-ratios
exceeded this value: citizen cooperation (F = 31.39), citizen distrust (F = 22.75), ser-
geants (F = 139.69), district management (F = 28.60), legal restrictions (143.72), law
enforcement (F = 57.04), order maintenance (F = 64.14), community policing (F =
60.52), aggressive patrol (F = 29.25), and selective enforcement (F = 11.76).
1016 TERRILL ET AL.

N C '~ N 'C 'C & n

C rFcl 4 rl - r- C4r - rT

r
C4 r ' 'C

&V '~
le \i5

n LN ) N NNNn cnN cn c

m C c C7C r{r~- &r

N- N0c C
N ' r-N C4 q r

NN 'C m N \ 0 rN 0l C N Nt x
N N -
N C-: 0 Nr'l0,,NN C N0

N00 OC1' N0 'C Nm0 000 en N N

c~00-.C Cc

N ) cn C4 cn Nl cn
CIO .1- N7 15 ) O 'V)~'CC

t
,I I 0- M 000 O= =0, N ~ 0- -t N, 'C N C1

N -N Ct 'Ci cl cliC r- r; c
aI

. C_ a0
= 1
a )t
C - UCO 0) o 0
 POLICE CULTURE AND COERCION 1017

important implications for the understanding of police culture among con-

temporary officers. That is, there are reactions to the work environment
in which groups of officers differ rather significantly, and there are some
components of the police world that generate higher levels of agreement
across the groups. This is consistent with some of the more recent police
culture studies that stress both the collectiveness and internal variation
among officers (Paoline et al., 2000).
The ANOVA results found in Table 3 also provide empirical evidence
for the contrasts that exist among the clusters that comprise the pro-, mid-,
and con-groups. For example, when we compare Cluster 1 of the con-
culture group with the clusters that comprise the pro-culture group, we
find that Cluster l's means significantly differed for seven of the ten
dimensions when compared with Cluster 3; seven out of ten dimensions
when compared with Cluster 5; and five of the ten dimensions when com-
pared with Cluster 7. Moreover, Cluster 6 of the con-culture group signifi-
cantly differed for eight out of ten dimensions, when compared with each
of the three pro-culture clusters. Finally, the mid-culture group contains
clusters of officers that share attitudes of both the pro- and con-groups,
thus falling somewhere in the middle of the two groups. For example,
officers in Cluster 2 are similar, in some respects, to Cluster 7 of the pro-
culture group (i.e., five out of ten significant mean differences), but also
similar to Cluster 6 of the con-culture group (i.e., six out of ten significant
mean differences). Similarly, officers in Cluster 4 appear to combine the
attitudes of officers in Cluster 3 of the pro-culture group (i.e., five out of
ten significant mean differences) and Cluster 1 of the con-culture group
(i.e., five out of ten significant mean differences).
PRO-CULTURE GROUP

Although we fail to find a cluster of officers that exhibits total congru-

ence with the tenets of the police culture, we do find that the attitudinal
composition of officers in these clusters are most similar to traditional
characterizations of police culture. Officers in Cluster 3 hold unfavorable
views of citizens and supervisors, are positively oriented toward aggressive
crime fighting patrol tactics, and endorse selective enforcement of the law.
In addition, officers in Cluster 5 are included in the pro-group, as they
moderately distrust citizens in the aggressive pursuit of enforcing the law,
maintaining order, and eliminating disorder violations. Finally, Cluster 7
contains officers who are the most distrustful of citizens, the strongest sup-
porters of the law enforcement role, and among the highest (with Cluster
5) proponents of aggressive patrolling tactics-all attributes typically asso-
ciated with traditional characterizations of police culture. What is impor-
tant, especially with respect to coercion, is that officers in these clusters
collectively hold the least favorable views of citizens, in performing their
1018 TERRILL ET AL.

law enforcement role aggressively, even if it means (for two of the clus-
ters) violating citizen rights. Our expectation is that officers in the pro-
culture group will display their coercive authority behaviorally more fre-
quently, and at higher levels, than other groups.

CON-CULTURE GROUP

The con-culture group consists of two clusters that, in many ways,

represent attitudes that are the antithesis of what is expected from mem-
bers of the police culture. Although varying in attitudinal intensity, the
favorable views of citizens, and (to a lesser extent) supervisors, as well as
the nonaggressive approach to policing that neither favorably accepts the
crime fighting role nor the violation of citizens' due process safeguards,
makes officers in Cluster 1 and Cluster 6 part of the con-culture group. To
the extent that officers in the pro-culture group exert more coercion, our
expectation is that officers in the con-culture group will be the least
coercive.

MID-RANGE CULTURE GROUP

Two clusters of officers comprise our mid-range culture group. As the

title implies, the constellation of officer attitudes in these clusters is not
exclusively regarded as part of either the pro- or con-culture group, but
instead it falls somewhere in the middle. For example, consistent with the
police culture, officers in Cluster 2 endorse a selective approach to aggres-
sive crime fighting. Contrary to the traditional culture, officers in this clus-
ter have very favorable views of citizens, supervisors, and procedural
guidelines, while performing roles other than just law enforcement. For
some of these attitudes, it is clear that officers in Cluster 4 do not tip the
scales in either direction. There does appear to be two distinguishable
attributes of Cluster 4 members-their extremely negative attitudes
toward supervisors in their organizational environment, and their very
strong positive attitudes toward citizens in their occupational environ-
ment. Our expectation is that officers in the mid-range group will be less
coercive than the pro-culture group, but more coercive than the con-cul-
ture group.
Previously we noted that we used a trichotomized measure over a
seven-category measure to better illustrate varying commitments to the
traditional police culture. ANOVAs for the seven clusters (see Table 3)
were used to provide empirical evidence for the reduction of the clusters
to the three groups. We also examined ANOVAs for the three-group
measure to provide further empirical evidence for this decision. Table 4
illustrates significant mean differences across each of the three groups of
varying cultural commitment for each of our ten measures. Table 4 reveals
 POLICE CULTURE AND COERCION 1019

that the three-group culture measure is an improvement over the seven-

cluster measure, as 66.6% of the possible pairwise comparisons are statisti-
cally significant compared with the 60.9% found in Table 3.

Table 4. Statistically Significant (p < .05) Mean

Comparisons Among Pro-, Mid-, and
Con-Groups
Dimension Pro Mid Con Overall
Citizen Distrust Mid, Con Pro - Pro - 2/3
Citizen Cooperation Mid, Con Pro, Con Pro, Mid 3/3
Sergeants Mid - Pro, Con - Mid 2/3
District Management Mid - Pro - - - 1/3
Procedural Guidelines Mid, Con Pro, Con Pro, Mid 3/3
Law Enforcement - Con - Con Pro, Mid 2/3
Order Maintenance Mid - Pro, Con - Mid 2/3
Community Policing Mid - Pro, Con - Mid 2/3
Aggressive Patrol Mid, Con Pro, Con Pro, Mid 3/3
Selective Enforcement . . . . . . 0/3

BEHAVIORAL ANALYSES
Given the creation of an attitudinal culture measure, we turn our atten-
tion to examining the extent to which these three groups of officers differ
in their application of coercion in day-to-day encounters with suspects.
Coercion was defined as acts that threaten or inflict physical harm on citi-
zens. This includes both verbal and physical force. As correctly noted by
Klinger (1995:173), "[b]ecause voice commands are viewed as force in law
enforcement circles, they are properly included in the universe of behavior
that researchers should consider in their studies of police use of force."
Thus, we broaden the scope of the dependent variable to include not only
physical forms of police force, but verbal force as well (for similar
approaches, see Alpert and Dunham, 1997; Garner et al., 1995; Klinger,
1995; and Terrill, 2001). Within this context, our dependent measure is
ordinally ranked in the following manner: none, verbal (commands and
threats), physical restraint (pat downs, firm grip, handcuffing), and impact
methods (pain compliance techniques, takedown maneuvers, strikes with
6
the body, and strikes with external mechanisms).

6. In terms of verbal force, a command was defined as a statement by an officer

that was in the form of an order, whereas threats involved a command followed by an
explicit or implicit intended consequence for not complying. For physical restraint, pat
downs were defined as instances when an officer physically touched a suspect as part of
a cursory search; a firm grip included an officer grabbing a suspect in a forceful manner
with a tight grip; and handcuffing involved placing restraints on a suspects' wrists.
1020 TERRILL ET AL.

BIVARIATE ANALYSIS

We began by looking at a simple cross tabulation (not shown) between

our measures of culture and coercion. As expected, officers in the pro-
culture group used force more frequently than those in both the mid- and
con-culture groups. Overall, pro-culture officers relied on force in 61.3%
of the observed police-suspect encounters, compared with 56.1% of mid-
culture officers and 50.4% of con-culture officers. This pattern held for
each type of force. The most pronounced difference was found within the
verbal force category where pro-culture officers used force in 39.0% of the
encounters, compared with 35.6% and 33.1% of the encounters involving
mid- and con-culture officers, respectively.

MULTIVARIATE ANALYSIS

Although a simple cross tabulation demonstrates statistically significant

differences across the three groups of officers with respect to the use of
force, there are numerous factors that may account for such differences.
Hence, we examined officers' attitudinal orientation toward culture (i.e.,
pro, mid, con) while controlling for situational factors that may affect
police use of force. Descriptions of independent variables and their
hypothesized relationship to coercion are shown in Table 5.
Legal justification for the use of force is most notably found when sus-
pects are resistant, there is a threat to citizen or officer safety, an increased
degree of culpability is present (as reflected in the strength of the evidence
measure), and in the course of making an arrest (Terrill and Mastrofski,
2002). Hence, we include each of these as potential determinants of force.
Suspect resistance was defined as acts that thwart, obstruct, or impede an
officer's attempt to elicit information; failure to respond or responding
negatively to an officer's commands or threats; and any physical act,
proactive or reactive, against an officer's attempt to control the suspect.
Suspect behavior that was cooperative and responsive to police direction
was considered and coded as no resistance. Resistance was measured
according to the severity of defiance posed to police, and placed along a
"continuum" ranging from least to most severe harm. Resistance was
ranked in the following manner: none, passive, verbal, defensive, and
active. 7

Finally, for impact methods, pain compliance techniques were defined as holds that
cause pain to a specific part of the body; takedown maneuvers included instances when
suspects were thrown, pushed, or shoved to the ground, against a wall, against a car or
any other surface; strikes with the body included hitting a suspect with the hands, fists,
feet, legs, or any other part of the body; and strikes with an external weapon included
the use of any item that was not part of the body.
7. Passive resistance was defined as suspect behaviors that were unresponsive to
police verbal communication or direction. Verbal resistance included a suspect verbally
 POLICE CULTURE AND COERCION 1021

Table 5. Description of Independent Variables

Hypothesized
Variable Effect Definition
Culture
1 = pro-officers, 0 = non-pro-officers
1 = mid-officers, 0 = non-mid-officers
1 = con-officers, 0 = non-con-officers

Control
Resistance + Level of suspect resistance: 1 = none, 2 =
passive, 3 = verbal, 4 = defensive, 5 -
active
Safety + 1 = suspect has weapon or within jump
and reach, 0 = all other
Conflict + Suspect in conflict with another citizen on
scene: 1 = none, 2 = calm verbal, 3 =
agitated verbal, 4 = threatened assault,
5 = assault
Arrest + 1 = suspect is arrested, 0 = not arrested
Evidence + Summative index (0-8) of the evidence of
the target's or requester's violation of
the law
Male + Suspect gender: 1 = male, 0 = female
Non-white + Suspect race: 1 = nonwhite, 0 = white
Age - Suspect age: 1 = 0-5 years, 2 = 6-12 years,
3 = 13-17 years, 4 = 18-20 years, 5 = 21-
29 years, 6 = 30-44 years, 7 = 45-59
years, 8 = 60+
Wealth - Suspect class: 1 - chronic poverty, 2 =
low, 3 = middle, 4 = above middle
Demeanor + 1 = suspect disrespectful to police in
language or gesture, 0 = all other
Drug/alcohol + 1 = suspect shows behavioral effects of
drug/alcohol, 0 = all other
# Officers +/- Number of officers on scene
# Bystanders +/- Number of citizen bystanders on scene
Violence anticipated + 1 = indication of violence from dispatcher,
other officers, or observed officers' own
knowledge (revealed by comments), 0 =
all other
Problem Type + 1 = problem involves a dispute, traffic
incident, or suspicious person, 0 = all
other
Proactive + 1 = officer initiated, 0 - all other
Site + 1 = Indianapolis, 0 = St. Petersburg
1022 TERRILL ET AL.

A citizen safety issue was coded when the suspect involved in the
encounter was in conflict with another citizen on scene. An officer safety
issue arose when a suspect had any sort of weapon on his or her person or
within "jump and reach." The evidence measure takes into account a
number of different forms of culpability, including eyewitness testimony,
physical evidence, and confessions. 8 Further, an arrest was defined as tak-
ing a suspect into custody for the purpose of charging him or her with a
criminal offense. 9
Several additional control variables posited to influence police force are
also included in the model. We limit these measures to "situational"
predictors of force giver that previous studies have found situational fac-
tors to be the most consistent predictors of force (see Sherman, 1980; Ter-
rill, 2001; and Worden, 1996). Four of the control measures involve
suspect characteristics (i.e., suspect gender, race, age, and wealth),
whereas another two (i.e., suspect demeanor and impairment) involve sus-
pect presentation behaviors. Demeanor involved the suspect doing some-
thing that showed disrespect to the individual or authority of the police
officer. 1° The alcohol and drug measure included any indication of use,
including the smell of alcohol on the breath, slurred speech, impaired
motor skills, or unconsciousness. In addition, both the number of officer
and citizen bystanders present on the scene of an encounter can influence
the likelihood or level of police force, although the impact of the effect is
open to interpretation. Further, when officers anticipate violence, they
may be quicker to resort to force themselves. If the dispatcher indicated

rejecting police verbal communication or direction. Defensive resistance was defined as

suspect attempts to evade police attempts at control. Active resistance included the
suspect either attempting or actually attacking or striking an officer.
8. A summative index (0-8) of the evidence of the target's or requester's viola-
tion of the law was used. Points were assigned for each factor present and summed:
officer observed suspect perform an illegal act (3), suspect gave officer a full confession
(2), suspect gave officer a partial confession (1), officer observed physical evidence
implicating suspect (1), and officer heard testimony from other citizens implicating the
suspect (1).
9. Those who are arrested are, of course, formally certified as criminal wrongdo-
ers, and both departments required a degree of physical force (handcuffing) when this
occurred. The inclusion of an "arrest" variable allows us to control for that level of
physical force required by department procedure to accomplish the arrest, an act that
legitimates physical control.
10. This included a variety of verbal statements: calling the officer names, making
derogatory statements about the officer or his family, making disparaging or belittling
remarks, slurs (racial, sexual, lifestyle). Ignoring the officers' commands or questions
did not constitute disrespect, but rather it was classified as passive resistance. In addi-
tion, certain gestures and actions were coded as disrespect. These including displaying
the middle finger in the direction of the police, obscene gestures, and spitting in the
presence of an officer (even if not in the direction of the officer).
 POLICE CULTURE AND COERCION 1023

the possibility of violence, or if the officer picked up on some other cue,

such as the suspect's reputation, the observer coded that violence was
anticipated.
Type of problem is included to account for those cases most often asso-
ciated with an increased likelihood of force. This is similar to what others
have done when selecting cases for study (Fyfe 1988). These researchers
hypothesized that certain types of cases (disputes, traffic stops, attempts to
question suspicious persons) are more likely to lead to force than others
(e.g., shoplifting). Therefore, they use problem types as the inclusion cri-
teria. By using "potentially violent" problem types as a predictor in the
model, it is hypothesized that such cases will predict force. Further, it is
hypothesized that when the encounter is officer-initiated (i.e., proactive),
officers may be quicker to assert their authority and to do it more force-
fully, perhaps because police legitimacy is lower than when the officer is
invited or called on; or alternatively, officers may simply observe some
sort of behavior requiring immediate police intervention thereby increas-
ing the probability of force (Reiss, 1972). The final measure takes into
account whether agency plays a role in the use of force. Finally, it is
important to note that all relevant independent variables are coded to
ensure a causal relationship (e.g., the highest level of suspect resistance
prior to the highest level of force).
Preferably one would investigate the effect of the situational factors by
nesting suspects' encounters within specific officers and subsequently
using hierarchical modeling techniques. However, the structure of these
data did not permit an analysis of this type. Prior research suggests that
ten or more level-one units (in this case, police-suspect encounters) are
necessary to estimate stable hierarchical models with acceptable levels of
bias (see Bryk and Raudenbush, 1992; Mok and Flynn, 1998). Although
the average number of suspects per officer was almost 12, too many
officers (58%) had an insufficient number (<10) to allow for nesting of
situational factors within individual officers. As a result, to assess the
effects of the various determinants, an ordered probit model was esti-
mated using LIMDEP version 7.0.11

11. Alternatively, one might run separate ordered probit models on each of the
three groups of officers (i.e., pro, mid, and con) or even on the two most pronounced
groups (i.e., pro and con) and then perform a model t-test comparing differences
between groups to ascertain whether officers in one group (i.e., pro) were significantly
more or less likely to use force compared with another group (i.e., con) in reference to
additional independent variables (e.g., resistance, safety, demeanor, etc.). However,
the central question posited is not whether there are differences between groups
according to the various control variables, but whether officers in one group are more
forceful than officers in another group on the whole. Performing separate regression
equations fails to answer this question.
1024 TERRILL ET AL.

Table 6. Descriptive Statistics

Variable Range Mean S.D.
Dependent
Highest level of force 0-3 .80 .81
Culture
Pro 0-1 .52 .49
Mid 0-1 .26 .44
Con 0-1 .21 .41
Control
Resistance 1-5 1.21 .66
Safety 0-1 .02 .12
Conflict 1-5 1.13 .57
Arrest 0-1 .11 .31
Evidence 0-8 1.31 1.69
Male 0-1 .72 .45
Non-white 0-1 .63 .48
Age 1-8 .52 1.34
Wealth 1-4 .24 .56
Demeanor 0-1 .09 .29
Drug/alcohol 0-1 .21 .40
# Officers 1-26 2.22 1.61
# Bystanders 1-100 4.23 5.72
Violence anticipated 0-1 .08 .28
Problem Type 0-1 .47 .50
Proactive 0-1 .44 .49
Site 0-1 .43 .49

Table 6 provides descriptive statistics for coercion and each of the inde-
pendent variables. Ordered probit results are presented in Table 7.12 The
overall model is significant as evidenced by the chi-square statistic.
Approximately 27% of the variance is explained, although caution is
required as ordered probit only generates a pseudo r-square statistic. As
expected all of the legal justification predictors were significantly related
to force. In addition, officers were also more likely to resort to higher
levels of force in encounters involving male, nonwhite, young, and poorer
suspects, as well as when the suspect showed signs of alcohol/drug impair-
ment. Further, an increased number of officers on the scene, proactive
encounters, and those interactions involving Indianapolis officers were all
more likely to increase the probability of force.
Most importantly, the culture measure is significantly related to force as

12. Collinearity diagnostics indicated no evidence of multicollinearity among the

independent variables based on their tolerance or variance inflation factors (VIF); the
highest VIF was 1.6, well below an acceptable level (see Neter et al., 1985:392).
 POLICE CULTURE AND COERCION 1025

Table 7. Ordered Probit Estimates-Combined Sites

Variable Coefficient Significance
Culture
Pro .200
Mid .155
Control
Resistance .352
Safety .705
Conflict .111
Arrest 1.180
Evidence .064
Male .259
Non-white .163
Age -. 096
Wealth -. 141
Demeanor -. 054
Drug/alcohol .339
# Officers .065
# Bystanders -. 001
Violence anticipated .038
Problem Type .049
Proactive .192
Site .158
Chi-square 1000.589, p < .001.
Intercept -. 512 (.165).
N 3223.
Pseudo-R-Square .267.

well. As hypothesized, encounters involving officers with pro-culture atti-

tudinal dispositions were significantly more likely to result in increased
levels of force when compared with con-culture officers-the reference
category. Further, encounters involving officers with mid-culture attitudi-
nal dispositions were also more likely (and significantly so) to involve
13
increased levels of force when compared with con-culture officers.
Besides estimating the model using the con-culture group as the refer-
ence category, we estimated two additional models using the pro- and mid-
culture groups as the reference category (not shown). When the pro-cul-
ture group served as the reference category, both mid- and con-officers

13. Note that an additional model, which included three officer level variables
(gender, race, and length of service) in addition to the control measures shown in Table
7, was also estimated to ensure model stability. Results showed that officers with less
years of experience were significantly more likely to rely on higher levels of force,
whereas officer race and gender had no effect. More importantly, however, all of the
situational control measures, as well as the pro- and mid-culture measures, remained
the same both in terms of direction and significance.
1026 TERRILL ET'AL. '

were less likely to use force, but only the con-culture group reached statis-
tical significance. When the mid-culture group served as the reference cat-
egory, pro-officers were more likely to use force, but showed no statistical
difference; conversely, con-officers were significantly less likely to use
force. These findings suggest that the use of force over citizens is a func-
tion of officers' varying commitments to the traditional culture of policing.
Moreover, what is driving this finding are those officers who reject tradi-
tional notions of culture (i.e., the con-culture group). The difference
between officers who most subscribe to traditional culture and those with
mixed views toward traditional culture are indistinguishable in statistical
terms.
Finally, because of the potential confounding influence of varying cul-
tures between the two cities, ordered probits were estimated separately for
each department, including in the model the culture measure and all of the
control variables (Table 8). This permits the opportunity to determine
whether the various culture groupings interact differently across depart-
ments. For instance, are pro-culture oriented officers in Indianapolis sig-
nificantly more likely to resort to force compared with pro-culture
oriented officers in St. Petersburg?
As demonstrated in Table 8, there is no support for the belief that
officers' attitudinal orientations toward culture differ in terms of coercive
behavior across departments. Pro-officers in both cities were significantly
more likely to use force compared with the con-culture group reference
category. Similarly, mid-officers in both cities were also more likely to
rely on force, although only St. Peterburg officers reached statistical signif-
icance. Nonetheless, as shown in the table's last column, neither coeffi-
cient was statistically distinguishable from its counterpart at the other site
at the conventional standard of p < .05 (requiring a t-value of 1.96 or
higher). Further, we see that there are few differences across any of the
predictors. Only arrest, suspect age, and the number of officers on the
scene produced differences across sites.
Predicted Probabilities. Another way to view the effects of the various
determinants is to consider predicted probabilities on each level of force.
Given the difficulty of interpreting ordered probit estimates, predicted
probabilities are used because of intuitive appeal and for direct compari-
sons between categories among independent variables (Long, 1997). They
offer a means to more easily grasp the effect of independent variables on
each of the response categories of the dependent measure.
Table 9 presents predicted probabilities for those factors found to be
significantly related to force (in the combined model). For example,
officers were most likely to deal with nonresistant suspects with no force
or verbal force. The probability of officers using some form of physical
force was .16 where the suspect was nonresistant. Conversely, suspects
 POLICE CULTURE AND .COERCION 1027

Table 8. Ordered Probit Estimates-Split Sites

Indianapolis St. Petersburg
Variable bt S.E. b2 S.E. tbl-b2

Culture
Pro .200* (.072) .198* (.079) 0.02
Mid .107 (.084) .232* (.087) -1.03
Control
Resistance .365* (.044) .357* (.058) 0.11
Safety .906* (.216) .439 (.244) 1.43
Conflict .071 (.046) .193* (.061) -1.60
Arrest .964* (.092) 1.482* (.111) -3.59
Evidence .077* (.017) .056* (.018) 0.85
Male .318* (.064) .202* (.070) 1.22
Non-white .199* (.060) .125* (.066) 0.83
Age -. 131" (.022) -. 066* (.024) -1.99
Wealth -. 115" (.051) -. 158* (.055) 0.57
Demeanor -. 043 (.107) -. 070 (.103) 0.18
Drug/alcohol .303* (.072) .376* (.078) -0.69
# Officers .105* (.019) .001 (.025) 3.31
# Bystanders -. 004 (.005) .004 (.006) -1.02
Violence anticipated -. 001 (.105) .038 (.115) -0.25
Problem Type .098 (.055) -. 037 (.067) 1.55
Proactive .112 (.057) .279* (.067) -1.89
Chi-square 583.328, p < .001 425.803, p < .001
Intercept -. 288 (.222) -. 606 (.254)
N 1814 1409
Pseudo-R-Square .275 .261
NOTE: b represents the regression coefficient; S.E. represents the standard error of each
coefficient; * indicates statistical significance at p < .05; tbl-b2 indicates t-difference test.

displaying increased levels of resistance were more likely to receive

increased levels of force. For instance, a suspect displaying defensive
resistance had a 52% chance of receiving physical force, whereby an
actively resistant suspect had nearly a 64% chance. Pronounced differ-
ences emerged in several of the other control measures as well. For
instance, suspects possessing some sort of weapon (i.e., an officer safety
issue) were over twice as likely to be physically restrained compared with
those who did not (38% versus 17%, respectively). Predicted probabilities
for the remaining control variables demonstrate similar, although gener-
ally less pronounced, differences.
Turning to the primary variable of interest, Table 9 shows probability
differences in the pro- and mid-culture groupings. Encounters involving
officers with pro-cultural attitudinal dispositions were more likely to
involve each form of force. Here, the greatest difference is found in physi-
cal restraint. Encounters involving pro-cultural officers had nearly a 20%
1028 TERRILL ET'AL

Table 9. Predicted Probabilities

Level of Force
No Force Verbal Restraint Impact
Culture
Pro
Yes .355 .445 .195 .005
No .432 .420 .145 .003
Mid
Yes .348 .447 .200 .005
No .407 .430 .160 .003
Control
Resistance
None .420 .424 .153 .003
Passive .290 .456 .246 .008
Verbal .183 .439 .358 .020
Defensive .104 .379 .472 .045
Active .053 .311 .547 .089
Safety
Yes .166 .431 .380 .023
No .396 .433 .168 .003
Conflict
None .397 .433 .166 .004
Calm Verbal .355 .445 195 .005
Agitated Verbal .315 .453 .225 .007
Threat .276 .457 .258 .009
Assault .240 .455 .293 .012
Arrest
Yes .092 .364 .493 .051
No .443 .415 .140 .002
Evidence
Level 1 .399 .432 .165 .004
Level 3 .351 .446 .198 .005
Level 7 .261 .456 .273 .010
Male
Male .364 .443 .188 .005
Female .465 .405 .128 .002
Nonwhite
Nonwhite .369 .442 .185 .004
White .431 .420 .146 .003
Age
13-17 Years Old .312 .454 .227 .007
21-29 Years Old .383 .437 .176 .004
45-59 Years Old .459 .408 .131 .002
Wealth
Chronic Poverty .320 .452 .221 .007
Low .372 .441 .183 .004
Middle .426 .422 .149 .003
Above Middle .482 .397 .119 .002
Drug/Alcohol Use
Yes .293 .456 .243 .008
No .419 .425 .153 .003
Officers on Scene
Two .397 .433 .166 .004
Four .348 .447 .200 .005
Six .300 .455 .237 .008
Proactive
Yes .351 .446 .198 .005
No .425 .422 .150 .003
Site
Yes .365 .442 .188 .005
No .426 .422 .149 .003
 POLICE CULTURE AND COERCION 1029

chance of physical restraint, compared with 15% in those encounters

involving non-pro-cultural officers. Comparatively, differences in verbal
and impact force were relatively small.
Encounters involving officers designated as mid-cultural were also more
likely to involve each form of force. As illustrated in Table 9, in
encounters involving mid-cultural officers, there was a 65% chance of
force being applied, whereas encounters involving non-mid-cultural
officers had about a 59% chance of force being used. Again, the greatest
difference is observed in physical restraint (i.e., 20% versus 16%,
respectively).

DISCUSSION

The present analysis illustrates that officers' attitudinal differences

toward the traditional view of police culture produce differences in coer-
cive actions over suspects. Officers who embody the values of the tradi-
tional police culture, or have mixed views toward the culture, were more
likely to use coercion compared with officers with nontraditional cultural
attitudes. In addition, although officers from Indianapolis (the depart-
ment that emphasized aggressive order maintenance) were more likely to
embody the values of the traditional culture, compared with officers from
St. Petersburg (the department that promoted a problem-solving/commu-
nity partnership approach), this had no effect in terms of applying coer-
cion. Pro-culture-oriented officers in both departments were more likely
to use force, but the difference between the two departments was statisti-
cally indistinguishable. Interestingly, however, officers in Indianapolis, on
the whole, were more likely to resort to force. Hence, it appears that the
style of policing promoted by top management in Indianapolis (i.e.,
aggressive order maintenance) does seem to matter. Nonetheless, officers
who subscribe to traditional notions of culture relied on coercion more
readily than those who do not, irrespective of the style of policing pro-
moted by top leadership.
In many respects, our analysis suggests that although broad-based,
almost caricatures of "police culture" have general utility, they are mis-
leading. Such caricatures have formed the basis for conventional wisdom
for over four decades. We agree that the occupation contains strains as
well as the organization in which officers work, but the ways in which
officers handle these strains appear to be more fragmented than depicted
by many police researchers. The findings reported here from the cluster
analysis support the notion that some groups of officers (i.e., the pro-cul-
ture group) are traditional culture carriers, but we also find variation in
adherence to "the" police culture (i.e., mid- and con-culture groups). As
1030 TERRILL ET AL.

such, reports of a universally shared police culture might have been over-
stated by police scholars. In this sense, this study adds to our understand-
ing of the complexities of culture, some of which researchers are now
beginning to acknowledge (see, for example, Herbert, 1998; Manning,
1994; Paoline, 2001).
Unless the matter at hand, what we have called awkwardly perhaps,
"the traditional view of police culture," is conceptualized more finely,
measured carefully, and associated with specific behavioral outcomes, it
will remain a misleading gloss on complexity. Typology research con-
ducted in the 1970s hinted at some of the variation among officers, but
systematic connections to potential differences in behavior were nonexis-
tent. Links between themes in the culture and specific outcomes are
needed rather than generalities about what police do, or feel, or their per-
sonalities. In this study, we have attempted to forge these vital links, and
future research should continue such endeavors. Moreover, as prior
research suggests (Reiss, 1972; Worden, 1995), situational- and encounter-
based studies help to disentangle the nuances of police behavior more
than relying merely on broad-based studies of generalized attitudes or
descriptions of extraordinary and rare violent events. As this implies, and
as our analysis shows, coercion is an interactive matter that takes place in
encounters or situations that vary and must be taken into account.
Although we believe these findings contribute to an overall understand-
ing of the relationship between culture and coercion, our conclusions
come with some caution. Although the attitudes included in our classifica-
tion scheme of officers are all rooted in the traditional view of police cul-
ture, there are some attitudes that were not captured (e.g., peer group
loyalty, social isolation, perceptions of danger, etc.). Further, the interplay
between the style of policing promoted by top leaders and officers' views
of culture is not sufficiently clear and requires additional attention. In
addition, as with any study of a small number of nonrandomly selected
departments, our ability to generalize from these findings is !imited.
Future research should work to add to the "incompleteness" of the current
research in contributing to a richer understanding of police culture and its
behavioral implications. This would undoubtedly contribute to cumulative
knowledge-building in the area of policing.
Finally, researchers should work to disentangle potential socializing
influences as transmitters of culture (e.g., peers, field training officers,
supervisors), as well as examine additional behaviors that are linked to
police culture (e.g., variations in citizen complaints, citizen support,
arrests, etc.), especially as the work environments continue to change dur-
ing the community era. Given the potential negative consequences of
coercive police-suspect incidents (e.g., highly publicized incidents, citizen
 POLICE CULTURE AND COERCION 1031

complaints, etc.), researchers and practitioners alike need to continually

focus on ways in which to reduce this form of behavior.

REFERENCES

Alpert, Geoffrey P. and Roger G. Dunham

1997 The Force Factor: Measuring Police Use of Force Relative to Suspect Resis-
tance. Washington, D.C.: Police Executive Research Forum.
Broderick, John J.
1977 Police in a Time of Change. Morristown, N.J.: General Learning Press.
Brown, Michael K.
1988 Working the Street: Police Discretion and the Dilemmas of Reform. New
York: Russell Sage Foundation.
Bryk, Anthony S. and Stephen W. Raudenbush
1992 Hierarchical Linear Models: Applications and Data Analysis Methods.
Newbury Park, Calif.: Sage.
Chan, Janet
1996 Changing police culture. British Journal of Criminology 36:109-134.
Crank, John P.
1997 Celebrating agency culture: Engaging a traditional cop's heart in organiza-
tional change. In Quint C. Thurman and Edmund McGarrell (eds.), Com-
munity Policing in a Rural Setting. Cincinnati, Ohio: Anderson Publishing
Company.
Crank, John P. and Michael Caldero
1991 The production of occupational stress in medium-sized police agencies: A
survey of line officers in eight municipal departments. Journal of Criminal
Justice 19:339-349.
Everitt, Brian S.
1980 Cluster Analysis. New York: Wiley.
Fielding, Nigel G.
1988 Joining Forces: Police Training, Socialization and Occupation Competence.
London: Routledge.
Frank, James and Steven S. Brandl
1991 The police attitude-behavior relationship: Methodological and conceptual
considerations.American Journal of Police 5:83-103.
Fyfe, James J.
1988 The Metro-Dade Police-Citizen Violence Reduction Project, Final Report,
Executive Summary. Washington, D.C.: Police Foundation.
Garner, Joel H., John Schade, Thomas Hepburn, and John Buchanan
1995 Measuring the continuum of force used by and against the police. Criminal
Justice Review 20:146-168.
Herbert, Steve
1996 Morality in law enforcement: Chasing 'bad guys' with the Los Angeles
police department. Law & Society Review 30:798-818.
1998 Police subculture reconsidered. Criminology 36:343-369.
1032 TERRILL ET AL.

Jermier, John M., John W." Slocum, Jr., Louis W. Fry, and Jeannie Gaines
1991 Organizational subcultures in a soft bureaucracy: Resistance behind the
myth and facade of an official culture. Organizational Science 2:170-194.
Kappeler, Victor E., Richard D. Sluder, and Geoffrey P. Alpert
1998 Forces of Deviance: Understanding the Dark Side of Policing. 2d ed. Pros-
pect Heights, Ill.: Waveland Press.
Kelling, George and Robert B. Kliesmet
1996 Police unions, police culture, and police use of force. In William A. Geller
and Hans Toch (eds.), Police Violence: Understanding and Controlling
Police Abuse of Force. New Haven, Conn.: Yale University Press.
Klinger, David A.
1995 The micro-structure of nonlethal force: Baseline data from an observational
study. Criminal Justice Review 20:169-186.
Long, Scott J.
1997 Regression Models for Categorical and Limited Dependent Variables.
Thousand Oaks, Calif.: Sage.
Manning, Peter K.
1994 Police occupational culture: Segmentation, politics, and sentiments. Unpub-
lished manuscript, Michigan State University.
1995 The police occupational culture in Anglo-American societies. In W. Bailey
(ed.), The Encyclopedia of Police Science. New York: Garland.
Mastrofski, Stephen D., Roger B. Parks, Albert J. Reiss, Jr., Robert E. Worden,
Christina DeJong, Jeffrey B. Snipes, and William Terrill
1998 Systematic Observation of Public Police: Applying Field Research Methods
to Policy Issues. Washington D.C.: National Institute of Justice.
Mok, Magdalena and Marcellin Flynn
1998 Effect of catholic school culture on students' achievement in the higher
school certificate examination: A multilevel path analysis. Educational Psy-
chology 18:409-432.
Muir, William K., Jr
1977 Police: Streetcorner Politicians. Chicago, Ill.: University of Chicago Press.
Neter, John, William Wasserman, and Michael H. Kutner
1985 Applied Linear Statistical Models: Regression, Analysis of Variance, and
Experimental Designs. Homewood, Ill.: Richard Irwin.
Nourusis, Marija J.
1990 SPSS Base System User's Guide. Chicago, Ill.: SPSS, Inc.
Paoline, Eugene A., III
2001 Rethinking Police Culture: Officers' Occupational Attitudes. New York:
LFB Publishing.
Paoline, Eugene A., III, Stephanie M. Myers, and Robert E. Worden
2000 Police culture, individualism, and community policing: Evidence from two
police departments. Justice Quarterly 17:575-605.
Parks, Roger B., Stephen D. Mastrofski, Christina DeJong, and M. Kevin Gray
1999 How officers spend their time with the community. Justice Quarterly
16:483-518.
 POLICE CULTURE AND COERCION 1033

Reiner, Robert
1985 The Politics of the Police. New York: St. Martin's Press.
Reiss, Albert J., Jr.
1972 The Police and the Public. New Haven: Yale University Press.
Reuss-Ianni, Elizabeth
1983 Two Cultures of Policing. New Brunswick, N.J.: Transaction.
Rokeach, Milton
1972 Beliefs, Attitudes, and Values: A Theory of Organization and Change. San
Francisco, Calif.: Jossey-Bass.
Rubinstein, Jonathan
1973 City Police. New York: Farrar, Straus, and Giroux.
Sampson, Robert J., Stephen W. Raudenbush, and Felton Earls
1997 Neighborhoods and violent crime: A multilevel study of collective efficacy.
Science 277:918-924.
Sherman, Lawrence W.
1980 Causes of police behavior: The current state of quantitative research. Jour-
nal of Research in Crime and Delinquency 17:69-100.
Skolnick, Jerome H.
1994 Justice Without Trial: Law Enforcement in Democratic Society. 3d ed. New
York: Wiley.
Snipes, Jeffrey B. and Stephen D. Mastrofski
1990 An empirical test of Muir's typology of police officers. American Journal of
Criminal Justice 14:268-296.
Sparrow, Malcolm K., Mark H. Moore, and David H. Kennedy
1990 Beyond 911: A New Era for Policing. New York: Basic Books.
Terrill, William
2001 Police Coercion: Application of the Force Continuum. New York: LFB Pub-
lishing.
Terrill, William and Stephen D. Mastrofski
2002 Situational and officer based determinants of police coercion. Justice Quar-
terly 19:101-134.
2003 Working the street: Does community policing matter? In Wesley Skogan
(ed.), Community Policing: Can it Work?. Belmont, Calif.: Wadsworth.
Toch, Hans
1969 Violent Men: An Inquiry into the Psychology of Violence. Chicago, Ill.:
Aldine.
Van Maanen, John
1974 Working the street: A developmental view of police behavior. In Herbert
Jacob (ed.), The Potential for Reform of Criminal Justice. Beverly Hills,
Calif.: Sage.
Westley, William A.
1970 Violence and the Police: A Sociological Study of Law, Custom, and Moral-
ity. Cambridge, Mass.: MIT Press.
White, Susan 0.
1972 A perspective on police professionalization. Law and Society Review
7:61-85.
1034 TERRILL ET AL.

Worden, Robert E.
1995 Police officers' belief systems: A framework for analysis. American Journal
of Police 14:49-81.
1996 The causes of police brutality: Theory and evidence on police use of force.
In William A. Geller and Hans Toch (eds.), Police violence: Understanding
and controlling police abuse of force. New Haven, Conn.: Yale University
Press.

William Terrill is an Assistant Professor in the College of Criminal Justice at North-

eastern University. He has directed several nationally funded research projects and has
published numerous scholarly articles on policing, crime in public housing, and system-
atic social observation. In 2001, he authored Police Coercion:Application of the Force
Continuum (LFB Publishing). He earned his Ph.D. in 2000 from the School of Criminal
Justice at Rutgers University, Newark.
Eugene A. Paoline III is an Assistant Professor in the Department of Criminal Jus-
tice and Legal Studies at the University of Central Florida. His research interests
include police socialization and culture, occupational stressors among criminal justice
practitioners, and theoretical development in criminal justice. He is the author of
Rethinking Police Culture: Officers' OccupationalAttitudes (2001, LFB Publishing) and
holds a Ph.D. in Criminal Justice from the University at Albany, State University of
New York.
Peter K. Manning received a Ph.D. from Duke in 1966. He holds the Elmer V. H.
and Eileen M. Brooks Chair in Policing at Northeastern University, College of Criminal
Justice. He has been a fellow of NIJ, the Rockefeller Foundation, Balliol and Wolfson
Colleges, Oxford, and the Socio-Legal Centre at Oxford. He is listed in Who's Who in
America and has been awarded many contracts and grants. He is the author and editor
of some 13 books. His research interests include the rationalizing of policing, crime
mapping and crime analysis, uses of information technology, and qualitative methods.
The second edition of Narc's Game (1979) will appear in 2003 (Waveland Press), and
his monograph, Policing Contingencies, is forthcoming with the University of Chicago
Press.