Policing domestic disputes in the south


Material Information

Policing domestic disputes in the south a study of Turk's norm resistance theory
Physical Description:
xi, 191 leaves : ; 29 cm.
Greenleaf, Richard G
Publication Date:


Subjects / Keywords:
Family violence -- South Carolina -- Charleston   ( lcsh )
Family violence -- Georgia -- Savannah   ( lcsh )
Police -- South Carolina -- Charleston   ( lcsh )
Police -- Georgia -- Savannah   ( lcsh )
Arrest -- South Carolina -- Charleston   ( lcsh )
Arrest -- Georgia -- Savannah   ( lcsh )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1993.
Includes bibliographical references (leaves 179-190).
Statement of Responsibility:
by Richard G. Greenleaf.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001923473
notis - AJZ9307
oclc - 30541248
System ID:

Full Text








I would first like to thank my supervisor Professor

Lonn Lanza-Kaduce, for his unwavering support, optimism,

patience and judgment. Without his inspiration, this

dissertation could not have gotten off the ground. He

enriched my sociological imagination, guiding me through

both methodological and theoretical mazes. His confidence

in me was unyielding; it was my privilege to have worked

with him for the last several years. The many distinguished

members of my committee--Professors Ronald Akers, Charles

Frazier, Benjamin Gorman and Gerardo Gonzalez--deserve my

enduring gratitude for their thoughtful comments and their

help in directing me around the potential mines of this


This study would have been impossible without the

permission of Chief Reuben Greenberg of the Charleston, SC

Police Department, and Dr. Michael Donahue and Chief David

Gellatly of the Savannah, GA Police Department. These

enlightened professionals provided me with the ideal

environment for research and I am indebted to them.

Many of the members of the Charleston and the Savannah

Police Departments have my appreciation. Special mention is

due to Charles Francis, Annette Nelson, Elaine Sherman, Lt.


Richard Hazel, Cpl. Donald Daquigan and Pfc. Dennis Potter

in Charleston. The data from Savannah could not have been

collected without the assistance of Maj. Dan Reynolds,

Palmer Davis and the Records Unit staff.

While completing this research I was on the faculty at

North Georgia College. I thank Professors James Wells, Ray

Rensi, Lee Downing and Linda August for all their

encouragement and support and Vicki Dowdy for her

administrative skills. My good friend and colleague Dr.

Barry Friedman, a political scientist, spent many weekends

reading drafts of a "foreign language" but always providing

me with sage advice.

I also owe a debt of gratitude to my family, David and

Tiffany Algaze, Fayzee Andria and Seymour Greenleaf. Thanks

are also extended to Linda, Ilysa and Kendra Lanza-Kaduce

for sharing their husband and father with me during the last

two years. There is one person, however, who made this

dissertation possible and always provided a receptive heart

and timely words of encouragement. My mother, Jessica

Algaze, who constantly encouraged me to reach for the stars

and achieve the impossible. She made enormous sacrifices

and shared all of the ups and downs over the last six years.

Without her love and contribution, "we" would never have

completed this dissertation.

In a final note, I recall my brother, Kenny. Although

he is gone, he provided me with the spiritual inspiration I


needed to complete this dissertation. I often felt his

presence during the long hours at the computer screen when I

felt like sleeping. Before starting my doctorate, I told

him that it would take me too long to complete a Ph.D.,

possibly five years. He replied, "How old will you be in

five years if you don't complete it?" I dedicate this

dissertation to the memory of my brother, Kenneth Greenleaf.



ACKNOWLEDGMENTS............ ............................. ii

LIST OF TABLES........................... ........ ...... vii

ABSTRACT................................................. x


1 INTRODUCTION........................ ................... 1

Statement of the Problem........................... 1
Summary............................................. 7

2 REVIEW OF LITERATURE.............................. 8

Introduction ..................................... 8
Police Use of Nonlethal Force....................... 9
Police Use of Lethal Force.......................... 15
Assaults on Police.................................. 22
Resisting, Refusing and Interfering................. 26
Domestic Disputes.................................. 28
Proarrest .......................................... 38
The Limitations of Past Research.................... 40

3 TURK ON NORM RESISTANCE............................ 43

Theoretical Framework.............................. 43
Critiques of Turk's Theory.......................... 49
Turk's Conflict Theory at Domestics............... 54

4 RESEARCH METHODOLOGY............................... 69

Charleston, SC ..................................... 71
Independent Variables.......................... 75
Data Analysis....... ....... ......... ... ..... 79
Savannah, GA........................................ 85

5 RESULTS............................................ 89

Charleston, SC............... .................... 89
Descriptive Statistics......................... 89

Crosstabulations............................. 96
Basic Culture Conflict........................ 96
Sophistication Indicators..................... 107
Organization Indicators....................... 114
Summary......................................... 125
Multivariate Analyses......................... 128
Savannah, GA........................................ 142

6 SUMMARY AND CONCLUSIONS............................ 145

Exhaustiveness or Theoretical Correspondence....... 146
Falsifiability and the Generation of Research
Hypotheses .......................................... 153
Policy Implications... ............................ 159
Internal Consistency--Problems of Theoretical
Specifications................................. 165


A POLICE REPORTS.. ................................. 174
B OTHER ANALYSES... ................................ 177

REFERENCES.................................................. 179

BIOGRAPHICAL SKETCH..................................... 191


Table page

1 Situational Variables and Officer Injuries
(Crosstabular Analyses)............................. 34

2 Situational Variables and Officer Injuries
(Regression Model) .................................. 35

3 Expected Relative Probabilities of Normative-
Legal Conflict.......................... ................. 50

4 Frequency Distribution Related to Basic Culture
Conflict and Cultural Norms of Subjects at
Domestic Disputes Charleston, SC Police, 1988 to
1991................................................. 91

5 Frequency Distribution of Sophistication
Indicators at Domestic Disputes Charleston, SC
Police, 1988 to 1991................................ 92

6 Frequency Distribution of Organization Indicators
at Domestic Disputes, Charleston, SC Police, 1988 to
1991................ .... ............................. 94

7 Arrest History of Subjects By Norm Resistance
Charleston, SC Police Department, 1988-1991.......... 97

8 Arrest History of Subjects for a Crime of Violence
or Disorder By Norm Resistance Charleston, SC Police
Department, 1988-1991 .............................. 97

9a Differences in Age of Officers and Subjects by Norm
Resistance Charleston, SC Police Department,
1988-1991 ........................................... 99

9b Officer-Subject Age by Norm Resistance Charleston,
SC Police Department, 1988-1991.................... 100

10a Differences in Race of Officers and Subjects By
Norm Resistance Charleston, SC Police
Department, 1988-1991.............................. 102

10b Officer-Subject Race By Norm Resistance Charleston,
SC Police Department, 1988-1991.................... 103


lla Differences in Gender of Officers and Subjects By
Norm Resistance Charleston, SC Police
Department, 1988-1991.............................. 105

llb Officer-Subject Gender By Norm Resistance
Charleston, SC Police Department, 1988-
1991................................. ............... 106

12 Officer Experience By Norm Resistance Charleston,
SC Police, 1988-1991............................... 108

13 Officer Age By Norm Resistance Charleston, SC Police,
1988-1991......... ................................ 109

14 Intoxication of Subjects By Norm Resistance
Charleston, SC Police Department, 1988-1991 ....... 110

15 In Progress Assault By Norm Resistance Charleston,
SC Police Department, 1988-1991.................... 112

16 How Dispute Came to Attention of Police By Norm
Resistance Charleston, SC Police Department, 1988-
1991............................................... 113

17 Relationship of Disputants by Norm Resistance
Charleston, SC Police Department, 1988-1991........ 115

18 Employment Status of Victim by Norm Resistance
Charleston, SC Police Department, 1988-1991........ 117

19 Visible Injuries To Victim By Norm Resistance
Charleston, SC Police Department, 1988-1991 ...... 118

20 Number of Disputants Arrested By Norm Resistance
Charleston, SC Police Department, 1988-1991 ....... 120

21a Presence of Witnesses By Norm Resistance
Charleston, SC Police Department, 1988-1991........ 121

21b Presence of Witnesses By Norm Resistance By
Location Charleston, SC Police Department,
1988-1991 .......................................... 122

22 Location of the Dispute By Norm Resistance
Charleston, SC Police Department, 1988-1991........ 124

23 Seriousness of the Charge By Norm Resistance
Charleston, SC Police Department, 1988-1991....... 124

24 Full and Reduced Logistic Regression Models of Factors
Relating to Norm Resistance, Charleston, SC Police
Department, 1988-1991.............................. 132


25 Classification Tables for Norm Resistance Charleston,
SC Police Department, 1988-1991.................... 133

26 Number of Culture Conflict Indicators By Norm
Resistance Charleston, SC Police Department, 1988-
1991............................... ................. 136

27 Number of Sophistication Indicators By Norm
Resistance Charleston, SC Police Department, 1988-
1991................................................. 137

28 Number of Organization Indicators By Norm
Resistance Charleston, SC Police Department, 1988-
1991................................................ 138

29 Standardized Discriminant Coefficients and
Classification Results, Charleston, SC Police
Department, 1988-1991.............................. 140

30 Presumptive Arrest By Norm Resistance Savannah,
GA Police Department--January 1, 1987 February 28,
1988............................................... 143

31 Bivariate Relationships from the Charleston Study
and Selected Research on Police-Citizen Conflict... 147

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy


Richard G. Greenleaf

August 1993

Chairman: Lonn Lanza-Kaduce
Major Department: Sociology

This study applying Austin Turk's norm resistance

theory examines police-citizen encounters at domestic

disputes in Charleston, SC and Savannah, GA. Turk's theory

is abstract enough to be able to account for various forms

of conflict (e.g., resisting arrest, verbal assaults,

obstructing an officer, deadly force). He argues that

normative conflict is inherent in all authority-subject

relations and is more likely to become overt when both

subjects and authorities act congruently with their beliefs,

when neither subjects or authorities are sophisticated in

their abilities to manipulate the situation, and when the

subjects are organized. Data for this research was obtained

from official police reports, police personnel records and

criminal histories of the batterers.

In Charleston, officers are more inclined to encounter

conflict when batterers are visibly intoxicated and the


disputes are still in-progress. This is associated with the

concept of sophistication or the knowledge and the ability

to manipulate the opposition. Police-citizen conflict is

also more likely when police arrest more than one disputant.

This is associated with the concept of organization or the

social support obtained from others present. Officers who

are younger, nonwhite or female may be more likely to

confront norm resistance at domestic calls when dealing with

older, white, male batterers. The emerging pattern can be

understood by the phenomenon of social deference or

deference reversals. Failing to defer automatically to the

officer's authority, some subjects instead evaluate the

officer's social status (age, race and gender) in deciding

whether to submit.

The final hypothesis that was generated for this

dissertation focused on the congruence between cultural and

social norms. When consistent, overt conflict is more

likely to follow. A domestic violence proarrest policy was

examined in Savannah, GA. The results indicate that patrol

officers continued to use discretion after administrators

instituted the proarrest policy. The unintended

consequences (i.e., authority-subject conflict) of the

policy could not be evaluated.


Statement of the Problem

Overt conflict between authorities and subjects may

take a variety of forms. These include: (1) the use of

deadly force by authorities (Blumberg, 1983; Croft, 1985;

Donahue, 1983; Fyfe, 1978, 1981; Geller & Karales, 1981;

Jacobs & Britt, 1978; Kania & Mackey, 1977, 1992; Sherman &

Langworthy, 1979), (2) the use of nonlethal force by

authorities (Croft, 1985; Friedrich, 1980), (3) assaults on

authorities by subjects (Choi, 1989; Geller & Karales, 1982;

Margarita, 1980; Uchida & Brooks, 1988), and (4) resistance,

interference or obstruction by subjects (Croft, 1985; Reiss,

1980). While the homicides of subjects or authorities can

be tragic, there are many forms of overt conflict that have

not received much notice by criminologists. Croft (1985)


[D]espite the importance attached to the issue of
police use of force, very little systematic and
reliable information is available on the subject.
Aside from a few studies on police use of force,
the research emphasis has been on "deadly force"--
police use of their firearms.1 (p. 41)

1 Croft (1985) advised that one of the major problems
researchers face who want to study police use of force is
the difficulty in obtaining information. "Only a handful of



The objective of this dissertation is to shed light on

authority-subject conflict. The research is done within the

context of domestic disturbances, which have been

traditionally viewed as one of the most explosive calls for

the police (Bard, 1974; Blount et al., 1992; Buzawa &

Buzawa, 1990, 1992; Davis, 1983; Ellis, 1987; Muir, 1977).

The types of overt conflict examined range from verbal

threats to lethal force.

In order to understand conflict between authorities and

subjects, it is insufficient to focus exclusively on deadly

force, which is rather infrequent and may be highly

politicized. What is called for is a framework that can

assist us in understanding all forms of conflict. One of

the theories that is abstract enough to be able to handle

conflict on several levels is Austin Turk's (1966, 1969)

theory of norm resistance.

Turk has been called "the most complex and unique

theorist within the conflict tradition" (Orcutt, 1983, p.

322). He has discussed the power of the authorities (e.g.,

police) and the use of coercion when subjects refuse to

grant deference.

There are those in society who constitute the
dominant, decision-making category--the
authorities--and those who make up the subordinate

police departments in the United States maintain records on
use of force incidents and even where such records are
maintained, they may not be available for scholarly
research" (p. 48).


category, who are affected by but scarcely affect
law--the subjects. (Turk, 1969, p. 33)

Unlike radical or critical criminologists (Platt, 1974;

Quinney, 1980), Turk (1969) does not see society as neatly

distributed between a ruling elite and a powerless mass;

instead he advocates a pluralistic conflict theory. In

addition, Turk dismisses ideology and focuses on a

scientific and objective conflict theory.

Enthusiasts of the idea that conflict theory is
and should be a partisan critique generally go on
to advance the self-defeating thesis that any
theory is nothing more than an ideology. (Turk,
1979, p. 459)

Akers has acknowledged that Turk's theory is loyal to

the conflict tradition and "capable of and intended for

testing free of its ideological freight" (1979, p. 539).

Like Dahrendorf (1959), Turk (1969) focuses on differences

of power rather than the means of production or economics

(see discussion of Turk in Akers, 1979; Lilly et al., 1989;

Liska, 1981; Orcutt, 1983; Vold & Bernard, 1986). He

focuses on the use of legal power to coerce subjects but not

the origins of that power.

Turk's (1966, 1969) use of the term "norm resistance,"

refers to overt conflict between authorities and subjects.

Turk's theoretical abstraction permits his arguments to be

extended in ways that allow drawing new and meaningful

hypotheses about police-citizen conflict.

Turk's (1966, 1969) conflict theory addresses extra-

legal factors, (e.g., employment, sex, age, class,


education, immigration) and the differences between

authorities and subjects. The potential for conflict exists

when authorities and subjects are dissimilar in race, age,

gender, education, and nationality. Conflict is more likely

to become overt when the authorities and/or subjects'

cultural norms, i.e., laws or beliefs, are congruent with

their social norms, i.e., actual behavior. A result of the

congruence of cultural and social norms is less compromise

and an increase in the potential for violence. For

example, males who believe they are the "kings of their

castle" (cultural norm) and use physical force to control

their spouses and/or family members (social norm) are likely

to have more difficulty deferring to authorities who must

arrest (under mandatory arrest policies).2 A thorough

discussion of Turk's theory is provided in a separate

chapter. At present, it is enough to highlight some basic


Police use of extra-legal factors for arrest decisions

has been well documented (Black, 1980; Ferraro, 1989;

Lundman, 1974; Piliavin & Briar, 1964; Powell, 1990; Smith &

Klein, 1984). To date, qualitative methods and

interactionist theories have been used to examine specific

exchanges (Davis, 1983; Manning, 1980; Margarita, 1980;

2 Gelles and Strauss believe that "the most telling of
all attributes of the battering man is that he feels
inadequate and sees violence as a culturally acceptable way
to be both dominant and powerful" (1988, p. 89).


Toch, 1969; Reiss, 1974). Turk takes a unique approach by

suggesting that rates of overt conflict should vary across

"structured situations." He directs us to examine the

variables that structure the situations, rather than look at

social psychological meanings of the participants or

symbolic details of particular interactions.

A sociological theory of interaction is required;
moreover, it must be a theory of interaction among
groups and categories of people rather than a
social psychological explanation of patterns in
interactions among individuals, even though
psychological assumptions will necessarily be
implicit in any such theory. (Turk, 1969, p. 53)

The police response to domestic disturbances has been

at issue, particularly since Sherman and Berk's (1984)

Minneapolis Experiment found that arrest reduced the

recidivism of batterers.3 Subsequently, police chiefs,

legislators and criminologists began to focus more on the

domestic dispute. Sherman and Berk's research has been

linked with many of the changes in police practices and

policies across the nation. These new policies advocate

arrest and are based on the concept of deterrence.4

3 The terms domestic disturbance, domestic dispute,
domestic violence or family fight may be used
interchangeably. These may involve husband-wife, boyfriend-
girlfriend, same sex partners or any combination of family
members or relatives. In general, this study relies on
police reports that label the call a "domestic."

4 Some studies refer to a mandatory arrest policy,
while others use the term presumptive arrest. The former
generally compels police to arrest the batterer, while the
latter suggests it. In either case, the officer may arrest
without witnessing the assault take place. Prior to these
new policies, officers could only advise the victim to file


According to Ferraro (1989, p. 61), "[T]he changes in the

statutes are a response to the fact that most assaults

against wives occur in private settings and fall in the

misdemeanor category." Recently, researchers have

questioned the conclusions of the Minneapolis Experiment and

have stated that formal arrest has no effect on recidivism

(Dunford, 1992; Dunford et al., 1990; Hirschel et al., 1992;

Pate & Hamilton, 1992; Sherman et al., 1992). However,

police officials at all levels remain convinced that

proarrest policies are effective (Blount et al., 1992;

Bouza, 1991). A South Carolina chief attributed a reduction

in the homicides in his city to a mandatory arrest policy.5

A recent study in a Midwestern city found that 88% of the

officers surveyed supported mandatory arrest (Friday et al.,


While there is disagreement over proarrest and

recidivism, its unintended consequences have been largely

ignored. After their study of violence toward officers in

Baltimore County, MD, Uchida and Brooks concluded:

A research question that should be examined
in the future is the extent to which different

a complaint, act as a mediator or separate the disputants
for the night unless the offense was a felony or occurred in
the presence of the officer. Both presumptive and mandatory
arrest are considered proarrest or preferred arrest policies
and depart from the in-presence requirement for misdemeanor

5Personal communication with Chief Reuben Greenberg
of the Charleston, SC Police Department on September 3,


police methods affect the number of officer deaths
and injuries taking place during domestic
disturbances. This question is especially
important today, given the increasing demand for
police departments to adopt an arrest approach for
dealing with domestic disturbances. (1988, p. 102)


In this dissertation, police-citizen conflict is

examined at domestic disturbances in Charleston, SC and in

Savannah, GA. Turk's (1969) theory of norm resistance is

used to help us understand why some disputes result in overt

conflict between officers and citizens, while others do not.

A domestic violence proarrest policy is then examined in

Savannah. Has presumptive arrest increased the probability

of police-citizen conflict, as the officers' cultural norms

and social norms become congruent?



Although the problem of conflict between officers and

citizens is examined within the framework of domestic

disputes, the findings are germane to several other police-

citizen encounters (e.g., arrests, general disturbances,

loud parties, fight calls). There has been a limited amount

of scholarly research of police-citizen violence at domestic

disturbances; however, more is known about police-citizen

conflict in other contexts (e.g., arrests, police use of

deadly force).

This section begins by reviewing research on police use

of nonlethal force. Many of the early studies were

conducted through direct observation; they primarily focused

on police discretion and the decision to arrest (Black &

Reiss, 1970; Lundman, 1974; Piliavin & Briar, 1964;

Skolnick, 1966; Sykes & Clark, 1980; Westley, 1970). In the

early 1980s, quantitative studies started to emerge, and a

popular subject was the police use of lethal as well as

nonlethal force (Blumberg, 1983; Donahue, 1983; Friedrich,

1980; Fyfe, 1978).1 Citizen violence toward the police

1 Deadly force is "the use of firearms or any other
object likely to cause death or great bodily harm.



also received some notice by researchers during the 1980s

(Chamlin, 1989; Choi, 1989; Geller & Karales, 1982;

Margarita, 1980; Uchida & Brooks, 1988). Unfortunately,

resisting arrest and other minor forms of conflict (e.g.,

refusing to obey, interfering with the police) have rarely

been researched by criminologists.2 Academic literature on

the police response to domestic disputes is also addressed

in the literature review. The emphasis is on the dangers of

policing domestics (Choi, 1989; Garner & Clemmer, 1987;

Uchida & Brooks, 1988) and preferred arrest policies (Friday

et al., 1991; Hirschel et al., 1992; Sherman & Berk, 1984).

Police Use of Nonlethal Force

No problem we R hungry we got a little
physical w/ a [name omitted] on Columbus it
was fun we had to teach him a little respect
for the police hahahahaha we had
fun no stick time though. (Los Angeles
Police Department Mobile Digital Terminal

Nondeadly force is the use of police batons, (e.g., PR-24s)
or other impact weapons, chemical agents (e.g., mace),
canine bites, or other objects used as weapons. In
addition, striking blows by officers with their hands or
feet is also an example of nondeadly force" (Lakeland, FL
Police Department, 1991, p. 2).

2 The differences among charges are illustrated in the
police policy for one city. Any subject who refuses to
comply with an officer's orders may be charged with
"refusing to obey." If the subject makes physically evasive
movements to defeat an officer's attempt at arrest (e.g.,
pushing or pulling away or not allowing the officer to apply
handcuffs), the charge may be "resisting arrest." Attempting
to or inflicting bodily harm upon the officer will usually
result in a misdemeanor or felony charge of assault or
battery (section adapted from Procedural Orders, Lakeland,
FL Police Department, February 3, 1991).


transmission, cited in Report of the Independent
Commission on the Los Angeles Police Department,
1991, p. 49)

In a very real sense, the patrolman-to-citizen
exchanges are moral contests in which the
authority of the state is either confirmed, denied
or left in doubt. (Van Maanen, 1980, p. 298)

Many of the first important studies of police-citizen

interactions focused on subject demeanor and arrest (Black,

1980; Piliavin & Briar, 1964; Sykes & Clark, 1980). Black

(1980) analyzed officer-complainant interactions and found

disrespectful complainants were not taken seriously.

However, officers did not always react to disrespectful

citizens with force. Unless there was a doubt as to who was

in charge, in the majority of encounters officers did not

perceive citizens as a threat to their authority (Piliavin &

Briar, 1964; Reiss, 1980). Black compares the differences

in how magistrates and police officers handle disrespectful


Unlike the judge, the policeman has no
special legal weapons in his arsenal for dealing
with citizens who refuse to defer to his authority
at a verbal or otherwise symbolic level. (1980, p.

Unofficial means may be used to manage defiant

subjects. Officers may use "street justice" (as Los Angeles

Police officers appear to have done to Rodney King in 1991).

Some officers perceive the government's authority as their

own personal power (Van Maanen, 1980). The officers' need

for status and respect has been associated with a greater

likelihood of violence (Sykes & Clark, 1980; Westley, 1970).


However, Westley (1970) believes that officers generally

know the consequences of abusing their powers and are

cautious about when they do it.

Many of the initial studies of police-citizen

encounters were conducted through direct observation (Black,

1980; Lundman, 1974; Piliavin & Briar, 1964; Reiss, 1974;

Skolnick, 1966; Van Maanen, 1980). Some of the researchers

also attempted to quantify their observations (Lundman et

al., 1980; Sykes & Clark, 1980). A seminal study conducted

by Albert Reiss (1980) in 1966 in Boston, Washington, DC,

and Chicago found the presence of additional officers

actually encouraged police violence and their use of force.

According to Westley (1970), officers rarely apply any

sanctions against a colleague who uses violence against a

citizen. The same phenomenon was also discussed by Toch,

who said, "Few officers may be violent, but these are backed

by peers who see police bonds as links to survival" (1977,

p. 31). The relationship between the number of officers

present and violence is complex and difficult to measure

since we do not know if the conflict erupted before or after

additional officers arrived at the call.

Other researchers have suggested that the number of

officers and witnesses present was correlated with the

police use of force (Brooks, 1989; Croft, 1985; Reiss, 1980;

Smith & Visher, 1981); however, the order of this

relationship may present methodological problems. Friedrich


(1980, p. 91) asks, "does the audience promote the use of

force or does the use of force attract [the] audience?"

Black (1980) could not find any evidence officers

discriminated on the basis of race when demeanor was

controlled. However, Skolnick (1966) found that racial bias

by the police was routine in an early observational study of

two police departments. He stated, "The average policeman

does not like Negroes; and since most policemen are

straightforward and outspoken, few would deny such a

generalization, at least among themselves" (Skolnick, 1966,

p. 81). Officers may be more hostile toward young black

males than black females. Skolnick (1966, p. 86) stated,

He may sometimes sympathize with the Negro
mother's lack of funds, especially if she is on
welfare, but he maintains a fundamental hostility
toward the young black male.

In a multivariate analysis of the observational data of

Reiss, Friedrich (1980) found that demeanor (e.g.,

intoxication) and the number of other police and citizens

present had a significant impact on the use of force by

police. Looking at 18 variables, Friedrich found that

individual traits of officers and suspects (e.g., sex, age,

race, experience and class) had almost no explanatory value.

However, Friedrich was pessimistic about these findings

because the full regression model only explained 12% of the

variance of the use of force.

The major conclusion which must be drawn from a
study that finds 18 factors thought to explain the
use of force accounting for slightly more than 10


percent of the variation in the use of force is
that either our theoretical understanding of the
phenomenon is very weak or our methodology for
studying that phenomenon is very weak, or both.
(Friedrich, 1980, p. 96)

Police use of force frequently involves minor

misdemeanors, such as intoxicated and disorderly

individuals, rather than serious felonies. Subjects are

often distraught or deranged at the time of the encounter

with officers (Croft, 1985). Subjects' use of alcohol was

associated with police use of force (Croft, 1985; Friedrich,

1980). A problem discovered by Croft (1985), however, was

that police reports of the Rochester, NY Police Department

did not always mention alcohol use by subjects, although

Croft believed that the irrational behavior by subjects

and/or the location of the call, i.e., outside a bar or at a

party, were sufficient indicators that the subjects were


Research conducted in New York City, Denver and Miami

all found that police use of force was infrequent (Bayley,

1986; Bayley & Garofalo, 1989; Fyfe, 1989a). Bayley and

Garofalo (1989) observed 350 tours of duty in three police

precincts in New York City and concluded that citizen-to-

citizen violence was usually over by the time officers

arrived. When violence was in progress, officers used

physical force 8% of the time; however, if the dispute was

over, force was used only 6% of the time (p < .01).


The use of force by citizens toward officers was also

uncommon. When violence was still in progress, citizens

assaulted officers 9% of the time; however, when violence

was not in progress, this decreased to only 1%; X2 = 17.84,

p < .01 (Bayley & Garofalo, 1989). Control variables,

(e.g., area of the city, suspect intoxication, suspect race,

or officer experience) were not used by Bayley and Garofalo.

Denver police used force just 5.4% of the time (Bayley,

1986), and Miami officers used force in 12% of their calls

(Fyfe, 1989a). These three studies were observational,

i.e., the researchers rode along with the patrol officers on

their shifts, and it is unknown if this affected the

officers' or the citizens' behavior.

Most individual characteristics of officers (e.g.,

race, sex and education) have been inadequate predictors of

the use of force (Croft, 1985; Grennan, 1987; Sherman,

1980). Officer age and amount of experience may be more

promising. However, caution should be used when analyzing

these two indicators, since they may be highly correlated

with each other, according to Sherman (1980).

Croft (1985) found that younger, less experienced

officers and those appointed at an earlier age were more

likely to use force in Rochester, NY. Multivariate analysis

was not used in Croft's study, so these findings may be

spurious. For example, younger officers are usually

assigned to the graveyard shift and are working in more


violent areas of the city, which may increase their use of

force (Fyfe, 1978).

In sum, studies indicate that the police use of force

is uncommon; however, when it does occur, subjects' demeanor

appears to be one of the better predictor variables.

Individual characteristics of officers (e.g., sex, race and

education) were not related to the use of force, with the

exception of officer age and experience. Control variables

(e.g., assignment, area and shift) should be used, since

younger officers with less experience are often assigned to

the most violent neighborhoods on the most active shifts.

Police Use of Lethal Force

The youngster who flees because police spells
danger may be gunned down by an officer who feels
threatened by the youngster's running. Such
occasions are doubly tragic because the two
parties are victims of reciprocal error. (Toch,
1977, p. 32)

Police use of lethal force has resulted in a great deal

of research during the 1980s. At least five doctoral

dissertations (Blumberg, 1983; Donahue, 1983; Rodez, 1980;

Taylor-Greene, 1988; Uchida, 1982), along with numerous

smaller studies have focused on this phenomenon.

Researchers have looked at individual, organizational,

and structural variables and their relationship to the use

of deadly force. Several studies have examined the

relationship between race and deadly force (Blumberg, 1981,

1983; Fyfe, 1981; Geller & Karales, 1981; Meyer, 1980;


Robin, 1963; Taylor-Greene, 1988); administrative policies

and deadly force (Fyfe, 1978, 1979, 1981, 1982, 1989a;

Matulia, 1982; Rodez, 1980; Scharf & Binder, 1983; Taylor-

Greene, 1988; Uchida, 1982); and economic inequality and

deadly force (Jacobs & Britt, 1978; Kania & Mackey, 1977).

James Fyfe (1989b), a prodigious writer on the police

use of deadly force, has distinguished between "extralegal"

and "unnecessary" violence. While extralegal violence

consists of abusive and unethical actions by officers,

unnecessary violence involves basic officer incompetence.

Fyfe explained that the motivation is different in these two


Extralegal violence involves the willful and
wrongful use of force by officers who knowingly
exceed the bounds of their office. Unnecessary
violence occurs when well-meaning officers prove
incapable of dealing with the situations they
encounter without needless or too hasty resort to
force. (Fyfe, 1989b, p. 465)

Frequently, officers fail to properly diagnose

encounters until it is too late. Fyfe believes that police

officers must take more time and preparation in order to

prevent violence. For example, officers who rush into

crowded supermarkets in response to robbery calls may be

provoking a violent confrontation. Officers should expand

their knowledge of their patrol areas and try to avoid

making "split second decisions" (Fyfe, 1989b).3

3 Fyfe (1989b) discusses the use of concealment and
knowledge. If officers know their beat, they can find places
to conceal themselves and not force a gun battle.


Many researchers have examined the relationship between

race and deadly force (Blumberg, 1981; Fyfe, 1981, 1982;

Geller & Karales, 1981). In Chicago and New York City,

black officers were more likely to use lethal force than

white or Hispanic officers (Fyfe, 1981; Geller & Karales,

1981). Like officer age and experience, the

disproportionate assignment of minority officers to higher

crime areas may explain this phenomenon. Many police

shootings occur when minority officers are off-duty within

their own neighborhoods (Fyfe, 1981).

There is evidence that black subjects are shot and

killed by police at a significantly higher rate than other

races.4 This may be associated with higher rates of

arrest, especially for crimes of violence (Geller & Karales,

1981; Matulia, 1982). Meyer (1980) found Los Angeles police

officers were more likely to shoot unarmed blacks than

unarmed whites or Hispanics. A decade ago, Fyfe (1982)

discovered blacks in Memphis were at a greater risk of being

shot by police, especially when they were fleeing from

property crimes. Takagi (1974) charged the American police

with having a separate trigger finger for blacks and whites.

However, proving police racism in shooting situations is

highly complex, especially given that baseline data is

4 For example, Alpert (1989) found 59% of the police
shooting victims were black in Miami, FL, and the population
of blacks at the time was just 25%. Kobler (1975) and
Geller and Karales (1981) found similar patterns in Seattle
and Chicago.


unavailable on the frequency of encounters or contacts

between police and citizens of all races (Blumberg, 1981).

All of the encounters where police elected not to use their

firearms but could have are unknown. Geller (1985) refers

to this as the "next frontier" of research on the use of

deadly force.

Officer gender, education and socioeconomic status

(SES) have all been found to be unrelated to the use of

deadly force (Blumberg, 1983; Sherman & Blumberg, 1981).

However, Blumberg (1983) found officer age and police

experience were significant, even after controlling for risk

exposures in separate studies he conducted in Atlanta and in

Kansas City, MO.5 Younger, less experienced officers were

more likely to be involved in police shootings. This

relationship was also discovered by Croft (1985) in

Rochester, NY. Perhaps younger officers are more aggressive

and lack the emotional maturity necessary for policing

(Blumberg, 1983). However, the layoff of 3,000 "junior"

officers in New York City did not reduce shootings. Rather

it increased the age and the amount of experience of the

officers who were involved, as older officers were forced to

replace younger ones on the streets (Fyfe, 1978).

Sherman and Langworthy (1979) found greater support for

structural variables than organizational variables. They

5 High risk or low risk was established by controlling
for arrest exposures and beat assignment (Blumberg, 1983, p.


concluded that deadly force was correlated with: (1) general

homicide rate, (2) violent index crime, (3) number of

officers per 1,000 population, (4) gun density, (5) arrest

rate, (6) violent arrest rates, and (7) the percent of

personnel assigned to Internal Affairs. The strongest

relationships were with the general homicide rate (r = .55),

officers per 1,000 population (r = .45) and the violent

crime index (r = .37). The number of supervisors per line

officer, percentage of complaints of police misconduct

investigated, percentage of complaints substantiated,

unemployment rate, suicide rate and population density were

all statistically insignificant. Since multivariate

analyses were not used, some of the findings could prove

spurious. However, these results are consistent with other

studies of police-caused homicides that have used structural

level variables (see Jacobs & Britt, 1978; Kania & Mackey,

1977, 1992).

Kania and Mackey (1977, 1992) and Jacobs and Britt

(1978) examined the police use of deadly force across the

United States. Using vital statistics from every state,

Kania and Mackey (1977) found that structural and community

level variables (e.g., welfare rate, police exposure to

violent crime, homicide rate, food stamp rate and percentage

of homes without televisions) were correlated with a higher

rate of police-caused homicides. However, their study had a

number of problems. "Race" was not examined and the


variable "welfare" was used as a proxy for social

stratification. Some highly stratified states may actually

have low welfare rates though. At the time, Kania and

Mackey's research was referred to as "the only quantitative

effort to examine police violence" (Jacobs & Britt, 1978, p.


Kania and Mackey (1992) recently examined the impact of

the Tennessee v. Garner (1985) decision, which found the

Tennessee statute unconstitutional that permitted police

officers to shoot a nonviolent fleeing felon. They

discovered this case had the greatest impact on the states

that have already been operating with more restrictive

policies prior to the ruling than on those states that were

less restrictive (i.e., applying the common law fleeing

felon rule). Kania and Mackey concluded that the law and

administrative policies have less effect on police-caused

homicides than do community and structural variables.6

These data suggest that Fyfe's 1988 assumptions
are in error and that the background levels of
public violence are stronger predictors of state
practices involving the police use of deadly force
than state law or policies redrawn to conform to
such laws. (Kania & Mackey, 1992, p. 12)

6 But for contrary findings, see Sparger and Giacopassi
(1992) who recently examined the effect of Tennessee v.
Garner (1985). They concluded that it was responsible for a
reduction in police shootings and less discriminatory use of
deadly force in Memphis. "The data suggest that between
five and 15 lives were spared by adoption of a restrictive
shooting policy" (Sparger & Giacopassi, 1992, p. 224).


Jacobs and Britt (1978) used multivariate analyses and

determined that states with higher economic inequality, a

higher violence index and a significant change in the

percentage of population had the most police-caused

homicides. They concluded that economic conflict theory

could explain deadly force in the United States. However,

Sherman (1980) cautioned that research using economic level

variables can be difficult to assess, as the indicators are

not selected in a systematic way. Additionally, economic

conflict theory was never developed for the specific problem

of police violence (Sherman, 1980).

Taylor-Greene (1988) found that officer shootings were

best predicted by the civilian homicide rate. Administrative

controls or civilian review were insignificant. Suspect race

was correlated with lethal force only at the zero order but

was insignificant in a multivariate model. These findings

are consistent with the earlier study by Sherman and

Langworthy (1979).

In summary, there seems to be a relationship between

officer and subject race and deadly force (Fyfe, 1981;

Geller & Karales, 1981; Matulia, 1982); however, it may be a

spurious one. Black subjects are more likely to be arrested

than whites and thus have higher exposure to potentially

lethal situations (Geller & Karales, 1981; Matulia, 1982).

Black officers are also more likely to work in violent

neighborhoods on more dangerous shifts. There is some


evidence that younger and less experienced officers are more

likely to be involved in the use of deadly force (Blumberg,

1983). However, this could again be related to assignment

and/or shift. Structural variables (e.g., homicide rate,

poverty rate) seem to provide more promise than either

individual or administrative variables (Jacobs & Britt,

1978; Kania & Mackey, 1992, 1977; Sherman & Langworthy,

1979; Taylor-Greene, 1988).

Assaults on Police

No administrative policy, no strategy, no training
program, no police procedure--not even bulletproof
vests--has ever been shown to reduce violence
against police. (Sherman, 1980, p. 9)

There has been less academic interest in violence

toward the police than in the police use of deadly force.

Uchida and Brooks (1988, p. 1) stated, "Unlike police use of

deadly force, assaults and homicides of police have not

received the attention by researchers that it merits." In

1988, almost 59,000 police officers were assaulted in the

United States; approximately 83% of those assaults were with

hands, fists or feet (Statistical Abstract of the U.S.,

1992). From 1980 to 1989, 801 officers were killed in the

United States.7 Seventy percent of the assailants of

police officers had a criminal record and 30% had a prior

7 This does not include the 713 officers who died
accidently in the performance of their duties (mostly by
auto accidents).


arrest for a violent crime. Officers were more likely to be

killed in the evening (between 6:00 p.m. and 6:00 a.m.) and

January was the most dangerous month of the year, while

August was the safest month (Major, 1991).

Most of the research on violence toward officers has

been descriptive and non-theoretical (Bristow, 1963; Fyfe,

1989b; Geller & Karales, 1982; Henley, 1987; Konstantin

1984; Major, 1991; McMurray, 1990). However, a few studies

have looked at the phenomenon from a symbolic interactionist

perspective (Bannon, 1976; Choi, 1990; Margarita, 1980) and

economic conflict theory (Chamlin, 1989; Peterson & Bailey,


Margarita (1980) examined a century of police homicides

(n = 245) of New York City police officers and found that in

the majority of cases, suspects who murdered officers were

"goal directed" and were trying to protect themselves from

arrest. She discussed a five category typology: (1)

perseverance, (2) autonomy, (3) protection, (4) defense of

others, and (5) contempt. These categories were formerly

used by Toch (1969) in his study of male violence.

Margarita's findings may have important implications for

mandatory arrest policies at domestics, since officers could

be at a greater risk when they are compelled to arrest.8

8 The Supreme Court of Oregon found that police
officers face civil liability if they fail to arrest at
domestics when probable cause exists that an assault has
been committed. In the case of Nearing v. Weaver, 1986, the


Austin, TX, police officers were also more likely to be

assaulted when making an arrest, with 65% of the assaults on

officers occurring in retaliation for applying handcuffs

(Henley, 1987).

Peterson and Bailey (1988) used a multivariate model to

examine the homicides of officers in all 50 states from 1977

to 1984. Police killings were regressed against several

structural variables (e.g., percent black, percent family

poverty, income inequality). They found police homicides

were associated with societal factors (e.g., the divorce

rate, the poverty rate); however, they cautioned additional

explanatory variables (e.g. officer competence, errors in

judgment) also may be necessary to understand the phenomenon

(also see Chamlin, 1989). A serious problem with many

cross-jurisdictional studies is the amount of variation that

occurs within each state or unit of analysis.

Several studies have found that younger police

officers, those with less than five years of experience,

were more likely to be the victims of assaults and injuries

(Bannon, 1976; Little, 1984; Uchida & Brooks, 1988). As

with studies of police force, this may be explained by

factors other than officer age or experience. Swanton

(1985) discovered that older officers usually left the more

"active" precincts and were working as police supervisors,

complainant alleged that officers knowingly failed to comply
with Oregon's mandatory arrest policy (Gundle, 1986).


which would reduce their street exposure and their risk of

being assaulted.

Forty percent of the suspects who assaulted officers in

Baltimore County were males between 19 and 25 years old.

When there were two subjects present, they tended to both be

males from different age groups and the same race (Uchida &

Brooks, 1988). Nationally, 70% of the suspects who killed

officers had a previous arrest of some kind, and 21% were on

parole or probation at the time of the event (Boylen &

Little, 1990).

Alcohol was a significant factor in 47.5% of the

assaults on officers in Austin, TX (Henley, 1987). Using

FBI statistics, Konstantin (1984) found that officer

initiated events or "on view" calls were more dangerous than

calls received through radio dispatch (Bayley & Garofalo,

1989; Fyfe, 1989a). However, Konstantin included a myriad

of police calls in the officer-initiated category (e.g.,

warrants, off-duty shootings and back-up calls not

dispatched), which probably inflated the results.

In sum, the research generally reports that structural

(e.g., poverty rate) and situational variables (suspect's

alcohol use, self-protection) relate to assaults on police.

Age or officer experience were the variables at an

individual level of analysis that seem to relate to assaults

on officers. Theoretical research on police assaults has

been meager. Those efforts that have been informed by


theory have usually applied social psychological

explanations (Choi, 1990; Margarita, 1980) or an economic

conflict theory (Chamlin, 1989). Sherman (1980, p. 94)

stated, "the problem is to develop a theory that not only

explains variance, but also offers the symmetry, coherence,

and elegance of formal theories. Whether such a theory is

possible remains to be seen."

Resisting, Refusing and Interfering

What is crisis for the citizen and diversion for
the outsider becomes routine for the patrolman.
(Reiss, 1974, p. 7)

The research on resisting arrest, refusing to obey and

interfering with police has been very limited. FBI

statistics (gathered from state and local police agencies)

traditionally have focused on more serious forms of police-

citizen violence (e.g., homicides, shootings). Since many

police agencies do not maintain specific files on resisting

arrests (as they do in the cases involving police use of

force or citizen complaints), obtaining this information can

be difficult. One method is by direct observation of

officers, as Reiss (1974) did in his classic study. Another

way is through the departmental incident reports themselves.

According to Reiss (1974), citizens behaved in ways

during or following an arrest that made them open to the


charge of resisting arrest 12% of the time.9 Reiss also

found that the majority of injuries to officers occurred

with interfering charges rather than resisting. Interfering

usually involves those who are not primary suspects, such as

witnesses or even victims. The individual may attempt to

stop the officer from making an arrest. Reiss found that

both resisting arrest and interfering were far more common

at misdemeanor calls than felonies (also see Croft, 1985;

Henley, 1987). Minor infractions, such as fights or

disorderly conduct, more commonly result in norm conflict

between police and subjects and provoke subjects into

believing that authority is being used unjustly (Reiss,

1974) .0

While Westley (1970) has claimed that some officers use

resisting or interfering charges to conceal brutality, Reiss

(1974) disagreed. Reiss (1974, p. 55) said, "An officer

generally does not cover himself for resorting to physical

coercion by charging an offender with resisting arrest."

He also found that resisting arrest was not even charged

when circumstances had warranted it and he attributed this

to the perception by many officers that resistance was only

9 It was unknown if the suspects were ever formally
charged with refusing, resisting or interfering, since Reiss
admitted that he and his associates did not follow the
arresting officers) to the jail to find out what the exact
charges were on the booking sheet.

10 Croft (1985) found that in 1,012 arrests involving
police use of force, 42.2% involved resisting arrest.


a routine part of their job. The frequency of resisting

arrest may vary with assignment or the crime rate of the

officers' beats. Officers working in high crime areas may

be accustomed to minor forms of resistance.

Sykes and Clark's (1980) study in 1970 involved 9,000

citizens and 1,466 encounters with police. They discovered

an asymmetrical relationship between officers and citizens.

First, police expected lower-class members to defer to them,

however, they were not likely to defer to these citizens in

return. Second, lower-class whites were more deferent

toward police than were lower-class nonwhites. Third, the

potential for conflict was lowest when officers confronted

upper-class members, as the wealthy and the police deferred

to each other. Finally, conflict was highest when officers

confronted lower-class nonwhites who had broken the law

(Sykes & Clark, 1980).

Domestic Disputes

Most of the first studies of domestic disturbances were

conducted through direct observation (Bard & Zacker, 1974;

Lundman, 1974). Bard and Zacker (1974) and Lundman (1974)

tended to dispel the "myths" surrounding domestic disputes.

According to Lundman (1974), domestics were not particularly

hazardous; he even described the interactions between police

and citizens as "polite."


One of the "myths" may concern alcohol. Bard and

Zacker (1974, p. 291) said that "police have tended to

emphasize the role of alcohol as a critical factor in family

conflicts." However, they found that just one-third to one-

half of all domestics involved alcohol use.

Station-house anecdotes may have contributed to some

myths surrounding domestic disputes (Buzawa & Buzawa, 1990;

Davis, 1981; Dolon et al., 1986; Ellis, 1987; Garner &

Clemmer, 1989). Ellis (1987) refers to it as the

"domestics-are-most-dangerous-to-police" myth. He states

that domestics may account for many assaults and injuries in

absolute numbers, however, we do not know how often officers

attend these calls. Therefore, we cannot assess their

dangers. Ellis also believes that the myth of domestics as

dangerous was a product of the government's attempt to win

the loyalty of the police during the 1960s and 1970s by

showing a disingenuous concern for their safety.

Attending domestic disturbances has been greatly
exaggerated. Attending domestic disturbances is
not associated with a disproportionate number of
deaths and injuries to police officers. (Ellis,
1987, p. 331)

Of 1,388 domestic violence calls observed in New York

City, officers (working in a special crisis unit) used

physical force to restrain or separate subjects only 1% of

the time (Bard & Zacker, 1974). A study conducted in the

Midwest found that the potential or actual act of physical

violence between police and citizens occurred in 12% of the


calls (Lundman, 1974).11 A recent survey found that 78.4%

of the officers said they had been physically assaulted at a

domestic dispute at least once during their careers (Friday

et al., 1991).

According to Garner and Clemmer (1989), the likelihood

of an officer being killed at a domestic dispute is

negligible. However, this research has been criticized for

how they calculated their danger rates (Choi, 1989; Elliott,

1989; Stanford & Mowry, 1990). Garner and Clemmer analyzed

several former studies of violence toward the police and

then "fit" these findings with two other studies of police

activity rates, i.e., the amount of time officers spent on

specific calls.12 Choi (1989, p. 61) says, "[B]y

arbitrarily imposing one set measure derived from one study

on another, the precision in estimating the danger rates is

in question." Elliott (1989, p. 431) adds, "The procedure

used for obtaining harm-activity ratios for the latter

estimates is questionable (the estimates of harm and

activity used to establish a given ratio are derived from

different jurisdictions)." When the studies by Bannon

(1976) and Chapman et al. (1974) were reanalyzed by Garner

and Clemmer (1989), domestic disputes actually ranked first

11 Lundman (1974, p. 26) defines potential or actual
violence as verbal threats, efforts at physical restraint,
fighting, making weapons ready for use and/or the use of
12 Garner and Clemmer's (1989) estimate of police
activity rates were based on studies conducted in Kansas
City, MO in 1975 and in three other cities in 1977.


in assaults on officers. However, the authors did not

elaborate on this finding except to strike a cautionary


Some caution is necessary in rejecting the earlier
claim that domestic disturbance calls are among
the most dangerous calls to which police respond,
if dangerousness is defined broadly to include
injury or assault. (Garner & Clemmer, 1989, p.

Robberies appear to be the most hazardous call in terms

of police deaths. For every officer who dies responding to

a domestic, three die at armed robberies (Ellis, 1987;

Margarita, 1980). While robberies are more lethal, assaults

and injuries are far more common at domestics (Choi, 1989;

Elliot, 1989; Friday et al., 1991; Uchida & Brooks, 1988).

According to Buzawa and Buzawa (1990, p. 30):

Officer deaths may be viewed as being a limited
measurement of a call's danger, because the
occurrence is relatively infrequent. Assaults
upon officers are far more common [at domestics].

Recent studies have found that domestics are risky in

terms of assaults and injuries (Choi, 1989; Uchida & Brooks,

1988). According to Choi (1989), research which has

suggested that the dangers of domestics are exaggerated

suffers from both internal and external validity problems.

The challenge lies in the fact that this area has
hardly been investigated. However, the price for
such a challenge is that there are very few
studies to rely on. Of the handful of studies
available, virtually all of them are restricted to
general assaults and injuries. (Choi, 1989, p. 65)

Choi identified the conditions which were likely to

result in police injuries at domestics in Ontario, Canada.


Using cross-tabular analyses of several situational

variables, Choi (1989) found that arrest, type of dwelling,

demeanor of principals, number of officers, and officer

prepared were all related to officer injuries (see Table 1).

Individual or personal attributes of officers (e.g., rank,

years of service, sex and age) were insignificant.13 Choi

did not examine race or socioeconomic status (SES) of the

officers or the violators.

Using multiple regression, Choi (1989) found that

police arrest, victim injury, detached house, alleged

assault, and hostile suspect were all statistically

significant (see Table 2). However, victim injuries were

inversely related to officer injuries. Choi calls this

finding "puzzling and totally unexpected," and he believes

it may be the result of the abusers having fled from the

scene before officers arrived when victims were injured.

Lone officer (officer responds to the domestic call alone)

and officer preparation were both insignificant. However,

because they came close to significance, Choi cautions that

they should not be rejected, especially given the sample


13 This table is not shown here. However, Choi (1989)
found that the differences between officers who were injured
and those who were not injured at domestics was not more
than 5% on each personal attribute.

14 P values for these two variables were not provided.
The explained variance (R Squared) was unreported by Choi,
although the overall regression model was significant (p =


Individual characteristics of officers (e.g., rank,

years of service, sex, height, weight and age) were not

significant in either crosstabular or regression analyses

(Choi, 1989). These variables were also insignificant in

previous studies of the police use of lethal and non-lethal

force (Friedrich, 1980; Sherman & Langworthy, 1979; Taylor-

Greene, 1988; but see Blumberg, 1985, or Croft, 1985 for

some contrary findings).

According to Choi (1989), "who called the police?"

(neighbor or victim) did not influence the decision to

arrest. However, this variable was not examined for a

relationship with police injuries. Since Choi focused on

injuries and not on assaults, his sample was not as large as

it might have been (n = 324). For example, only one in five

assaults on officers in a Baltimore County study resulted in

any injuries (Uchida & Brooks, 1988).

Choi's discovery that alcohol was not related to

police-citizen violence differs from most other research

(Croft, 1985; Friedrich, 1980; Henley, 1987; Reiss, 1974).

However, he did find that suspect drinking was related to

the officers' decision to arrest. Police officers who

arrested at domestic disputes were more likely to be injured

(Choi, 1989). This may have significant training and policy

implications, given the current trend toward proarrest

(Dunford, 1992; Hirschel & Hutchison, 1992). Choi stated:

Table 1: Situational Variables and
(Crosstabular Analyses)

Officer Injuries

Situational Injured Non-Injured Total
Variables Officers Officers Cases
# % # %

Type ***


Injury to



Hostile ***

Number of *

Officer **

Arrest **


30 14
7 9






10 18
27 12

30 13
5 11

24 19
13 8




135 92
99 81

182 86
70 91

150 89
100 85

46 82
206 88

210 87
40 89

103 81
150 92

25 78
236 89

187 90
63 80

104 82
149 92

*** Significant at .01 level
** Significant at .05 level
* Chi-square = 2.1, p = .15

Source: Alfred Choi, Policing Domestic Disturbances:
Police Responses and Risks, Ph.D. dissertation,
York University, 1989.










Table 2:

Situational Variables
(Regression Model)

and Officer Injuries

Situational Regression t-Value
Variable Coefficient

* Detached house -.113 -2.82
* Alleged assault .106 2.05
** Injury to victim -.128 -2.76
Weapon involved .004 .08
Alcohol involved .029 .52
* Hostile principal .100 2.45
Lone officer .128 1.93
Officer prepared -.088 -1.94
** Arrest .123 2.85

* P < .05
** P < .01

F = 4.52

P = .001

Source: Alfred Choi, Policing Domestic Disturbances:
Police Responses and Risks, Ph.D. dissertation,
York University, 1989.

Arrest is clearly shown to have significant
effects on police injury and is the best predictor
of all situational variables. The probability of
the officer being injured increases by 12.3% if
the officer decides to arrest the suspect. (1989,
p. 115)

Choi (1989) used two social-psychological theories, the

practicality thesis and the patriarchy thesis, to try to

explain police-citizen encounters. The practicality thesis

examines the police reluctance to arrest, given both

organizational and criminal justice demands (e.g., the need

to be accessible for more "important" calls, seriousness of


the offense, lack of victim's assistance).15 The

patriarchy thesis focuses on male stereotypes about the

victims' role in domestic violence and the male social order

(Gelles & Cornell, 1990). Choi found greater support for a

practicality thesis. Both of these theories focused on

arrests and not officer assaults or injuries.

Uchida and Brooks (1988) examined 1,550 assaults of

police officers in Baltimore County, MD from 1984 to 1986.

They discovered that domestic disputes accounted for only

5.3% of police activity, but 25% of all the assaults of

officers. More than 21% of these assaults resulted in

injuries to an officer. According to Uchida and Brooks:

While recent research studies indicate that
domestic disturbances are not as dangerous as
originally believed, our data show that domestics
present a high risk of danger to police, at least
in terms of assaults. (1988. p. 101)

Choi (1989), Elliott (1989) and Uchida and Brooks (1988) all

believe Garner and Clemmer (1989) were premature in de-

emphasizing the dangers of domestic disputes.

Uchida and Brooks (1988) found that handcuffing and

struggling with suspects were less likely to precede

assaults than were investigating and resolving disputes.

These categories were rather vague and were not discussed in

great detail by the authors. Injured officers may have been

15 Buzawa and Buzawa (1990) advise that an arrest is
less likely if the domestic occurs at the end of the
officer's shift because this means the officer would have to
stay late.


caught off guard or may not have been prepared. According

to Grennan:

Officers responding to a less overtly threatening
situation (family dispute, disorderly man, and so
on) are not always prepared to face a violent
conflict in which they may be assaulted and
injured. An officer, due to laxity or the
unanticipated confrontation, can become an easy
target of an assailant. (1987, p. 81)

Other studies, however, have concluded that attempting to

arrest and handcuffing suspects were more dangerous

activities for officers (Choi, 1989; Henley, 1987;

Margarita, 1980).

Uchida and Brooks (1988) compared domestics with non-

domestics. For example, they found that assaults on

officers at domestics were more likely to occur indoors than

was the case for non-domestic assaults. However, they did

not examine why some domestics resulted in violence between

officers and citizens while others were resolved without

overt conflict.

There are very few studies to rely on when examining

police assaults and injuries at domestic disputes (Choi,

1989). This appears to be even more obvious if only

theoretical research is considered. The use of Turk's

(1966, 1969) theory in this dissertation is an attempt to

fill that void. It not only attempts to predict the kinds

of conditions which increase the potential for overt

conflict but offers an explanatory framework which can

integrate various research findings.


In sum, the evidence provided by previous studies on

the dangers of policing domestics is varied. While several

researchers have maintained that domestics are very

hazardous (Choi, 1989; Elliott, 1989; Friday et al., 1991;

Uchida & Brooks, 1988), others seem to believe that these

risks are overstated (Bard & Zacker, 1974; Ellis, 1987;

Garner & Clemmer, 1989). How "danger" is operationalized

may be one of the obstacles. The studies which have focused

exclusively on the deaths of officers have concluded that

domestics were not dangerous; however, those that have also

looked at minor assaults and/or injuries have found that

domestics were indeed hazardous.16 Structural and

situational variables (e.g., type of residence, arrest)

rather than individual officer or suspect variables (e.g.,

rank, sex, age) are generally more predictive of injury at



During the 1980s, researchers claimed that proarrest

policies at domestic disputes significantly reduced the

recidivism rate of batterers (Berk & Newton, 1985; Sherman &

Berk, 1984; Sherman & Cohn, 1989). Recently, however,

studies have questioned this relationship. According to

16 In a study of violence toward police officers in
Baltimore County, MD, Uchida and Brooks (1988) found that
domestic disputes were less likely to involve firearms than
were non-domestic calls; however, they were more likely to
involve sharp, blunt, or thrown objects.


Dunford (1992, p. 122), "Arrest in Omaha, by itself, did not

appear to deter subsequent domestic conflict. ." Hirschel

and Hutchison (1992, p. 118) found, "arrest is not a

significant deterrent for misdemeanor spouse assault" in

Charlotte, NC. Other studies in Dade County, FL and in

Milwaukee found that arrest actually increased the

recidivism rate of batterers when they were unmarried and

unemployed (Pate & Hamilton 1992; Sherman & Smith, 1992)

indicating the importance of social bonding. Sherman (1992,

p. 45) acknowledged, "Undoing the effects of initial results

may be much harder than some criminologists imagined."

In a study of Phoenix police officers, Ferraro (1989)

found that the department's presumptive arrest policy was

being habitually evaded, as police arrested only 18% of the

time. "Mandatory and presumptive arrest policies

implemented in the absence of change in other parts of the

legal system will probably have little lasting impact on how

police respond to domestic violence" (Ferraro, 1989, p.

72) .17

While most of the research has focused on recidivism,

the unintended consequences of proarrest policies have been

overlooked. For example, has proarrest had any effect on

police-citizen conflict? Buzawa and Buzawa (1992, p. 214)

argue "there is a heightened probability of conflict between

17 Similar findings were found in Minneapolis, where it
was discovered that officers arrested just 20% of the time
(Buzawa and Buzawa, 1990).


the officer and the offender and even the victim when the

victim does not want arrest." While stating that domestics

were rarely fatal for officers in Tampa, FL, Stanford and

Mowry (1990, p. 248) admit, "departments with pro-arrest

policies in domestic disturbance situations may demonstrate

a different potential for assault." Several researchers

have found a relationship between officer assaults and

arrest (Choi, 1989; Henley, 1987; Margarita, 1980). In

Duluth, MN, however, police injuries reportedly went from 15

to 0 after a mandatory arrest policy was implemented (Buel,

1988).18 According to Buel, abusers are less inclined to

commit assaults on police officers if they know that the

officers are following a "strict legal duty imposed by law"

(1988, p. 221).

The Limitations of Past Research

Past studies of police-citizen conflict have had

problems related to (1) external validity, (2) data sources,

(3) failure to use control variables or multivariate models,

and (4) the absence of theory to identify variables and

explain results.

Most of the research has traditionally been conducted

at a single police department (Croft, 1985; Fyfe, 1989;

18 The Duluth mandatory arrest policy began in 1982.
After several telephone conversations in the Autumn of 1991
with a Duluth Police Department Crime Prevention Sergeant
(who advised he was instrumental in the study mentioned),
these figures could not be verified.


Geller & Karales, 1981, 1982; Margarita, 1980; Stanford &

Mowry, 1989; Uchida & Brooks, 1988). Given the sometimes

potent influence of local cultures on criminal justice

phenomena, caution must be used in interpreting these

findings and offering generalizations.

Uniform Crime Reports (UCR) have been the sole data

source of many police researchers (Kania & Mackey, 1977;

Konstantin, 1984; Matulia, 1978; Sherman & Langworthy, 1979;

Taylor-Greene, 1988). However, they generally provide

little detail and are not immune from politics and

embellishment. According to Hagan (1982, p. 52), "Crime

rates have a mysterious way of dropping if required for

political purposes" (see also Sherman & Langworthy, 1979).

Although susceptible to reporter bias (Donahue, 1983),

departmental offense/incident reports provide a superior

source of information for research.19

Many of the studies of police-citizen conflict have

been descriptive or have examined bivariate relationships

(Bayley & Garofalo, 1989; Croft, 1985; Donahue, 1983; Fyfe,

1981; Geller & Karales, 1982; Uchida & Brooks, 1988). Since

19 In addition, national level data often use the state
or city as the unit of analysis. Differences within cities
cannot be examined this way. Peterson and Bailey (1988, p.
216) admit, "Although states are heterogenous in comparison
to counties, SMSAs, and cities, it is not possible to
conduct the analysis at a lower level of aggregation because
police homicide figures provided by the FBI cannot be
disaggregated below state level for all the jurisdictions
and years under consideration."


multivariate analyses or control variables were not used,

these findings may prove to be spurious.

When researchers have provided theories of police-

citizen violence, they are often left undeveloped or are too

difficult to assess (Sherman, 1980). However, some studies

have applied interactionist theories (Bannon, 1976; Choi,

1989; Margarita, 1980; Toch, 1977) and economic conflict

theory to this phenomenon (Chamlin, 1989; Jacobs, 1979;

Jacobs & Britt, 1978; Kania & Mackey, 1977). Economic

conflict theory, however, has been criticized for being

unclear and difficult to evaluate for this problem (Orcutt,

1983; Sherman, 1980). Orcutt (1983, p. 334) stated, "some

of these theories are so vague and so deficient in specific

empirical implications that it is unclear what kinds of

evidence would be required for their assessment."


Theoretical Framework

Austin Turk has been cited as, "the deviance theorist

who has persisted longest in an effort to develop a non-

Marxist framework for the analysis of conflict processes"

(Orcutt, 1983, p. 321). Turk (1966, 1969) has focused on

the process of criminalization and the relationship between

authorities and subjects. Authorities are the decision-

makers and subjects are those affected by the decisions.

Subjects are empirically distinguished from
authorities by their inability to influence the
legal processes of norm creation or enforcement.
(Turk, 1969, p. 33)

As such, authority relations are inevitable according

to Turk and so the potential for conflict is inherent. Not

all authorities have an equal opportunity to act or

influence the law. Turk (1969) views the "firstline

enforcers," as those who have the greatest impact on

criminality. Although the norms of deference and submission

maintain the balance of the authority-subject relationship,

in specific situations, authorities must rely on coercive

means to gain compliance; Turk refers to this as "nightstick



Sellin (1938) has argued that criminality is the result

of cultural distinctions. Specific offenders may not know

or may not accept specific legal norms. Turk also discusses

the variations in the cultural norms of authorities and


The greater the cultural difference between the
evaluator and the violator, the less likely are
psychological sanctions which assume a capacity
and readiness to subtle cues to get through to the
violator, and therefore sanctioning will have to
be more physically coercive in order to enforce
the norm. (Turk, 1966, p. 345)

Non-violent forms of persuasion (e.g., body language,

verbal statements) by officers may be insufficient to gain

the compliance of those who adhere to unlike norms. For

example, when police encounter subjects of a different

culture, age, sex, or race, the probability of overt

conflict should increase (Turk, 1966; 1969).1 Turk states

that conflict is more likely when there are "interactions

and relations between people and others not of their own

kind" (1969, p. 285).

Turk's conflict theory has gone through several

revisions (Orcutt, 1983). Norm resistance may be applied to

situations "within" the authority structure itself. For

example, Turk (1966, p. 343) states, "specific persons in

official law enforcement agencies, as well as persons in

purportedly criminal populations and organizations, may be

Norm resistance and overt conflict will be used
interchangeably throughout the discussion of police-citizen


in either the law-enforcing or the law-resistant party in

empirical instances." It follows that officers who fail to

comply with the policies of their departments are acting as

resisters (Turk, 1966). Some attributes may be less

distasteful to lower-level enforcers (i.e., patrol

officers), while they may be highly offensive to management

(i.e., chiefs and captains). Ferraro's (1989) study of

Phoenix police officers illustrates an example of this

phenomenon as most patrol officers failed to comply with the

department's arrest policy.2 While Turk's theory permits

the analyses of all types of authority-subject relations,

this dissertation focuses primarily on the encounter between

authorities (i.e., police) and subjects (i.e., batterers).

In certain structured situations, conflict is more or

less likely to occur and it hinges on cultural norms, i.e.,

statutes, written law, or policies, and social norms, i.e.,

actual behavior or how a policy is being enforced. The

probability of overt conflict is greatest when there is

congruence between cultural and social norms, since conflict

is inherent in the structure of the authority-subject

relationship. When officers comply with a proarrest policy

at domestics, the likelihood of conflict with subjects

2 Turk (1969, p. 39) said, "no norm enforcer (e.g.,
policeman, judge, priest, elder) can be fully aware of and
able to control all that conditions his own behavior,
implying that his behavior will always differ somewhat at
least from what he and others expect and believe it to be."
In an earlier essay, Turk (1966, p. 345) stated, "only
rarely will announcers and enforcers be the same persons."


should increase. If authorities and subjects both hold to

disimilar beliefs and act in accord with these beliefs,

there is no room left for compromise. However, conflict is

less likely when cultural and social norms are incongruent.

Here, although cultural beliefs clash, authorities and

subjects do not practice what they preach. Turk (1969, p.

56) states, "neither party is very prone to fight over an

essentially meaningless set of symbols."

Turk (1969) describes four basic structural situations

affecting the probability of authority-subject conflict and

each situation has a different potential for conflict.

These situations are: (1) the congruence of norms for both

authorities and subjects (high-high), (2) the congruence of

norms for subjects but not authorities (high-low), (3) the

congruence of norms for authorities but not subjects (high-

low), and (4) the congruence of norms for neither

authorities or subjects (low-low). In 2 and 3, the

probability of conflict falls into the middle ranges;

however, conflict is more likely if it is the authorities'

cultural and social norms that are congruent because the

authorities are more intolerant than subjects in accepting


Flagrant, persistent disregard for the law will,
nevertheless, force authorities to act to
demonstrate that they are still in charge, that
they are still able to assert their will against
resistance in a showdown. (Turk, 1969, p. 64)


In the event that cultural norms conflict, both

authorities and subjects appeal to their own separate

standards. While authorities are likely to appeal to legal

norms (e.g., the law or policies), subjects are likely to

appeal to non-legal norms (e.g., natural law, the right to


Conflict between authorities and subjects is also

related to organization (Turk, 1969). Organization is

determined by both the "complexity of the relationship" and

by "social support."

When an attribute or act has been integrated into
a system of relationships, implying that it is a
part of some role which the individual performs,
then we can expect that some kind and degree of
coercion will be required to break the behavior
pattern. (Turk, 1969, p. 58)

Authorities are organized "by definition," since the

authority structure implies a significant degree of

organization by the more powerful members of society (Turk,

1969).3 However, subjects may or may not be highly

organized. The greater group or social support subjects

have for their actions, the less likely they will be to

compromise with the authorities. Turk (1969) also mentions

"complexity" in his discussion of organization. Activities

or behaviors that are more complex (rather than simple) are

more organized and harder to break by the authorities.

3 Turk (1969) admits to variation in the organization
of authorities, although this does not necessarily undermine
its organized character.


An individual who has group support for his
behavior is going to be more stubborn in the face
of efforts to make him change than is someone who
has only himself as an ally. (Turk, 1969, p. 58)

Turk (1969) defines sophistication as the "knowledge

used in order to manipulate others." The likelihood of

overt conflict is lessened when subjects or authorities are


Sophisticated norm resisters are more accurate in
assessing the strengths and weaknesses of their
position relative to authorities, and consequently
better able to avoid open warfare with the
superior enemy. (Turk, 1969, p. 59)

Sophisticated subjects are less likely to engage the

powerful authorities in battle. Authorities who are

sophisticated are also able to achieve their goals without

physical violence.4 Turk sees greater variation in the

levels of sophistication than organization [for


Clearly, sophistication is a prerequisite for
enduring authority; so it could be argued that
just as authority implies organization, it also
implies sophistication. However, control agencies
demonstrably vary a great deal in the extent to
which their policies and practices are affected by
appreciation of the need for knowledge. (1969, p.

The various combinations of organization and

sophistication lead to different probabilities of conflict

within the respective congruence and incongruence conditions

4 Muir (1980, p. 52) states, "It is important to
appreciate the unusual skills of eloquence which policemen
in authority need if they are going to escape the violence
implied by the neighborhood beef." He emphasizes the need
for language skills and police training.


discussed above. This results in a matrix (see Table 3).

Note that the probability of conflict is highest whenever

subjects are organized and both authorities and subjects are


Turk cautions that the cells in his matrix (see Table 3

below) do not represent empirical observations but only

relative probabilities of conflict:

[W]e do not know just how unlikely is conflict in
this minimum chance situation. All that can be
postulated is that this will be the lowest value
of the 32 possibilities and that the value will be
somewhere above .00, because the authorities are,
after all, on public record against an attribute
actually found in some part of the population.
(Turk, 1969, p. 63).

Critiques of Turk's Theory

Turk's (1966, 1969) theory has been criticized by both

radical conflict theorists and consensus criminologists.

His effort to create a value neutral conflict theory has

been considered overly submissive, illusory, deterministic,

abstract and tautological (Gibbons, 1979; Orcutt, 1979;

Taylor et al., 1973).

In a discussion of Turk's theory, Taylor et al. call

it, "one of permanent adjustment of the subordinate to the

powerful under present social arrangements" (1973, p.

251).5 Since radical theorists profess to affiliate with

5 Gibbons (1979, p. 170) argues that Taylor, Walton and
Young set up "straw men" who "were laboriously constructed
and then demolished. and lacked fidelity to the
theorists' actual arguments."

0 0

z 0

0 H


.c o

* En*
430 04

u a t
c o
a) 0

(0 '0 t(

C CP C t
0) C)O
0 d 0 N
0).'4 ) '-
-4 C (

O ko co



30 30
c 0

1* ( *oe

0 t) 0)
*<1 0 o- (








-4- >4
-H .-
*-4 i-


4 4-
a0 (

O4 0


N 40



the working classes in resisting the ruling elite, Turk is

seen as an ally of the government (Bernard, 1981; Taylor et

al., 1973). Gibbons states:

Radical criminologists claim that effective power
in American society is wielded by members of a
corporate ruling class, which is a very different
view from that espoused by mainstream
criminologists, including conflict theorists such
as Turk, who opt for a pluralistic conception of
interest groups and power relationships in modern
society. (1979, p. 159)

Taylor et al. (1973) suggest that criminologists pay

attention to the origins of conflict, which they see as

inequality and social stratification.6 However, Turk is

less concerned with the origin of conflict than with the

conditions in which it becomes overt. According to Turk

(1969, p. 41), "people assent to the order because it is

built into them through socialization." His theory attempts

to explain differential rates of overt conflict between

authorities and subjects and the likelihood of

criminalization. In the process, Turk does not sound like

an apologist for the elite.7

"Nightstick law" and "assembly line justice" are
commonly experienced by the very weak, and
official records are not kept with the same care
as in conflicts where the resisters have enough

6 Richard Quinney (1980) views crime and criminal
behavior as the products of capitalism and social
inequality. He sees the role of criminology as rejecting
the capitalist order.

7 Paul Walton claims that "when you unpack the
ambiguous language of Quinney, Turk and Matza, you find that
they are by no means as radical as their language would have
them appear" (Mintz, 1974, p. 44).


power to demand and get some degree of adherence
to procedural norms. With the helpless, enforcers
can also be careless because mistakes do not cost
much and indeed may cost less than would even-
handed justice. (1969, p. 70)

Consensus theorists have criticized Turk for neglecting

a significant body of knowledge which has shown that

consensus and not conflict characterizes society (Akers,

1979). However, Turk views consensus as only illusory, and

more recently stated, "whether they realize it or not,

people are inevitably involved in intergroup struggles over

who shall have what resources in a finite world" (1982, p.


Radical theorists, like Quinney (1980) and Taylor et

al. (1973) reject Turk's position on the inevitability of

the authority-subject relationship; and are critical of

Turk's approach for being deterministic (Bernard, 1981).

[T]he acts of men who have not been "conditioned
to accept the authorities as fact of life" require
something more in the way of a description of
human consciousness than is allowed by the
adoption of the terminology of behaviorist
psychology. (Taylor et al., 1973, p. 249)

The "condition" relevant to Turk's theory lies in his

discussion of norms of deference--something that was not

specifically developed. The criticism is less germane to

Turk's discussion of the structured circumstances in which

these norms of deference will not check the inherent

conflict. Indeed, the probabilities for conflict presented

in Table 3 above are largely independent of whether subjects

are "conditioned to accept authorities."


The role of determinism (as one of the hallmarks of

positivistic science) is disputed even as it applied to

radical conflict theories. While Groves (1985) states that

radicals are hostile toward positivism, Bernard (1981)

believes that Marx himself was a positivist describing the

objective processes of history. Like Dahrendorf (1959),

Turk views inequality as a fact and its elimination as

idealistic. Turk would probably not be too concerned if

others see his efforts to explicate conflict in authority

relations as being deterministic.

Turk's level of abstraction has raised concerns about

tautology in his basic conflict formulation. For example,

the assertion that systems of control serve the needs of

authorities seems true by definition (Orcutt, 1983, p. 334).

So does the assertion that "lawbreaking is taken to be an

indicator of the failure or lack of authority; it is a

measure of the extent to which rulers and ruled, decision-

makers and decision-acceptors, are not bound together in a

stable authority relationship" (Turk, 1969, p. 48). Taylor

et al. have argued that Turk's overall abstraction

exaggerates what are really modest theoretical


[T]he fact that the young, the masculine, and the
members of minority racial groups are more likely
than most to be criminalized [makes Turk's
theory] superficially relevant. (1973, p. 246)

One of the purposes of this study is to demonstrate how

Turk's theory may be applied in ways that hold policy and


theoretical significance. The challenge must be to convert

abstract concepts into concrete operationalizations, so they

may be tested and subjected to falsification.8 Akers

(1977) maintains that even conceptual tautologies can

provide the foundation for testable hypotheses. Police-

citizen interactions at domestic disputes provides a

situation where specific hypotheses may be advanced.

Turk's Conflict Theory at Domestics

Recall that Turk (1969) focused on "specific structured

situations" and not on social psychological meanings. He

stated that in order to understand conflict, "A sociological

theory of interaction is required. [and] it must be

[one] of interaction among groups. rather than a social

psychological explanation of patterns in interactions. .

even though psychological assumptions [are] implicit in any

such theory" (Turk, 1969, p. 53).

Turk hypothesizes that under conditions of dissensus,

conflict is most likely to become overt when both

authorities and the subjects behave consistently with their

cultural norms. For example, males who in thought and deed

consider themselves to be the "masters of their houses"

cannot be expected to defer to police who view spouse abuse

8 In an early discussion of norm resistance, Turk
explains that his propositions are only a "rough first
approximation of a theory that will hopefully offer not only
plausible but precise, systematic, and therefore testable
explanations of criminality and punishment" (1966, p. 347).


as plainly criminal. The domestic dispute call places

authorities in positions where they must confront general

American privacy norms about the family and the home that

are combined with norms of male dominance. According to

Bard and Zacker (1974, p. 290):

Intervening officers are in an extremely sensitive
situation, one which lends them to being seen as
outside "intruders" (and, particularly, with power
and authority) into the sacrosanct domain of the
home. The dilemma for the police officer is that,
although he has been summoned, he moves into
highly volatile emotional "territory." His
presence in the home can be regarded as either
helpful or threatening. Clearly, under such
circumstances, there are enormous potentials for
either insight or incite.

Turk assumes there is a potential for conflict between

authorities and subjects. Bard and Zacker graphically

detail how this potential for conflict applies to the

domestic call. However, Turk saw some circumstances where

this basic conflict is more pronounced and where conflict

was more likely to become overt. When cultural differences

are greater, the potential for conflict increases (culture

conflict) and it will be most likely to become overt when

authorities and subjects act congruently with their own

respective norms (congruence of norms and behavior).

Prior criminal behavior and arrests demonstrate the

willingness for subjects to oppose the normative order.

Although we usually cannot measure the norms they hold

directly, we might expect these subjects to have developed

an oppositional philosophy and to be more likely to offer


resistance when confronted by control agents. In a

discussion of Turk's theory, Vold and Bernard (1986, p. 281)

stated, "The probability of conflict will be greatest when

subjects have a full blown language and philosophy with

which to defend their behavior." As indicated in FBI data,

70% of the subjects who killed police officers between 1980

and 1989 had a previous arrest and 20% were on probation or

parole at the time of the homicide (Boylen & Little, 1990;

Major, 1991). Donahue (1983) found suspects who were shot

by police averaged 2.63 felony arrests prior to the shooting

event.9 The initial hypothesis of this dissertation

relates to circumstances where there may be greater conflict

between subjects and authorities inherent in the situation

so that there is a greater likelihood for overt resistance.

H1 Overt conflict between authorities and subjects will be
more likely when subjects have a criminal arrest

If a prior arrest was for a crime of violence or

disorder (e.g., domestic violence, assault, battery,

interfering with police, disorderly conduct, refusing to

obey, resisting arrest, robbery, homicide), the probability

of authority-subject conflict should be greater."1 The

9 The majority of the subjects shot by police officers
in Kansas City, MO and Atlanta, GA had been previously
arrested (Blumberg, 1983).

10 Disorderly conduct was included under crimes of
violence or disorder since the police often place this
charge on subjects who were verbally abusive and aggressive
toward officers and others present. The following was taken
from a domestic dispute call which occurred in the City of


next hypothesis concerns the potential for conflict and the

congruence of the subjects' cultural and social norms.

H2 Overt conflict between authorities and subjects will be
more likely when subjects have a criminal record
involving an arrest for a crime of violence or

Recall that Turk also argued that coercion and control

would vary by the extent of cultural differences.

Indicators of cultural differences are numerous and may

include differences in the age, sex, and race of subjects

(suspects) and of authorities (police). Turk states that

authorities and subjects must "learn and continually re-

learn to interact with one another as, respectively,

occupants of superior and inferior statuses and performers

of dominating and submitting roles" (1969, pp. 41-42).

According to Martin (1993), officers sometimes face

deference reversals, where citizens refuse to acknowledge

the officers' official status but instead focus on age, sex

or race. Since the norms of deference are crucial in

maintaining the balance of the authority-subject

relationship, it is less likely that older subjects who

assault their family members will acquiesce to younger

officers. Therefore, the following hypothesis relates

Charleston. "[Officer] attempted to speak with the
suspect when the complainant became even more abusive and
loud. She started yelling profanities and a crowd started
gathering. She was advised by [officer] that if she did not
control herself that she would be placed under arrest for
being disorderly." This narrative conforms to the author's
personal experiences as a patrol officer and sergeant with
the Albuquerque, NM Police Department.


conflict with differences in the age of the officers and the


H3 Overt conflict between authorities and subjects will be
more likely when subjects are over 30 years of age and
officers are under 30 years of age.

The potential for overt conflict should be higher when

nonwhite officers encounter white subjects who are law

violators. In this situation, not only are there cultural

differences, but deference norms which maintain the

authority-subject balance may be in jeopardy." The

following hypothesis relates differences in the race of

authorities and subjects.

H4 Overt conflict will be more likely when officers are
nonwhite and subjects are white.

According to Turk (1969), it is the learning of

superior and inferior stations and of dominating and

submitting roles which maintains the authority-subject

balance. Male subjects who view themselves as the "rulers

of their castle" may not readily defer to female officers

who are mandated to arrest at domestics. Martin (1993, p.

338) said, "By virtue of their office, female officers

expect deference to their position from citizens; by virtue

of their female status they are subordinates of men."

Therefore, the next hypothesis relates conflict with gender

differences of the officer and the subject.

11 Turk (1969, p. 61) explains, "[Overt conflict is
greater [when] the basis of legitimacy is the social norm of
deference rather than 'norm internalization,' or consensus."


H5 Overt conflict will be more likely when officers are
female and subjects are male.

For Turk (1969) sophistication referred to "Knowledge

used to manipulate others." Several researchers have

found that less experienced officers are more likely to be

assaulted and to use physical force (Bannon, 1976; Bard &

Zacker, 1974; Blumberg, 1983; Croft, 1985; Muir, 1980;

Uchida & Brooks, 1988). According to Muir (1980, p. 52),

"It is important to appreciate the unusual skills of

eloquence which policemen in authority need if they are

going to escape violence implied by the neighborhood beef."

More experienced officers should possess greater

sophistication in handling sensitive calls, such as

domestics. However, officers who are "not trained in

intervention subtleties may provoke aggressive, territorial

responses in citizens" (Bard & Zacker, 1974, p. 290).12

The following hypothesis relates norm resistance with the

officers' professional experience--an indicator of their


H6 Overt conflict will be more likely when officers have
48 months or less of police experience.

Croft (1985) found that younger police officers were

more likely to be involved in use of force calls. Blumberg

(1983) also discovered younger officers were involved in

more deadly force incidents than older officers. Uchida and

12 Turk (1969, p. 60) cites Trebach (1964) who found
that the FBI is less prone to violence than state and local
officers because of their superior training.


Brooks (1988) determined that younger officers were at a

greater risk of being assaulted as well. While these

findings may reflect differential assignments or "bad"

shifts of younger officers, it may also reflect their

abilities, skills or sophistication (Fyfe, 1978; Swanton,

1985). It is less likely that younger officers have the

sophistication of older officers in defusing potentially

explosive encounters. The following hypothesis relates

officer age (as a indicator of sophistication) with norm


H7 Overt conflict will be more likely when officers are
under the age of 30.

Another indicator of subject sophistication is the use

of alcohol or drugs. Intoxicated subjects who have

assaulted their spouses should be less capable of

manipulating officers when confronted with their behavior.

From Turk's theory, suspect intoxication is a predictor of

sophistication; much of the research is consistent with this

hypothesis (Croft, 1985; Friedrich, 1980; Henley, 1987;

Reiss, 1974; Uchida & Brooks, 1988; but see Choi, 1989 for

contrary findings). The following hypothesis relates to

subject sophistication and intoxication.

H8 Overt conflict will be more likely when subjects are
intoxicated or under the influence of alcohol or drugs.

Although citizen-to-citizen violence has usually

terminated by the time officers arrive, police-citizen

conflict is more probable if the dispute is still in-


progress (Bayley & Garofalo, 1989). In this situation, both

authorities and subjects have less opportunity to develop

plans or strategies (Fyfe, 1989a); their efforts will be

less sophisticated. Therefore, this hypothesis results:

H9 Overt conflict will be more likely when the domestic
dispute is in-progress when police arrive at the call.

When officers are sent to a domestic dispute call, the

police radio dispatcher often provides valuable information

(e.g., weapons present, suspect intoxication, prior calls to

that location, victim injuries).13 However, there are

situations where officers observe disputes while they are

patrolling (on sight) or simply drive up on them.14 In

these instances, officers often have less information and

less time to prepare themselves (Scharf & Binder, 1983) and

so will have lower sophistication. The next hypothesis

relates to authority sophistication and how the domestic

dispute came to the attention of the police.

H10 Overt conflict will be more likely when the police
receive the domestic as an on-view rather than a
dispatch call.

Recall that for Turk the dimensions of organization are

complexity and social support. Turk (1969) argues that the

more an attribute or an act has been integrated into a

13 Proactive interventions are likely to be in
progress and outside.
14 In a discussion on police-citizen violence, Scharf
and Binder (1983, p. 83) stated, "[An] advantage of
dispatched encounters is the time available to the officer
immediately prior to entry on the scene."


system of relationships, there is greater organization.

There is a greater level of social complexity, therefore,

when officers challenge the authority of a "head-of-

household" who has used violence against family members than

when they intervene in a dispute between friends or

acquaintances. From Turk's analysis, it is predicted:

H11 Overt conflict will be more likely when the disputants
are married or related to each other rather than

Victims of assault who are unemployed may grant more

support to the batterer due to financial or emotional

dependence. Victims have been known to intervene on the

side of the batterer or refuse to prosecute after police

arrive. In San Diego, in spite of a special police domestic

violence unit (made up of 21 officers), 60% of the victims

refused to participate in prosecuting their batterers

(Benson, 1993). Thus, the following hypothesis relates to

the organization of the subject.

H12 Overt conflict will be more likely when the victims are

Another consideration is the visibility of injuries.

Whatever makes a dispute more visible alters how it is

interpreted and the potential for support. The organization

of the subjects should be reduced when their victims are

visibly injured, since the subjects are likely to have less

social support for their behavior and resistance is less

realistic. The result is likely to be a decrease in overt


conflict with police. Therefore, the following hypothesis

refers to subject organization in terms of victim injuries.

H13 Overt conflict will be more likely when the victims do
not have visible injuries.

Turk (1969, p. 58) explained that, "An individual who

has group support for his behavior is going to be more

stubborn in the face of efforts to make him change than is

someone who has only himself as an ally." When police

arrest more than one individual, the potential for conflict

should be greater, since more potential opposition now

exists. According to Buzawa and Buzawa (1992), police-

citizen conflict may be related to dual arrests, especially

if there is a mandatory arrest policy. Thus, the following

hypothesis is related to the degree of organization in terms

of the number of persons arrested.

H14 Overt conflict will be more likely when more than one
person is arrested at the dispute.

Domestic disputes which are visible to a wider public

are inherently more complex. According to Muir (1980),

police-citizen encounters in front of crowds present

difficulties for managing such situations. "A policeman who

attempts to restore order at the expense of a citizen's

honor leaves the citizen no recourse but to attack" (Muir,

1980, p. 48). The same may be true for authorities, as

Friedrich (1980) found visibility had a substantial impact

on police-citizen violence. "Police sometimes rely on force

to demonstrate their authority and protect their image, and


these things are in greater jeopardy when more people are

around" (Friedrich, 1980, p. 91). In Rochester, NY,

witnesses failed to support police in 18.9% of the calls and

in 17% of the calls, they actually attempted to help the

suspect by stopping police from arresting (Croft, 1985).

The following two hypotheses relate to organization in terms

of witnesses and the location of the dispute.

H15 Overt conflict will be more likely when witnesses are

H16 Overt conflict will be more likely when the dispute
occurs in public.

The probability of overt conflict should be greater if

the domestic dispute was a misdemeanor rather than a felony.

Croft (1985, p. 69) found, "both numerically and

proportionately use of physical force incidents involve

misdemeanor and violation offenses." Reiss (1974) found

that subjects were more likely to offer resistance at

misdemeanor arrests because of their perception that the law

was being unjustly applied to them, as others avoided arrest

for similar acts. Therefore, when the domestic assault is a

misdemeanor, subjects are likely to believe they have more

social support (organization). The following hypothesis

relates to organization in terms of the level of criminal


H17 Overt conflict will be likely when the domestic assault
is a misdemeanor rather than a felony.

The real world is quite different from the scientific

laboratory, and it is unlikely that human relations can be


understood by the examination of a single variable. Several

indicators of Turk's culture conflict, sophistication and

organization themes are likely to be involved in each

domestic encounter between officers and citizens. In

general, the greater the number of indicators of Turks' norm

resistance theory, the greater the potential for police-

citizen conflict. In multivariate analysis, indicators of

sophistication should have more impact on norm resistance

than the organization indicators according to Turk. "The

available evidence suggests that skill in avoiding head-on

collisions in social interaction is more crucial than the

degree of organization in affecting the chance of

authorities-subjects conflict" (Turk, 1969, p. 59).

Accordingly, variables related to sophistication (e.g.,

intoxication, officer age, officer experience, on view

encounters and how the call came to the attention of

officers) should have more explanatory value than the

organization variables (e.g., victim injuries, location,

victim employment, witnesses present, number arrested,

seriousness of charge). Thus the following hypotheses are


H18 Overt conflict will be more likely when there are a
greater number of indicators of culture conflict,
organization and a lack of sophistication at the

H19 Overt conflict will be more likely when there is a lack
of sophistication rather than when there is greater


Turk (1969) hypothesized that under conditions of

normative dissensus, conflict was more likely to become

overt when both authorities and the subjects acted

consistently with their cultural norms.15 Congruence on

the part of authorities is somewhat more predictive than

congruence of the subject (see Table 3 above). To the

extent that presumptive or mandatory arrest policies are

achieved, authority's cultural and social norms become more

congruent. Turk would predict greater overt conflict

between subjects and police in such a circumstance. Thus,

the last hypothesis concerns the cultural and social norms

of the authorities.

H20 Overt conflict will be more likely when the police
department has a proarrest policy in place.

Turk's theory has given rise to 20 hypotheses--

something which by itself counters his detractors, who

complained about theoretical abstraction and tautology.

Those hypotheses are recapitulated below and the Turk theme

which each addresses is included in parentheses. In sum,

authority-subject conflict is most likely to become overt


15 When the police department has a proarrest policy,
cultural and social norms should become congruent [if there
is compliance with the policy]. However, subjects may not
act consistently with their own cultural norms. In this
situation, the probability of norm resistance would be
somewhere in the highest 50% of Turk's matrix (from 1 to
16), and would depend on the sophistication and organization
indicators (see Table 3).


(1) Subjects have an arrest history for any type of crime
(culture conflict and the congruence of norms and

(2) Subjects have an arrest history for a crime of violence
or disorder (culture conflict and the congruence of
norms and behaviors);

(3) Officers are less than 30 years old and subjects are
over 30 years old (culture conflict);

(4) Officers race is nonwhite and subjects race is white
(culture conflict);

(5) Officers gender is female and subjects gender is male
(culture conflict);

(6) Officers have no more than 48 months of police
experience (sophistication of authorities);

(7) Officers are under 30 years old (sophistication of

(8) Subjects are intoxicated (sophistication of subjects);

(9) The domestic dispute is still in progress when police
arrive (sophistication of authorities and subjects);

(10) Officers receive the domestic dispute on view rather
than through the police dispatcher (sophistication of

(11) Disputants are married or relatives (organization of

(12) Victims are unemployed at the time of the call
(organization of subjects);

(13) Victims do not have visible injuries (organization
of subjects);

(14) Multiple arrests are made at the dispute (organization
of subjects);

(15) Witnesses are present at the dispute (organization of

(16) The dispute occurs in a public area rather than within
a dwelling (organization of subjects);

(17) The citizen-to-citizen assault is a misdemeanor rather
than a felony (organization of subjects);


(18) There are a greater number of indicators of culture
conflict, organization and a lack of sophistication
present (culture conflict, sophistication and

(19) Authorities and subjects are less sophisticated rather
than more organized (sophistication of authorities and
of subjects and the organization of subjects);

(20) The police department has a presumptive arrest policy
in place (congruence of norms).


The unit of analysis for this research is police-

citizen encounters at domestic disturbances. Two research

sites were chosen: The Charleston, SC Police Department and

Savannah, GA Police Department. The sampling frame consists

of official police reports from both the Charleston Police

(January 1, 1988, through Dec. 31, 1991) and Savannah Police

(January 1, 1987, through February 28, 1988). Calls for

service where officers were dispatched to a domestic dispute

but did not write a report were not included in the study.

Many of these calls were considered "unfounded" by the

officers or else the complainant could not be located.

There are several limitations in using police incident

reports (Croft, 1985; Donahue, 1983; Fyfe, 1981). Since

official reports detailing police-citizen encounters may be

used for supervisor monitoring, disciplining, or civil

lawsuits, officers no doubt cast themselves in the most

favorable light. In addition, Croft (1985, p. 53) stated,

"[I]t can be expected that some dynamics of the actual

situation may be forgotten or overlooked." However, police

reports are likely to be as informative as personal


interviews with the participants and far superior to many

newspaper accounts.'

The data collected from the Charleston Police

Department were used to address all of the hypotheses with

the exception of hypothesis 20, which involves the

presumptive arrest policy. Data collected from the Savannah

Police Department relate to hypothesis 20 (the congruence of

social and cultural norms for authorities). The dependent

variable in both of the research locations is norm

resistance. This is operationalized as verbal attacks on

officers, physical assaults on officers, citizens' refusal

to obey a lawful order, resisting arrest, interfering with

officers, and the use of force by police (both lethal and

non-lethal). The information was obtained from the

officers' incident reports themselves and coded as either

norm resistance present or not present during the domestic

dispute. Incident reports where subjects were described by

officers as "belligerent," "uncooperative," "hostile," or

"abusive" were considered to be acts of norm resistance,

even when formal charges were not generated. As pointed out

by Reiss (1974, p. 54), "[B]ooking on a charge of resisting

1 For example, the news media usually obtains
information from the Police Department Public Information
Officer. However, it is not uncommon for a story to be
incorrect by the time it appears in print or on the air.
While the author was a member of the Albuquerque, NM Police
Department from 1979 through 1986, he observed this
phenomenon regularly.


an arrest or interfering with an arrest, as for all arrests,

is a matter of officer discretion."

Charleston, SC

The 1991 population of the city of Charleston, SC was

83,167 residents in 100.08 square miles. Approximately

29.78 square miles are water areas. The racial breakdown of

the city was 57.2% white (47,572); 41.6% black (34,597) and

1.2% other (998).2 In 1988, there were approximately 250

sworn officers and 100 civilian employees on the

department.3 The study period started on January 1, 1988

and concluded on December 31, 1991. The Charleston Police

Department had a mandatory arrest policy for domestic

disturbances in effect throughout the four years examined.

The data from the Charleston Police Department was used

to compare the universe of calls where there was evidence of

norm resistance (N = 89) with a sample of calls where no

norm resistance was exhibited (n = 81). The Charleston

Police Department Planning and Research Unit supplied the

author with a computerized listing of all domestic disputes

from 1988 through 1991; the bulk of these reports were found

on microfilm. All incident reports were read by the author

or several research assistants, who worked for the

2 This information was acquired from the Charleston
Police Department Monthly Report (June, 1992).

3 This information was obtained from a report to
Savannah Police Chief David Gellatly (Knoche, 1990).


Charleston Police. Each of these assistants had several

years of experience in the department's Records Unit.

Periodic checks were made by the author to ensure that the

assistants were using standardized procedures.

There were 749 domestic violence reports during the

four year study period, the large majority of which involved

spouses or cohabitating partners. The population of reports

showing norm resistance was selected (N = 89). Every sixth

case of the remaining reports was collected, the selection

being made from a random start. However, many of the

"nonresistance" calls had to be discarded for a variety of

reasons. Often the batterer had already left by the time

officers arrived, so there was not an officer-subject

encounter. In other cases, the officer determined that the

call was unfounded, so there were no grounds for authority-

subject conflict. Only those calls in which potential

conflict between officers and subjects was inherent in the

encounter were then retained (N = 81).

An ex post facto design was used to look at the

relationship of several independent variables and norm

resistance. According to McMillan and Schumacher (1989, p.


Ex post facto designs are used to compare two or
more samples that are comparable except for a
specified factor. The possible causes are studied
after they have occurred. Rather than
manipulating what will happen to subjects, as in
experimental designs, the research focuses on what
has happened differently for comparable groups of


subjects, then explores if the subjects in each
group are different in some way.

It was impossible to determine the frequency of

domestic disputes which the Charleston Police Department

responded to during the study period. This is because

domestics are initially coded by the police radio dispatcher

under "general disturbances" (along with parties, neighbor

disputes or fight calls) until officers write an incident

report. Therefore, the 749 police reports are not

indicative of the actual number of dispatches to domestic

disturbances. Officers may arrive at a location and decide

that there was insufficient evidence to make an arrest or

even to write a report. The complainant or victim may have

left, or the location may have been wrong. The domestic

dispute policy in Charleston states:

The officers will take the assailant into custody
for a misdemeanor assault with probable cause, a
misdemeanor assault committed in their presence, a
felony assault, and an arrest with a warrant
(Charleston Police Department, Procedural Order
Number 89 Handling Domestic Dispute Calls).

In addition, according to a Charleston Police

Department training memorandum (undated) on domestic

violence law:

a. Where an assault is evident or committed in the
officer's presence: an arrest will be made in all
cases regardless of the desire of the victim.

b. Where an assault is alleged by the victim but no
physical evidence is present: an arrest will be made if
requested by the victim.


Charleston officers are required to write incident

reports on all domestic disturbances; however, this does not

always occur. For example, the author spent one evening

riding with an officer in a downtown area. The officer

responded to a domestic dispute call on the street. Both

disputants (an older black man and woman) were told to

depart in opposite directions. The officer did not write a

report; however, there were no visible injuries to either


Police reports list the type of offense and the

individual characteristics of the suspect, victim/

complainant (e.g., age, race, sex, height, weight), and also

provide a narrative of the event. In January 1991, the

Charleston Police Department revised their report forms to

conform with the Uniform Crime Reporting system (UCR).

Although the newer reports are similar to the older ones,

there are some obvious differences. For example, there is

now a space allotted for the officers to check a box if the

subject was under the influence of alcohol or drugs

(Appendix A). On the earlier report forms, the sobriety of

the subjects could only be established if the officer

4 Buzawa and Buzawa (1990) advise that even in the
Minneapolis Police Department in 1986, there were only 3,645
arrests out of 29,948 domestic assault calls (i.e., arrests
were made 20% of the time). Officers also commonly failed
to write reports in spite of the mandatory arrest policy.


decided to mention it in his or her narrative (Appendix


Criminal histories of subjects arrested for domestic

assault were collected for the population of norm resisters

and for the sample of nonresisters. This information was

acquired from the I.D. Section of the department. Individual

characteristics of the officers in the sampled encounters

were obtained from Police Personnel and provided the

reporting officer's name, rank, race, sex, date of birth and

date of hire.6

Independent Variables

The independent variables in the hypotheses were

analyzed to determine if they relate to norm resistance.

The variables examined were batterers' prior arrests for any

crime (PREARRST), arrests for only crimes of violence or

disorder (PREVIOL), the intoxication of subjects (SUSINTOX),

subjects' age (SUSAGE), subjects' race (SUSRACE), subjects'

5 The narrative allows reporting officers to describe
the event or offense in their own words. The length of
these narratives varies from a couple of sentences to
several pages. Police supervisors, prosecutors and defense
attorneys may read the officer's narrative. In a study on
the use of force in Rochester, NY, Croft (1985, p. 53)
explained, "The narrative is based on the officer's
perception of the situation and, therefore, is subject to
the several inaccuracies common to a participant's
perception and to recording past recollection."

6 The list was dated September, 1991. In a few cases,
the reporting officer's name and characteristics were not
found on the list. This meant that they were no longer
working for the department. This was coded as missing data.


gender (SUSGEN), officers' age (OFFAGE), officers' race

(OFFRACE), officers' gender (OFFGEN), officers' police

experience (EXP), if the domestic was still in progress

(INPROG), the kinship relationship of the disputants (REL),

victims' employment status (EMP), victims' injuries

(VVISINJ), witnesses present (WITNESS), number of disputants

arrested (NUMARR), seriousness of the assault (FELONY),

location of the dispute (LOC) and how the call was received


The operationalization of these variables requires

further attention. PREARRST refers to whether the subject

does or does not have an arrest history prior to the

incident report. Any prior misdemeanor or felony (both

property and violent crimes) available to the Identification

Section of the Charleston Police Department is included in

this dichotomous variable.

PREVIOL refers to whether or not the subject has been

arrested for any public order or personal crimes. This

includes homicide, robbery, assault, domestic violence,

resisting arrest, interfering with police, refusing to obey

and disorderly conduct. In the City of Charleston, a person

is charged with disorderly conduct if he or she causes a

danger or a breach of the peace, makes offensive, disgusting

or insulting remarks, or molests and interferes with any

person lawfully in any public area (adapted from the

Charleston City Code, Sec. 21-191). This information was


obtained from the Identification Section and is coded as

either present or absent.

The variable SUSINTOX refers to whether or not the

reporting officer noted that the subject had been drinking

prior to the assault. This was either indicated by a box

which could be checked (1991 reports) or by the officer's

narrative (1988-1990 reports). The information is coded as

either present or absent.

Officers (OFFAGE) who were 29 years old or less were

considered "younger"; those 30 or older were coded as

"older."7 The same coding convention was used to

dichotomize the ages of the subjects (SUSAGE).

OFFEXP was based on the number of months the reporting

officer had worked for the Charleston Police Department. If

the officer had prior policing experience at another agency,

it was not included in this study. Officers with more than

48 months of experience with the Charleston Police

Department were considered more experienced; all others were

less experienced.

DIS refers to how the officers received the dispute

call. The coding reflected whether the encounter resulted

from the police radio dispatcher or was on sight or on view.

If the disputants were still involved in an altercation when

officers arrived, the dispute was coded as in progress

7 In a study of police use of force in Rochester, NY,
Croft (1985) used the age of 30 to divide older and younger


(INPROG). This information was taken from the officer's

report narrative.

REL refers to the relationship between the subject and

the victim. Information was coded as related or married

(husband-wife; brother-sister, father-son) or acquaintances

(boyfriend-girlfriend, neighbors).8

The variable EMP refers to whether the victim was

employed at the time of the dispute. The information was

taken from the business phone and the employer/occupation

box listed for the victim on the report. This was coded

either employed outside the home or not so employed.

The variable VVISINJ refers to whether the officers

observed any visible injuries to the victims. This was

obtained from the narrative of the incident report and coded

as either present or absent.

The variable FELONY refers to whether the initial

domestic assault was a misdemeanor or a felony. Aggravated

assaults, i.e., assaults involving the use of deadly

weapons, are felonies; however, minor or simple assaults,

i.e., involving hands, are usually misdemeanors.

Both the variables WITNESS and LOC refer to the

visibility of the police-citizen encounter. The first

addresses individuals actually present (other than

disputants); the second deals with the potential visibility

8 Acquaintances were rarely living in the same
household and had different addresses on the police incident


of the encounter (Friedrich, 1980). Witness was coded as

present or absent; location was coded as in a dwelling or

outside a dwelling in a public area.

Several hypotheses refer to differences between

subjects and authorities (see Hypotheses 3-5). AGEDIFF was

computed by subtracting SUSAGE (suspect's age) from OFFAGE

(officer's age). For GENDIFF, the suspect's gender (SUSGEN)

was subtracted from the officer's gender (OFFGEN). RACEDIFF

was constructed by taking the difference between suspect's

race (SUSRACE) and officer's race (OFFRACE). The resulting

difference measures in each case were coded same or

different for analysis.

Data Analysis

Descriptive, bivariate and multivariate analyses were

used to present the data. Crosstabular analyses and Chi-

square tests assessed whether norm resistance related to

each independent variable. For example, do racial

differences between police and batterers affect the

likelihood of norm resistance occurring? Are subjects who

have been arrested previously more likely to offer

resistance? The main assumptions of the chi-square test

permit it to be used with nominal-level variables. Random

sampling (or a stratified random sample) is recommended;

there should also be at least five cases in each cell of the

cross-tabulation (Agresti & Finlay, 1986). While the


significance level commonly used in social sciences is p

<.05, Blalock (1979, p. 161) advises, "there is nothing

sacred or absolute about these levels [I]f one

actually did not want to reject the null hypotheses, he or

she would be on safer ground using the .10, .20 or even .30

level, thereby reducing the risk of a type II error."9

Since this study is exploratory, the significance level is

set at .10.10

T-tests were also used to examine the differences in

the means of the norm resister group and the nonresisters on

several interval level variables, e.g., officer age, officer

experience and subject age (see Appendix B). The main

assumptions of a t-test are that the population distribution

is normal, the variables can be measured on an interval

scale and the sample was randomly selected (Norusis, 1988).

According to Agresti and Finlay (1986, p. 141), "The t-test

for a mean is quite robust against violations of the

assumption that the population distribution is normal."

The number of indicators present in each of the three

constructs (basic conflict, sophistication, organization)

were added into separate indexes. More indicators of

cultural differences, organization, and a lack of

9 Norusis refers to a Type II error as "not finding a
difference when there is one" (1988, p. 209).

10 In their exploratory study of police executive
stress, Crank et al. (1992) used a .10 significance level
order to avoid making a Type II error.


sophistication were hypothesized to relate to norm

resistance at the domestic dispute. Each indicator was

coded as either + 1 or 0 (present or absent). There were

five indicators of basic conflict (CONFLICT) so the conflict

index ranged from 0 to 5 (see Hypotheses 1-5). The

indicators are: differences in the race of authorities and

subjects (RACEDIFF); differences in ages of authorities and

subjects (AGEDIFF); differences in the gender of authorities

and subjects (GENDIFF); previous arrests of batterers

(PREARRST); and batterers' previous arrests for crimes of

violence or disorder (PREVIOL).

There were five indicators of the sophistication

construct (SOPH) and its index also ranged from 0 to 5. The

indicators of lack of sophistication are: suspect

intoxication (SUSINTOX); whether the dispute was in progress

when police arrived (INPROG); how the dispute came to the

attention of the police because it was on-view (DIS); less

experience of the officers (OFFEXP); and younger age of the

officers (OFFAGE).

A similar organization index was constructed from six

organization indicators. The range of the organization of

subjects (ORGAN) was from 0 to 6. Indicators of more

organization are: no visible injuries to victims (VVISINJ);

the disputants were related or married (REL); the presence

of witnesses (WITNESS); more than one disputant was arrested


(NUMARR); a public location of the dispute (LOC); and the

charge was or would be only a misdemeanor (FELONY).

These indices were used in several different analyses.

First each one of them was crosstabulated with norm

resistance. Second, they were simultaneously entered into a

discriminant function analysis to see how well they could

discriminate norm resistance cases from nonresistance cases.

According to Klecka (1980), discriminant function

analysis has seven assumptions: (1) two or more groups are

required, (2) there are at least two cases in each group,

(3) any number of discriminating variables can exist

provided it is less than the total number of cases minus

two, (4) the discriminating variables are at the interval

level, (5) no discriminating variable is a linear

combination of other discriminating variables, (6) the

covariance matrices for each group is approximately equal

and (7) each group was drawn from a population with a

multivariate normal distribution on the discriminating


The first three assumptions are obviously satisfied

with these data. Assumption 4 is also met in that the

indices are counts of how many indicators are present. The

zero order correlations show that the three indices were not

linear combinations of each other. The correlations were

-.03712 between sophistication and organization, -.03308

between organization and culture conflict, and .04378


between culture conflict and sophistication; assumption 5 is

satisfied. The covariance matrices for the indices are not

significantly different according to Box's M (Box's M =

8.2586, p = .23). Thus assumption 6 is met. Klecka (1980,

p. 62) maintains that discriminantt analysis can be

performed when the assumptions of multivariate normal

distributions and equal group covariance matrices are not

satisfied." Violation of these assumptions may make it more

difficult to determine how many functions help discriminate

among groups (not applicable to our two group analysis). In

addition, it may lead to some loss of ability in

classification of cases; however, the technique is robust if

the percentage of correct classifications is high. Thus,

the technique is robust even if the multivariate normal

distribution (Assumption 7) is not met.

Discriminant function analysis yields a variety of

helpful summary statistics. It produces standardized

discriminant function coefficients that indicate the

relative magnitudes of the respective variables in

discriminating between groups. Cannonical correlations

provide estimates of the overall predictive utility of the

variables that are included in a discriminant function. The

square of the cannonical correlation can be interpreted as

the amount of variation between the groups accounted for by

the discriminant function. The technique also performs a


classification analysis to reveal how many cases are

correctly classified by the discriminant function.

One way to establish the efficacy of discriminant

function analysis is to conduct a proportional reduction in

error (PRE) statistic for the classification. Klecka (1980,

pp. 50-51) offers the formula for such a statistic, tau.

Tau is used to see how much better than chance the Turk

indices perform. If the reduction in error is substantial

we know that any violation of assumptions did not change the

analysis or findings.

A logistic regression model was used to regress the

norm resistance (a binary dependent variable) on each of the

respective indicators of culture conflict, sophistication,

and organization. Logistic regression requires fewer

assumptions than multivariate techniques built on least-

squares, such as multiple regression or discriminant

analysis (Norusis, 1988). Logistic regression analysis

requires the number of cases be at least four times the

number of independent variables (Tabachnick & Fidell, 1983).

A parsimonious set of independent variables is preferred.

Multicollinearity may be a problem in most regression

analyses, as some independent variables are likely to be

intercorrelated, (e.g., officer age and officer experience).

However, Schroeder et al., (1986, p. 72) caution, "there is

no single preferable technique for overcoming multi-

collinearity, since the problem is due to the form of the


data." A correlation matrix or phi associations are often

used to search for multicollinearity between interval or

nominal level variables (see Appendix B). The independent

variables that were insignificant in the crosstabulations

were excluded from the logistic regression analysis. Logit

coefficients, standard errors, and p values are reported for

the effects of each independent variable on norm resistance.

Savannah, GA

The Savannah data were collected to investigate

hypothesis 20. The Savannah Police Department serves a

community of 151,288 residents with blacks comprising 56% of

the city population. The patrol area is almost 51 square

miles and there are 328 sworn police officers and 89

civilian employees." One very important reason for

selecting the Savannah Police for this research is because

it used a separate report code for domestic disputes before

and after the proarrest policy began.12 A one-group

before-and-after design was used to examine the arrest

policy and its impact on police-citizen conflict. This pre-

experimental design is often called a one-group pretest-

This information was obtained from a report
submitted to Savannah Police Chief David Gellatly (Knoche,
12 If the police department did not separately code
domestics, the author would have had to read every general
disturbance call (e.g., stranger assaults, bar fights, loud
parties, disorderly persons) to filter out the domestic
assault cases.


posttest design (Babbie, 1989). One inherent weakness of

this design is that it fails to control for other factors

(e.g., historical events) that may cause changes from the

pretest to the posttest. However, after interviewing

several members of the Savannah police, the researcher is

unaware of any unusual circumstances that occurred in

Savannah during this time period.

The independent variable is Savannah's presumptive

arrest policy, which was implemented on May 27, 1987. It


Anytime any officers respond to a domestic
disturbance where a criminal act has occurred, and
there is sufficient probable cause to identify the
perpetrator along with the substantiating of the
charge, an arrest may be made by officers as
defined by State Code. (Savannah Police Standard
Operating Procedure 1-20-11)

A sample of 253 domestic violence reports from January

1987 through February 1988 was collected from the Savannah

Police Records Unit. Officers in Savannah are requested to

write reports on all domestic disputes to which they are

dispatched. However, some of these calls were "unfounded"

in the judgment of the responding officers.13 Therefore,

only those reports where officers found evidence of a

domestic assault were included in this study. In the five

months prior to the implementation of the presumptive arrest

13 The researcher's prior experience as a police
sergeant was very useful in making sure that unfounded
reports did not satisfy the elements of a domestic assault
crime. Unfounded reports generally involved calls where
victims were not located.


policy (January 1, 1987 to May, 26, 1987), 326 reports were

found dealing with domestic problems. A systematic sample

of half those reports was drawn. Of these 163 reports, 57

dealt with child abuse, child neglect, or abandonment and

were dropped. Another 10 reports were incomplete, leaving a

sample of 96 valid domestic violence incident reports.

Reports written prior to January 1, 1987 were not collected,

since the police department no longer had these on their

computer.14 During the first nine months after the policy

(May 27, 1987 to February 28, 1988), 843 reports were

available according to the computer lists. A one-in-four

systematic sample was taken and after all of the reports

dealing with child abuse, neglect and those that were

incomplete were deleted, 157 incident reports remained that

could be analyzed. With these samples, the impact of

Savannah's proarrest policy was examined to see if there was

any variation in the rate of police-citizen violence.

Of course, impacts can only be anticipated if the

arrest policy was complied with by the patrol officers.

Therefore, the arrest rate was also examined before and

after the policy. If officers did in fact arrest more

frequently after the policy began, the probability of

14 A computer list of the domestics was available for
the current year (1992) and the previous five years (1987-
1991). Therefore, only five months of reports during the
pre-presumptive arrest period could be obtained.


conflict should increase, as the authorities' cultural norms

and social norms become congruent.

Norm resistance was crosstabulated by the presence or

absence of a proarrest policy. A chi-square test was used

to assess the significance of differences in the number of

cases of norm resistance and nonresistance before and after

Savannah had implemented the proarrest policy.


Charleston, SC

Descriptive Statistics

This section describes the variables relevant to Turk's

conflict, sophistication, and organization themes. The

frequencies and percentages for each variable are introduced

below. Similar descriptive information from other studies

is presented in footnotes to provide some points of


Turk's (1969) basic conflict assumption suggests that

when great differences exist between the experiences of

authorities and of subjects, conflict is more likely to

become overt. Table 4 displays the descriptive information

relevant to the basic conflict assumption and cultural norms

of the subjects. Slightly more than half (56%) of the

officers in the study were white and most of the officers

were males (90.6%). While whites constituted 57.2% of the

general population of Charleston, only 28.7% of the subjects

(suspects) were white.1 Most of the subjects (94.1%) in

1 Due to the small number of Hispanics and Asians in
the sample, blacks, Hispanics and Asians were combined into
a non-white category. In a recent study of racial