:2.

:23. $53.5:
. I

. D, .s. 5.5... Z:— 93..

 

 

 

 

 

"THESIS

54W” 5/

This is to certify that the
dissertation entitled

POLICE RESPONSES TO MISDEMEANORS: GENDER
DIFFERENCES AT POLICE ARREST

presented by
Alex Obi Ekwuaju

has been accepted towards fulfillment
of the requirements for the

Ph. D. degree In Criminal Justice

 

 

flaw

MajorP essor’s nature

W 003

Date

MSU is an Affirmative Action/Equal Opportunity Institution

 

 

 

‘--.-.--—A- -.—.- .—

on-vQ-A-n-n-n-n-.-‘—-.-‘-_I-.-1--c-n--o-.----a-o--c-.-n-n-a-a-n-.-.--n-c--n--n--n.

 

 

LIBRARY ‘
Michigan State

University

 

 

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

 

DATE DUE

DATE DUE

DATE DUE

 

 

H 3' U I”;

ha-
pn-

 

APR 6 2 2003

 

.—
"‘ -7- .— w
b'

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/ClFlC/DaIeDue.p65-p.15

POLICE RESPONSES TO MISDEMEANORS:
GENDER DIFFERENCES AT POLICE ARREST

By

Alex Obi Ekwuaju

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
School of Criminal Justice

2003

ABSTRACT

POLICE RESPONSES TO MISDEMEANORS: GENDER DIFFERENCES AT
POLICE ARREST

By

Alex Obi Ekwuaju

A review of studies of police behavior at arrest indicates that two main separate
traditions exist regarding differential treatment relative to gender. These traditions are
contradictory in their description of how police respond to male and female suspects.
The chivalrous perspective asserts tlut police officers have historically treated female
suspects more leniently than male suspects. This protective stance is a result of
culturally defined sex role differences and depends on set traditional gender
expectations and clmracteristics that are said to exist between men and women.

An opposing view suggests, contrary to the chivalry argument but consistent with
the labeling argument, that female suspects are more likely tlmn their male counterparts
to be dealt with in a more severe and formal way. Labeling theorists suggest that
individuals with lower status and less power are more likely to have their criminality
detected, labeled and sanctioned because these individuals do not have resources to
manipulate the system to their benefit. Since women generally occupy less powerful
positions in society and have fewer economic and political resources at their disposal,

they may be more likely than males to be treated in a more severe and formal way.

Given these two opposing perspectives (chivalry and labeling perspectives), a
general hypothesis can be derived regarding police reactions to male and female
suspects: police will process female suspects differently than male suspects at arrest.

This research reports an empirical study of police processing of offenders in low-
level offenses where officers can exercise a great degree of discretion. Confidential
questionnaires were completed by police officers and detectives attached to a relatively
large police department in a Midwestern state. The officers were asked to indicate how
they would respond in four hypothetical offense scenarios that varied by types of
offense, and gender, race, and demeanor of the suspect.

Findings indicate that overall, the gender of the suspect and tlmt of the officer are
not directly related to police arrest decisions. Thus, neither the chivalry nor the labeling
perspective was supported by the results of this study. Police arrest decisions are not
contingent on the race of the suspect. The major variables in determining how police
officers will respond are the nature of the offense and the demeanor of the suspect.
There is additional evidence that the race of the officer is related to police processing of
suspects at arrest. Finally, this study finds that police have a high degree of confidence
in the arrest decisions they make. Findings are reconciled with existing literature.
Implications for official indicators of male/female arrest statistics and policy are

discussed.

Copyright by
Alex Obi Ekwuaju
2003

This work is dedicated to the Blessed Virgin Mary, “Mater Dei,” and to my father,
Akunne Afam Ekwuaju (deceased) and my two amiable sons, Alex Obi Ekwuaju, Jr.
and Michael Olisa Ekwuaju.

ACKNOWLEDGEMENTS

I have no doubt in my mind that it would take a lot of space to thank all those
whose assistance, support and encouragement were vital to the completion of this
study. I am gratefitl to all concerned.

Some people, however, were particularly most instrumental to the completion of
this work and I would like to acknowledge and thank them. From the depth of my heart,
I thank the members of my dissertation committee: Drs. Vincent Hoffman, David
Wiley, and Ken Williams. Worthy of special mention is Dr. Charles Corley, chair of the
committee. At all material times, these accomplished and highly respected scholars,
voluntarily placed at my disposal, out of the abundance of their expert knowledge of the
issues, invaluable advice and suggestions. Their guidance and encouragements kept me
on the right track and ultimately ensured the success of this journey. I am immensely
grateful to all of them.

I would also like to thank the Chief and all officers of the police department
where I conducted this study for their assistance, professionalism, and cooperation.
Lastly, but really first, I thank Almighty God for His infinite mercies, blessings,
guidance, and protection. I alone, however, remain responsible for the contents of this

doctoral dissertation and the views herein expressed.

TABLE OF CONTENTS

LIST OF TABLES .............................................................................. ix
LIST OF FIGURES ............................................................................ xi
CHAPTER ONE: Statement of the Problem .............................................. l
The chivalry perspective .................................................................. 4
The labeling perspective .................................................................. 6
CHAPTER TWO: Review of the Literature ............................................... 9
Introduction ................................................................................. 9
Gender and police arrest decisions ....................................................... 11
Intersection of Race and Gender ........................................................ 17
Summary .................................................................................... 26
CHAPTER THREE: Methodology ......................................................... 27
Data collection techniques ............................................................... 27
The questionnaires ........................................................................ 30
The sample ................................................................................. 33
Research question ......................................................................... 34
Research hypothesis ....................................................................... 35
Null hypotheses ............................................................................ 35
Variables .................................................................................... 35
Questionnaire coordination and control ................................................ 36
CHAPTER FOUR: Analysis and Results ................................................. 38
Demographic characteristics of police respondents .................................. 38
Summary of responses of the officers .................................................. 42
Preliminary Findings from a summary of the oflicers’ responses .................. 49
Gender differences in police arrest decisions .......................................... 53
Impact of gender of the suspect .................................................. 53
Impact of the gender of the officer .............................................. 61
Race differences in police arrest decisions ............................................ 63
Impact of race of the suspect ..................................................... 63
Impact of race of the officer ...................................................... 66
Summary of findings ..................................................................... 72
CHAPTER FIVE: Discussion and Conclusions ......................................... 73
Cautionary remarks ...................................................................... 80

vii

Implications for official arrest statistics ................................................ 84

Policy implications ....................................................................... 85

Directions for future research ........................................................... 86
APPENDIX A: The Questionnaire (Type 1) .............................................. 88
APPENDIX B: SPSS Coding Guide ....................................................... 96
APPENDIX C: Alphabetical List of Variables .......................................... 100
APPENDIX D: Other Tables and Figures ............................................... 102
BIBLIOGRAPHY ............................................................................. 107

viii

LIST OF TABLES

Table 1. Characteristics of police respondents .......................................................... 40

Table 2. Officers’ responses to the four public drunkenness scenarios by
suspect race and gender (demeanor is constant = negative) ....................................... 45

Table 3. Officers’ responses to the four shoplifting scenarios by suspect race
and gender (demeanor is constant = positive) ............................................................ 46

Table 4. Officers’ responses to the four assault scenarios by suspect race and
gender (demeanor is constant = negative) ................................................................. 47

Table 5. Officers’ responses to the four tratflc scenarios by suspect race and
gender (demeanor is constant = positive) .................................................................. 48

Table 6. Summary of the distribution of officers’ responses to all offenses as
contained in all four types of questionnaires (suspect demeanor is constant) .............. 50

Table 7. Oflicers’ levels of confidence in their decision-making across all
four types of offenses ............................................................................................... 51

Table 8. Gender differences in police arrest decisions controlling for type
of offense, race and suspect demeanor ...................................................................... 58

Table 9. Independent t-test comparing arrest decisions of male and female
officers across all four offenses ................................................................................ 62

Table 10. Race differences in police arrest decisions controlling for type of
offense, gender and suspect demeanor ...................................................................... 65

Table 11. Independent t-test comparing arrest decisions of black and white
oflicers across all four offenses ................................................................................ 68

Table 12. Independent t-test comparing arrest decisions of white and Hispanic
officers across all four offenses ................................................................................ 69

Table 13. Independent t-test comparing arrest decisions of black and Hispanic
officers across all four offenses ................................................................................ 70

ix

Table 14. Independent t-test comparing arrest decisions of white and
minority ofl’icers across all four offenses .................................................................. 71

Table 15. Number of police officers and types of questionnaires completed

and returned ........................................................................................................... 104
Table 16. Gender differences in police arrest decisions: SPSS group statistics ....... 105
Table 17. Race differences in police arrest decisions: SPSS group statistics ............ 106

LIST OF FIGURES

Figure 1. Distribution of hypothetical offense episodes, demeanor, race,

and gender across the four types of questionnaires. ........................................ 102
Figure 2. Race and gender composition of all “police oflicers” in the department
Where data was collected ..................................................................... 103

xi

CHAPTER ONE
STATEMENT OF THE PROBLEM

In less serious offenses such as misdemeanors, the gender of the suspect has long
been recognized as playing a critical role in police arrest decisions (Walker et al., 2000;
Visher, 1983; Pastor, 1978; Price, 1977; DeFleur, 1975). In spite of this awareness,
research on how female suspects fare at arrest has for the most part been neglected.
Even with the "recent" increase in interest in the study of female criminality (e.g.,
Meyer 1981; Sloan, 1991), most of the studies focus almost exclusively on how
women, as compared to men, fare at the hter stages of the criminal justice process (e.g.,
sentencing decisions, as in Moulds, 1980; Kruttschnitt, 1981). There is therefore a
"general failure to study how women fare during initial stages of the criminal justice
processing, such as arrest" (Sloan, 1991: 106). Arrest is an important issue and stage in
the criminal justice system. In general, it marks the entry point into the sanctioning
apparatus of the system. A police decision not to arrest, in essence, constitutes an
"acquittal" which most of the time immunizes the suspect from firrther action by the
criminal justice system (Thomson-Rogers, 1991). The vacuum created by the absence
of research in this area is even more significant when one realizes that the percentage of
female arrestees is continuously increasing. According to the Uniform Crime Reports
(U CR) for example, the percentage increased from 10.6 percent of all arrests in 1958, to
16 percent in 1978, and to 21.8 percent in 1998 (FBI, 1993: 234; FBI, 1998: 227). Do
these figures represent changes in actual behaviors or are they more related to public
attitudes and police practices?

Research on gender differences at arrest have been criticized on methodological

grounds. There has been virtually no previous research attempting to quantitatively
amlyze this issue. Most of the analyses to date have taken the form of ethnographic
research and case studies (see Nees, 1986; Visher, 1983; Lundman, 1974). Moyer
(1981: 237) asserts that most research in this area is "based on speculation and
theoretical assumptions about the meaning of data acquired fi'om secondary sources, for
example, the Uniform Crime Reports." These studies, according to Visher, (1983z7),
"are often inadequate because of incomplete data or poor design." According to Meyer
(1981 :237), “the need for original empirical research is greatest in the area of the
treatment of female offenders by police officers where there appears to be an
insufficient interest in research.” Sloan, (1991: 107), put it succinctly:

“. . .many of the studies [on the response of the criminal justice system to female
offenders] suffer fi'om design flaws, incomplete data collection, and other
methodological problems that have resulted in a lack of consensus among researchers
as to how women fare during criminal processing.”

It is therefore submitted that there is a general need for original empirical
research on gender differences at arrest utilizing a strong research design. It must be
added however, that although the existing studies are immensely useful, more
systematic quantitative evidence is also needed to investigate this issue.

Another compelling reason that justifies further inquiry into gender differences at
arrest is that results of studies that focus exclusively on female processing at the later
stages of criminal justice processing (e.g. sentencing), may be compounded by the
effects of what occurs at initial stages, including arrest. For example, if one assumes
that the female/male population ratio is 50:50, and a study finds that a significantly

smaller number of females, compared to males, are processed at the judicial (court)

stage of the criminal justice process, this may not necessarily mean that females

commit fewer offenses than males. A plausible alternative hypothesis may be that
females commit as many offenses as males but they are afforded preferential treatment
at the earlier stages (e. g., arrest and probably also during the District Attorney’s
decision whether to initiate criminal prosecution or not). It now appears well settled
that one effect of eliminating discretion at one stage of the criminal justice processing is
to foster it at other stages (Alschuler, 1978). The reverse of this may also be true. This
is because the criminal justice system is composed of interdependent and connecting
agencies or components rather than separate or independent agencies or arms. What
occurs in one stage therefore affects the others. According to Feeley (1979:204), the
criminal justice system “is a complex hydraulic system, and pressure exerted at one
point inevitably produces deformation at another.” It is therefore necessary to study
gender differences not only at the later stages, but also at the initial stages of entry into
the criminal justice system. In this way, the effects of any prior occurrence that
compounds the results of gender differences at later stages are more likely to be
identified and explained.

Visher (1983) makes an additional point that, while some studies have analyzed
the effects of gender on police arrest decisions, the conclusions reached by researchers
concerning the effect of gender on arrest decisions are mixed. For example, in
attempting to explain why women might be less likely to be arrested than men, some
researchers have concluded that this is due to male officers fearing that charges of
“sexual harassment” might be brought against them by the female suspect (see Sloan,
1991). Some analysts have concluded tint the reason why women are less likely to be

arrested is that male officers believe that women are “unpredictable” in arrest situations

(see Sloan, 1991). Still others have concluded that the reason why women are less
likely to be arrested is because eflicers believe there is a “cultural prescription” against
the use of coercion against women, such as that required to nuke an arrest, which
serves to “check” the officer’s behavior (c.f. Bayley & Mendelsehn, 1969; Friedrich,
1977; Niederhoffer, 1967; Rubinstein, 1973). In addition to the above, there is a
growing number of criminologists who report that instead of receiving preferential
treatment by the criminal justice system, female offenders actually fare much worse
than their male counterparts (Feinman, 1979; Smart, 1976). Other studies report that
there are no direct effects of gender on the arrest decisions of police officers. Instead,
the seriousness ofthe offense and the demeanor ofthe suspect appear to be the best

predictors of arrest decisions (Meyer, 1981; Visher, 1983).

The Chivalry Perspective

Given the above, a prudent and meticulous review of studies of police behavior at
arrest indicates that two main separate traditions exist regarding differential treatment
relative to gender. These traditions are contradictory in their description of how police
respond to male and female suspects. A long line of serious researchers and
academicians who belong to the chivalrous perspective (e.g., Visher, 1983; Lundman,
1974; Pastor, 1978; Sealock and Simpson, 1998; Moulds, 1980; Datesman & Scarpetti,
1980; Anderson, 1976; Pollak, 1950), assert that police officers and other law

enforcement officials have historically treated female suspects more leniently than male

suspects. Supporters of the chivalry perspective suggest that females are less likely than
males to have their criminal acts detected and sanctioned (Moulds, 1980). In fact, it has
been noted that police are less inclined to arrest female suspects (Pollak, 1950; Visher,
1983). According to Haskell and Yablonsky (1973:6) “police are less likely to arrest
women than they are men under identical circumstances.”

Moreover, it was observed that when arrested, females are more likely than males
to have their charges dismissed (Bernstein, Cardascia, & Ross, 1977). Furthermore, it
has been suggested that, among convicted offenders, females received less severe
sentences than their male counterparts even for similar crimes (Corley et al. 1980). This
protective stance is a result of culturally defined sex role differences between males and
females: women are defined as sexual beings whose primary purpose is to perform
utilitarian functions in the home (Kleinl973:3). The arrest and subsequent removal of
women from the home has been discouraged because it would threaten the solvency of
the nuclear family (Crites, 1976). Since many female offenders have children, to
incarcerate women would place a burden on the rest of society (Crites, 1976).
Similarly, the widespread practice of incarcerating females may expose them to an
environment detrimental to their role as mothers (Chesney-Lind, 1977). Thus, the
chivalrous treatment of females by the criminal justice system is intended to preserve
the social order (Harris, 1977). In fact, Visher (1983: 6) suggests that a “chivalrous
relationship should be thought of as a barter or exchange,” that is, women receive
preferential treatment in return for displaying appropriate sex role behavior. Chivalry
then exists because of the ways in which women have been defined. These definitions

depend on sets of traditional gender expectations and characteristics that are said to

exist between men and women (Corley, Cernkovich, & Giordano, 1989). The line of
authorities cited above raises a number of important implications. For example, if this
chivalrous attitude is true, then differences in the levels of officially recorded
criminality of males and females could be explained in part by patterns of police

responses toward female ofl'enders.

The Labeling Perspective

An opposing view suggests that law enforcement oflicials treat females more
harshly than males. In this view, female suspects are more likely than their male
counterparts to be dealt with in a more severe and formal way - that is, arrested and
referred to court — especially when their alleged criminal acts violate appropriate sex
role behaviors (Chesney-Lind, 1977). There is evidence that a woman who commits the
same crime as a man is more likely to: (1) be incarcerated; (2) receive a longer term of
sentence; or (3) actually serve a longer sentence (see Feinman, 1982; Srmrt, 1976).
Instead of reacting to female offenders in a “chivalrous’ manner, criminal justice
officials may in fact perceive female felons as being “double offenders”: they have not
only violated the law, but have also violated appropriate gender role behavior (Schur,
1983). This position suggests, contrary to the chivalry argument presented above but
consistent with the labeling argument, that females are more likely to be labeled deviant
and receive harsher sanctions than their male counterparts. Labeling theorists suggest

that individuals with lower status and less power are more likely to have their

criminality detected, labeled, and sanctioned because these individuals do not have
resources to manipulate the system to their benefit (see Curran, 1983). Because women
generally occupy less powerful positions in society and have fewer economic and
political resources at their disposal, they may be more likely than males to have their
deviance detected, labeled, and sanctioned (see Corley, Cemkovich, and Giodano,
1989:542).

To further complicate matters, Chesney-Lind (1977:51-52) argues that courts
actively “sexualize” offenses by reclassifying specific violations as sex-role or sexual
violations. According to her, the tendency is to punish female behavior more severely
when it falls within these categories. She asserts, for example, that juvenile court is less
tolerant of female runaways than male runaways. The absence of supervision and the
possibility of sexual relations occurring while away from home seem to elicit more of a
negative response toward female than male runaways. Therefore, female runways are
often incarcerated “for their own protection,” or for purposes of instilling a sense of
“sexual morality” (at page 23).

Apart from the above‘two main views, some evidence supports the proposition
that female offenders are treated no differently than males. In the juvenile jurisdiction
for example, Curran found tint over a ten-year span (1966-1976), females and males
received equal treatment by the juvenile court (Curran, 1983). In addition, Corley,
Cemkovick and Giordano (1989: 553) assert that, “contrary to the labeling and chivalry
perspectives, findings suggest that judicial sanctions operate independently of sex, race,
and age. The research offers support for a legal model in which legal variables such as

the seriousness of the offense are important factors in judicial sanctions.” — (See

Klinger, 1994; Meyer, 1980). Overall, however, the literature has emphasized two
extremes in sanctioning, thereby implying that male and female suspects are rarely
treated similarly. Given these two main opposing perspectives (chivalry and labeling
perspectives), a general hypothesis can be derived regarding police reactions to female
suspects: police will process female suspects differently than male suspects at arrest.

Lastly, it is well-known that gender has been identified as one of the most
consistent extralegal factors that influences criminal justice personnel and juries
concerning offensive behaviors (see Muraskin and Roberts, 2002; Walker, 1998).
Against this background, Muraskin and Roberts (2002), most recently pontificated that
“no work in criminal justice is complete without referring to issues of gender ...”
Among other things, this dissertation is a modest attempt to help fill these lacunae or
gaps in the literature as described above.

Thus, this research focuses upon the treatment of women, relative to men, at the
point of arrest. Specifically, this research attempts to throw more light on the issue of
whether women are treated chivalrously, more punitively, or the same as males at the

point of arrest.

CHAPTER TWO

REVIEW OF THE LITERATURE

Introduction

According to Goldstein (1993), police decisions not to arrest offenders largely
determine the outer limits of law enforcement. By making such decisions, the police
define the ambit of discretion that can be exercised by other decision-makers involved
in the criminal justice process decision — prosecutor, grand jury, judge, probation
officer, correction authority, and parole and pardon boards. The police decision not to
arrest, unfike the decision to arrest, is generally of extremely low visibility and
consequently seldom the subject of review. Goldstein argues that an opportunity for
review and appraisal of non-arrest decisions “is essential to the functioning of the law
in our system of criminal justice” (p.78). Goldstein however suggests the necessity of
police discretion in processing offenders. He presents the difficulties of total law
enforcement (limitations of time, personnel, etc), to support his argument tlmt selective
law enforcement is essential for the criminal justice system to function effectively.
“Full enforcement,” according to him, “is not a realistic expectation.” (p. 81).

Bittner (1970; 1990), and the National Advisory Commission on Criminal Justice
Standards and Goals (1973), both state that police officers exercise considerable
discretion and that their decisions not only help define the limits of the criminal justice
system, but also have a profound impact on the overall administration of justice.

According to Nees (1986), law enforcement officers exercise broad discretion in

carrying out their duties. While most researchers, authors, practitioners, and the public

recognize this as fact, no one has developed a method or approach to help manage the
exercise of enforcement discretion. Herein lies the importance of Nees’ (1986)
research. His research involved the use of vignettes of enforcement incidents. Law
enforcement officers from Boulder County were asked to indicate from a series of
response options how they would handle the various incidents in the questionnaire.
Other groups of respondents (law enforcement supervisors, administrators, prosecuting
and defense attorneys, judges, probation officers, community leaders, and members of
the community) were asked to select fi'om a series of responses, the option which best
fit what they expected or desired officers to do in various hypothetical situations
contained in the questionnaire (Nees, 1986). The study was descriptive and exploratory
in nature. Results indicate that while differences do exist among the different groups of
respondents, the differences were not great. In brief, the differences were among the
options which had a lesser impact on the suspects (take no action, warn, or refer to a
social service agency). Thus, no significant differences exist among the various groups
of criminal justice personnel on how they think law enforcement discretion should be
exercised.

In addition to the works cited above, Thoresson-Rogers (1991) suggests tint
police work is essentially and basically discretionary. That discretion, however, is
almost always unregulated by formal guidelines. According to Powell (1990), police
exercise of discretion to arrest or to utilize other alternatives is often guided by the
values, role conception, biases, and emotions of individual officers.

Meyer (1981), however, found that there are some general "guidelines" that are

connected with the police fimction and the officer's concept of his/her role. Generally,

10

in serious cases involving a felony or certain misdemeanors, there will be an arrest. In
less serious cases, there may or may not be an arrest.

Smith (1982) examined 37 studies of police decisions to invoke the law and found
that existing studies clearly suggest that police arrest decisions in less serious
misdemeanor cases vary with the attributes of the offender such as race, sex, age, and
demeanor. Powell (1990), asserts tint a police officer may discriminate between
potential arrestees based upon factors such as race, age, or nationality. According to
Powell (1990), a police officer may arrest only those who prove uncooperative, sparing
the penitent; he may, in fact, make his arrest decision for the best or worse possible
reasons. Apart from the factors mentioned above (i.e., race, sex, and demeanor of the
suspect), other influential variables in the arrest process identified in the literature
include: the presence and wishes of the victim (Black, 1971; LaFave, 1965); and the
presence of an audience - arrest more likely in the presence of audience than otherwise

(Thoresson-Rogers, 1991).

Gender and police arrest decisions

What does the literature say about the relationship between the specific variable
gender and police arrest decisions? As noted earlier, this area has generally been
neglected in prior research. In a study on the criminality of women conducted more
than fifty years ago, Pollak (1950: 151) asserted that "men hate to accuse women and
thus indirectly to send them to their punishment, police officers dislike to arrest them,

judges and juries to find them guilty and so on." It is however submitted that this

I]

statement does not lmve what may seriously he considered an empirical backing.
Similar remarks were mde by Haskell and Yablonsky (1973), in an attempt to explain
the differences in the levels of officially recorded criminality of males and females. The
authors gave as one of their reasons for the lower female arrest rate that "police are less
likely to arrest women than they are men under identical circumstances." (p.6). To back
up this assertion, the authors argued that a man walking alone at night in a strange
neighborhood may be perceived by police and residents as suspicious, dangerous, or
potentially involved in a crime, but that a woman would seldom be viewed as a
potential criminal under those circumstances.

Price's research, based on UCR data, found tha "... the criminal justice system
has related to women differently fiom the way it does to men at every stage of its
process" (Price, 1977: 104). As an example, Price stated that although the ratio of
arrests was approximately six males to one female in 1970, admissions to state and
federal correctional institutions during that year showed a ratio of nearly 21 males to
one female. However, Price did not control for the nature of the offense in his study. It
is submitted that the three works cited immediately above (that is, Pollak, 1950; Haskell
and Yablonsky, 1973 and Price, 1977) consist mainly of research based on secondary
data such as the UCR, and, as described by Meyer (19812237), on "speculation and
theoretical assumption."

In his study of arrest discretion for the President's Crime Commission, Black
(1980) found that police officers consistently under-enforce the law, arresting only 58
percent of felony suspects and only 44 percent of misdemeanor suspects. The decision

to arrest is primarily influenced by situational factors. Officers are more likely to arrest

12

when the evidence is strong; the crime is of a more serious nature; the complainant or
victim requests an arrest; the relationship between the victim and offender is distant
(e.g., strangers rather tlnn acquaintances or spouses); and the suspect is disrespectful
toward the officer. In Black's study, the gender of the suspect was not a major
determinant of arrest decisions.

Pastor (1978) in his study of police-suspect interaction in public drunkenness

encounters included gender, race and demeanor of the suspect as independent variables.

The study was conducted in Seattle and Boston over a period of fourteen months in
1973 and 1974. Data were collected through participant observation on 1,321 cases.
Pastor found that female suspects are "very rarely arrested for public drunkenness” (p.
380). He also found that non-white suspects are more likely to be arrested and that
hostility increases the likelihood of arrest for public drunkenness.

Lundman (1974), undertook a study of routine police arrest practices in "a large
Midwestern city" over a 15 month period beginning in June 1970. Like Pastor, his
study also involved public drunkenness. A group of seven observers trained in the use
of an interaction code traveled with police on a random time sample basis and recorded
the data using portable electronic field encoding equipment. Random selection also
determined the police cars in which the observers rode. The final data consisted of
2,835 potential police-citizen contacts, and 1,978 actual encounters, 195 of which
involved public drunkenness. Lundman found that "the sex of the drunkenness offender
is not one of the conditions relating to the exercise of police arrest discretion" (p. 132).

Both studies reviewed immediately above are limited to public drunkenness and

both have relatively small samples of female suspects. Pastor's study involved 89

I3

I

females (13.6% of total data), while Lundman's study involved only 15 females or nine
percent of the total data used for the study.

Like Lundman (1974), Visher (1983) in her study of the extent of preferential
treatment toward female offenders at arrest also used a field research design. Trained
civilians riding during 900 patrol shifts observed 5,688 police-citizen encounters in 24
police departments in the St. Louis, Missouri; Rochester, New York; and Tampa-St.
Petersburg, Florida metropolitan areas. The data were collected in 1977 and were part
of a larger evaluation of police services. The observers recorded citizen and officer
characteristics and behavior, in addition to other features of the encounter (e.g., time,
place, and nature of offense). Offenses considered were grouped into four — violent,
property, disputes and public order. The final sample contained data on 785 police-
suspect encounters with 643 male and 142 female suspects. Visher’s study included
variables such as race and demeanor of the suspect, type of offense, victim-suspect
relationship, and location of the encounter. Results indicate that the type of offense was
found to be a powerful predictor of arrest for both males and females. Arrest probability
increases with the seriousness of the offense. Also race of the suspect was found to
influence the arrest decision for both male and female suspects. Blacks are more likely
to be arrested than whites. Results further indicate that chivalry exists at the stage of
arrest for those women who display appropriate gender behaviors and characteristics
(e.g., white, older, submissive, apologetic etc. - see pp. 17; 23). In general, the findings
indicate that female suspects who deviate fi'om stereotypic gender expectations lose the
advantage that may be extended to female offenders. Specifically, older, white,

submissive, and apologetic female suspects are less likely to be arrested than are

14

younger, black or hostile women. Based on the above findings, Visher (p. 24)
concluded:
"Thus the official picture of female criminality is not representative of
crime committed, in general, by females. Second, chivalrous attitudes
appear to play a role in patterns of police response toward female criminals.
Whether these patterns have diminished in the last several decades, thus
giving the perception that female crime is increasing is an open question.”

Friedrich (1977) in his study relied on observational data on the behavior of
patrolmen gathered by Reiss, for a study for the President’s Commission on Law
Enforcement and the Administration of Justice in the summer of 1966. In an attempt to
understand transactions between policemen and citizens, Reiss deployed a total of 36
observers to ride and walk with policemen in three large American cities: Boston,
Chicago, and Washington, DC. The actual observation of police behavior took place
over a six week period in each of the three cities (12 observers in each city). The
observers had two basic tasks. One was to record in a booklet the details of each and
every encounter that they observed between police and citizens, including the number,
characteristics, and behavior of the citizens involved, the reason for the encounter, and
both the formal and informal actions taken by the police in the situation. The second
task was to record in a separate booklet, which was filled out at the end of each shift,
summary descriptions of what had occurred during the shift - the number of
encounters, the kinds of people encountered, the characteristics of the territory patrolled
and the characteristics of the patrolmen observed. At the end of the six week period, a
total of 3,955 encounters involving a total of 11,422 citizens were observed.

The basic issue tackled by F riedrich’s study was why policemen act as they do?
More specifically, how do organizational, individual, and situational factors affect

police behavior?

15

Analysis of the data indicate that patrolmen in a traditional, “watchman-like”
department invoke the legal process less fi'equently, are sometimes more informal in
their manner toward citizens, and exert less effort than patrolmen in a more modern,
“legalistic” department. Offrcers’ actions also vary with their individual characteristics.
More experienced offrcers, compared with less experienced ones, invoke the legal
process less often, treat citizens more personally, and work less hard. Morale has little
effect on behavior. Further, black officers invoke the legal process, treat citizens

impersonally, and work hard more often than white officers work. Racial prejudice

w “ ant—men fl

leads to higher arrest and ticketing rates for blacks and has a complex impact on the
policeman’s manner toward blacks. Among situational factors - overall, the most
potent determinants - seriousness of offense and the preference of the complainant
most influence formal decisions, while the officer’s manner toward the citizen depends,
for the most part, on the citizen’s manner toward the officer. Other situational factors
play a role too. Perhaps most interestingly, the presence of other citizens and of a police
partner significantly influence police behavior.

Two studies focused on the impact of gender on police arrest decisions in drug
cases. Johnson, Peterson, and Wells (1977) examined arrest statistics retrieved from
police files in conjunction with self-reported data collected by the Response Analysis
Corporation in 1971. The study took place in Cook County, Illinois; Douglas County
(Omaha), Nebraska; and the Washington DC. metropolitan area. The sample consisted
of a total of 637 cases, out of which 19 percent were females. The authors concluded
that males have a disproportionally higher arrest probability (about five times higher)

for marijuana use than do females. However, fewer females are arrested because of

16

their predominant pattern of drug use in private residences, which is less likely to
attract police attention. Males are more likely to use marijuana in public which will
attract police attention.

DeFleur (1975), in a study of Chicago police practices in drug enforcement, found
that police officers tend to be lenient toward young female offenders if they display
certain sex role behaviors, such as crying. However, women who act in an aggressive or
hostile manner are labeled as uncooperative and are arrested.

Some ethnographic research suggests that police are reluctant to arrest women
because women are considered unpredictable in such situations, and because using
coercion as a means of controlling females is contrary to cultural norms (Bayley and
Mendelsehn, 1969). Additionally, charges of sexual harassment fi'om arrested females

are possible and are not easily defended by patrolmen (Rubinstein, 1973).

Intersection of race and gender
Do black female suspects receive any leniency at police arrest? Some studies

indicate that black women are more independent than many white women are because
they fiequently occupy the role of head of household, and females in black households
usually have equal or greater status than male members (see Black, 1980: 124-128;
Roberts, 2000). According to Roberts (2000: 63), “most black children in America are
born to unmarried mothers.” Gender expectations for black males and females are less
difl‘erentiated than for white males and females (see Black, 1980; Smith and Visher

1980). Therefore, black females may be less inclined than white females to display

l7

traditional gender behaviors in their encounter with police. If this is so, the implication
is that in the criminal justice system, black females may not receive preferential
treatment. This argument is supported by one study of court decisions which found that
black females, relative to white females, were given more severe dispositions than
black males relative to white males (Datesman and Scarpitti, 1980). In fact, in more
recent studies, Mann (1989, 1995), studied the treatment of minorities from arrest to
incarceration and found that arrest rates for black women exceed those of other ethnic
groups. In other instances, black women received higher bails, were not adequately
represented in court, received longer sentences, and served more time in prison.
According to Mann (1989: 95), “minority women offenders [are] doubly discriminated
against because of their gender and race/ethnicity status.” (See also Muraskin &
Roberts, 2002).

Meyer (1981), using similar variables to Visher (1983 - discussed in detail above)
but with a survey research design, studied police arrest decisions in September -
December, 1978. 282 confidential questionnaires were completed by police officers and
detectives in "a large metropolitan area in a Southeastern state." Each officer was asked
to indicate a police decision in five hypothetical situations or scenarios that varied by
type of offense, race, sex and demeanor of the offender. To measure the effects of the
independent variables (type of offense, race of the offender, sex of the offender, and
demeanor of the offender) on the decisions made by patrol officers and detectives (the
dependent variable), forty vignettes were constructed. An analysis of the responses of
the police officers indicates that decisions of police to arrest are not contingent on the

sex and race of the offenders. The major variables in determining how police officers

l8

will respond are the nature of the offense and the manner in which the offender behaves
when confi'onted by the officer. The author therefore concluded that it is possible that
the alleged "differential treatment of women by police is simply a myth. On the other
land, it is also possible that more data is needed to test the hypotheses." (p. 245).

The race of the officer does not appear to influence arrest decisions. African
American, Hispanic and Caucasian officers arrest people at similar rates and for
generally the same reasons. Black (1980), however found some evidence that Afiican
American officers were slightly more likely to arrest African American suspects, in part
because they appeared to be more willing to comply with requests for arrests made by
Afiican American citizens. He admitted, however that the subject has not been
researched as thoroughly as it needs to be (Walker, Spohn, and DeLone, 2000).

According to F inn and Stalans (1997), studies on gender biases in police decision
making have not directly examined how assailants’ gender affects offrcers’
interpretation of situational cues or their decisions in domestic assault situations. That
was prior to their study. Therefore, using hypothetical scripts and experimental
manipulation, their study examined how disputants’ gender and mental state affected
130 offrcers’ inferences about dangerousness, responsibility, credibility, and control
and, ultimately, their decisions to arrest or refer for involuntary civil commitment in
domestic assault cases. Ofiicers inferred that male victims of domestic assault had more
control over their actions and were more responsible than female victims. Mentally ill
assailants were viewed as more dangerous and less in control of their actions than
normal assailants. Gender influenced arrest decision through officers’ assessments of

disputants’ credibility and responsibility. Female mentally ill assailants were more

19

likely than males to be referred for involuntary civil commitment. The findings suggest
that both gender and mental state affect oflicers’ inference and decisions in domestic
assault cases.

Sealock and Simpson (1998), investigated police decisions to arrest and how bias
in police decisions can occur. Data are based on 1968-1975 police records incorporated
in the 1958 Philadelphia birth cohort. The analysis is based on the assumption that
police use type-script (and countertype) heuristics. According to the concept of type-
scripts, people, as social actors, are assigned roles according to “type,” which is
shorthand for any ascribed or achieved characteristics. Commonly identified types in
criminological research include gender, race, and socioeconomic status (SES). logistic
regression analyses were performed on police arrest decisions for offenses categorized
according to their gender type-script. In the aggregate and when other variables were
controlled, females were less likely to get arrested than their male counterparts, and
race and SES significantly affected the arrest decision. Among all offenses, the gender-
typing variable explained a large portion of the effect of gender alone on the arrest
decision. Within gender-type offense categories, evidence was found that officers
consider offense seriousness and, most notably, the number of prior police contacts in
arrest decisions. The latter played a slightly larger role in the arrest decision for females
than for males. Results were confounded by interaction with race and SES.

In 1980, Lawrence Sherman undertook a first attempt to organize, review and
codify the published quantitative research examining the causes of police behavior in
the United States up to the yearl980 (Sherman, 1980). That review provided a synopsis

of the factors known to influence service, detection, arrest, and force behavior. Since

20

Sherman’s summary of findings in 1980, scholarly interest regarding the causes of
police behavior has increased. Moreover, more sophisticated modes of analysis have
been utilized, producing a large body of findings on how individual, situational,
organizational, and community-level variables influence police behavior. Against this
background, Riksheim and Chermak (1993) reviewed relevant quantitative research
from the 1980s and compared and contrasted their findings with those cited in Sherman
(1980). In general, their findings indicate that although our understanding of the causes
of police behavior has become more refined since 1980, a number of questions remain
unanswered. Riksheim and Cherman’s research is considered a major and very
significant work. F or this review of the literature, a prudent and meticulous computer
and manual reference search of most criminal justice/social science indices was
conducted and Riksheim and Cherman (1993) remains the most thorough, and
comprehensive updated review of studies on the causes of police behavior since
Sherman’s review in 1980. Much of the remaining part of this concluding section of
this review of literature is therefore largely from Riksheim and Chermak’s 1993 review
of post-1980 literature on causes of police behavior and the authors’ comparison of
their findings with the position prior to 1980, as reviewed and codified in Sherman
(1980).

Riksheim and Chermak (1993) found there have been mixed findings between
pre-1980 and post 1980 research regarding suspect characteristics as causes of police
behavior. On the effect of gender on police behavior, for example, research prior to
1980 indicated that female suspects were less likely to be arrested than males

(Sherman, 1980:82). Sherman cited Friedrich (1977), and Lundman (1974) to support

21

this view. Further research conducted during the 19808 supported these findings (Sykes
and Brent, 1983:217; Smith and Klein, 1984:474). However, a considerable amount of
research from the 19805 indicated that gender was not an important predictor of arrest
(Smith and Visher, 1980:174; Meyer, 1981:244; Visher, 1983:22), and it remained
insignificant across department types, that is, fiatemal, legalistic, service, and
militaristic (Smith, 1982:28).

Riksheim and Chermak’s 1993 review also revealed mixed findings on the impact
of suspect race on the police arrest decision. Utilizing a variety of data sets and
examining various offenses, most studies from the 19805 found that race had no effect
on police arrest decisions. For example, utilizing the Police Service Study (hereafter
PSS) observational data, researchers found no relationship between race and decisions
to arrest when no complainant was present (Smith, 1986:330), when both complainant
and suspect were present (Smith 1984:246), in domestic disturbance encounters
(Worden and Politz, 1984:110), or in interpersonal disputes (Smith and Klein,
1984:475). The results held that regardless of whether departments were legalistic,
service, militaristic, or fi'aternal (Smith and Klein, 1984:89; Smith, 1984:28). Analysis
of other data sets indicated a similar lack of effect for drunk driving offenses
(Hollinger, 1984:178) and for data collected fiom police responses to vignettes (Meyer,
1981 :240).

A few studies, however have found a relationship between race of suspect and the
police decision to arrest. In analyzing encounters involving physical interpersonal
violence, Smith (1987:776) found that police were more likely to use a “penal style” of

control in situations in which both combatants were white than they were when both

22

were nonwhite. This contradicts the research cited by Sherman (1980:80) indicating
that police were more likely to arrest black suspects than white. Among other studies,
Sherman had cited Black (1971) to support his view that police were more likely to
arrest black suspects tlmn whites. Some recent research, however, has also indicated
that black suspects are more likely to be arrested (Smith and Visher, 1981:172; Visher,
1983:21; Smith et al., 1984:244). Visher (1983 :21) found the relationship to be much
stronger for females than males, while Smith (1984:244) found suspects’ race
influenced police arrest decisions for females only.

Some research fiom the 1980s has suggested that the suspect’s age is not a
significant predictor of police decisions to arrest (Smith and Visher, 1981:172; Visher,
1983:15; Smith, 1984:27). These results differ from those obtained in the previous
decade (pre 1980 years), which indicated that young suspects were less likely to be
arrested by the police (Sherman, 1980:82). Sherman had cited Lundman (1974) to
support this view. However, other research from the 19808 indicated that age does
influence arrest when the suspect is female (V isher, 1983: 15), when misdemeanants are
adults (Krohn et al., 1983:428), or when a department is legalistic (Smith and Klein,
1983:89; Smith, 1984:30).

Most research Item the 19803 examining the influence of a suspect’s demeanor on
police arrest behavior supported previous conclusions (Sherman, 1980:81) that
“uncooperative,” “abusive,” and “antagonistic” citizens were more likely to be arrested
than those who were “calm,” “cooperative,” and “quiet” (Smith and Visher, 1981:172;
Meyer, 1981:240; 1982:380; Visher, 1983:16; Smith and Klein, 1983:90; Smith et al.,

1984:244; Wooden and Pollitz, 1984:113: Smith, 1986:330; 19872778; Wooden,

23

1989:700). Furthermore, male suspects were found to be more likely to be arrested if
they had been drinking (Worden Pollitz, 1984:] 13). Interestingly, some research has
suggested that antagonistic behavior and intoxication have little effect on the arrest
decision (Waaland and Keeley, 1985364) and no effect in militaristic departments
(Smith, 1984:30). Police also have been found by some research to be more likely to
separate dispute when the combatants have been drinking (Smith, 1987:778).

Overall, individual officers’ characteristics appear to have little influence on
arrest although the influence of many variables remains unresolved. For example,
Sherman (1980: 75-76) cited research indicating that better educated offrcers made
more arrests. Sherman cited Bozza (1973) to support this view. The findings from the
1980s are inconclusive. Two analyses of the PSS data (Smith and Klein, 1983:84—85;
Worden, 1989:701) showed that an individual officer’s level of education had no effect
on arrest behavior. However, departments in which officers had higher levels of
education (operationalized as an indicator of police professionalism) had lower arrest
rates (Smith and Klein, 1983:84-85). Research by Sykes and Brent (1983:217,221)
firrther complicated the issue by indicating that higher mdian levels of officer
education led to increases or decreases in the severity of police sanctions depending on
differences in situational factors.

Sherman (1980:73) cited research indicating that less experienced officers made
more arrests and cited Friedrich (1977) to support this view. Research from the 19803
produced mixed findings. Some studies showed that officers’ length of service
increased the likelihood of arrest (Sykes and Brent, 1983:217; Meyers et al., 1989:182

for juvenile drunk driving suspects). Some studies showed that length of service

24

decreased the likelihood of arrest when measured at the department level (Smith and
Klein, 1983:84), and some indicated that length of service had no effect on arrest when
measured at the individual level (Smith and Klein, 1983:84); Worden, 1989:701) or
when the arrests involved adult drunk driving suspects (Meyers et aL, 1989:]82).

Sherman (1980:74-75) cited research (Friedrich 1977) suggesting that officer race
and attitudes influenced police arrest behavior. Research item the 19808 contradicted
these findings. Analysis of the PSS data indicated that race of officer had no effect on
police arrest behavior (Worden, 1989;710) and remained non-significant across
department types, that is, legalistic, service, militaristic, or fi'aternal (Smith and Klein,
1983:87-89). Similarly, most research indicated that attitudes had no statistically
significant effect on arrest decisions (Smith and Klein, 1983:88; Worden, 1989:687-
702; Meyers et al., 1989:182; Smith, 1990:43) although they may have affected arrest
behavior indirectly by interacting with situational variables (Worden and Pollitz,
1984:118; Worden, 1989:700).

Research in the 19808 on the effects of officer gender has clarified previous
research cited by Sherman. Sherman (1980:73-74) reported mixed findings, citing
studies that found female officers made fewer arrests than male officers and one study
that reported no difference (Bloch and Anderson, 1974). Research from the 19808
indicated that officer gender had no effect on arrest behavior (Worden, 19892701).

Finally, Walsh (1986:278) found that officers with high arrest rates were more
likely to be married and their wives were less likely to be working. The author

speculated that such officers may be making arrests to earn overtime pay.

25

Summary

As mentioned in chapter one - Statement of the problem, previous research shows
that two separate traditions exist regarding differential treatment by police relative to
gender. These traditions (chivalry and labeling perspectives) are contradictory in their
description of how police respond to male and female suspects. A general hypothesis
can be derived from the chivalry and labeling perspectives discussed earlier regarding
police treatment of males and females at arrest: Police officers will arrest male suspects
difi‘erently than female suspects for most types of mid-level, non-felony offenses

(controlling for type of offense and demeanor of the suspect).

26

CHAPTER THREE

METHODOLOGY

Data collection techniques

Different research methods, either as a whole by themselves, or in combination
with other methods, have been used to study gender differences at arrest. As seen from
the review of the literature above, the two most common research methods that have
been employed in this area are the survey and field research methods (see Pastor, 1978;
Lundman, 1974: Visher, 1983; Friedrich, 1977 and Meyer, 1981). Notwithstanding
that field observation which as said above has also been used in similar studies (e.g.,
Visher, 1983), is relatively inexpensive, generally yields a deeper understanding of the
issues involved, and is fairly flexible, the survey design was chosen in this study
mainly because of the problems of objectivity in measurement (instrumentation), which
are generally inherent in any field study. Field research may also sometimes be weak in
reliability and is also far more time consuming. Lastly, according to Maanen (1973),
“observation of police in naturally occurring situations is difficult, lengthy, and often
threatening ...” (p. 407).

This study employs the survey method of data collection utilizing closed-ended
survey. The survey is the most widely used data-gathering technique in sociology, and
it is used in many other fields, as well, including criminal justice (Neuman, 2000). In
fact, surveys are almost too popular, but despite the popularity, it is easy to conduct a

survey that yields misleading or worthless results. Surveys are appropriate for research

27

questions about self-reported beliefs or behaviors. According to Neuman (2000: 247),
“they are strongest when the answers people give to questions measure variables.”

Every researcher collects data using one or more techniques. The main techniques
may be grouped into two categories: quantitative — collecting data in the form of mostly
numbers - and qualitative, collecting data in the form of words or pictures. Some
techniques are more effective when addressing specific types of questions or dealing
with specific kinds of respondents. It takes skill, practice, and creativity to match a
research issue to an appropriate data collection technique. The discussions that fo llow
will mke clearer the rationale for choosing the data collection method mentioned
hereunder.

Data for this dissertation was obtained fiom police responses to self-administered
confidential questionnaires. Robert Groves, a leading expert on surveys, had remarked
that "surveys produce information that is inherently statistical in nature. Surveys are
quantitative beasts." (see Groves, 1996: 389; see also Neuman, 2000). Surveys produce
predictive data and have been extensively used in criminal justice. They are a powerful
tools for obtaining quantitative data for both descriptive, inferential studies and for
addressing the issue of causality and, as briefly mentioned above, are the most widely
used data-gathering technique in the social science. As briefly mentioned above, they
are appropriate for research questions about self-reported beliefs or behaviors. They are
strongest when the answers people give to questions measure variables. A survey may
therefore be defined as a method of gathering information from a number of
individuals, a 'sample', in order to learn something about the larger population fiom

which the sample is drawn. Survey can be open or closed-ended. Each form has

28

advantages and disadvantages. The crucial issue is not which form is best. Rather, it is
under what conditions a form is most appropriate.

The closed-ended survey research method of data collection has its advantages
and disadvantages. One important attractive feature of a closed-ended survey is that it is
easier and quicker for respondents to answer because it asks the questions and also
gives the respondent fixed responses fiom which to choose. The answers of different
respondents are easier to compare, code and statistically analyze. The response choices
can clarify the meaning of a question for respondents and there are fewer irrelevant or
confused answers to questions thus making it easier for the researcher to undertake
meaningful data analysis. Respondents are more likely to answer about sensitive topics
and there are fewer irrelevant or confused answers to questions. Further, collecting
data through the use of a closed-ended survey does not put less articulate or less literate
respondents at a disadvantage as other methods could (Neuman, 2000). It might also be
added here that replication is easier in studies that collect data by closed-ended survey
(Neuman, 2000).

On the other hand, despite the above advantages of data collection strategy by
using the closed-ended survey method, the procedure has some disadvantages. Since
the questionnaire is closed-ended, ideas can be suggested that the respondent would not
otherwise have. Respondents with no opinion or no knowledge can answer anyway.
Respondents can be frustrated because their desired answer is not a choice and it is
confirsing if many response choices are offered. Misinterpretation of a question can go
unnoticed, and they may force respondents to give simplistic responses to complex

issues. Thus, surveys tend to be superficial because the researcher is unable to probe for

29

firrther understanding in relation to the respondent’s actual perception.

The questionnaires

The quantitative data for this study were obtained from police responses to self-
administered, confidential questionnaires. The questionnaires contained hypothetical
offense scenarios or vignettes. This approach has been used in many studies of police
decision making. Nees (1986), for example used a series of 20 vignettes covering a
series of mid-level offenses in his study of police arrest discretion in Boulder County,
Colorado. Meyer (1981), in her study of police arrest decisions, also utilized a series of
vignettes similar to those used by Nees (1986).

Vignettes were selected as a valid method for obtaining original empirical data to
determine how police respond, or at least say they will respond to a variety of people
and situations. The use of hypothetical cases permits the systematic manipulation of the
relevant variables. As stated by Alexander and Becker (1978: 93):

". . .such a stimulus would more closely approximate a real-life decision-
making or judgement-making situation. Furthermore, by holding the
stimulus constant over a heterogeneous respondent population, the survey
researcher gains a degree of uniformity and control over the stimulus
situation . .. [T]he vignette technique makes possible an analysis of the

effects on people's judgements systematically varying the characteristics
used on the situation description"

It should be pointed out here however that vignettes may sometimes be inflexible and
respondents may sometimes be biased, or at times outright lie. Another major problem
with vignettes is that responses may differ from how officers will respond when
actually faced with similar situations in real life.

Each of the questionnaires used in this study was prefaced with a letter thanking

30

the emcers for participating. Each questionnaire contained four hypothetical offense
scenarios with four different types of mid level offenses (public drunkenness,
shoplifting, assault, and traffic offense). All the offense episodes contain low visibility
incidents and were selected because they lend themselves to the broadest range of
police discretion (Goldstein, 1993; Bittner, 1970). In the shoplifting and traffic
scenarios, the suspects displayed a cooperative (positive) demeanor while in the
drunkenness and assault scenarios, they displayed a hostile (negative) demeanor - see
Figure ] (Appendix D). This means that suspect demeanor was constant in each of the
four offense scenarios as described immediately above. The race and gender of the
suspect(8) in each of the scenarios in the “base” questionnaire (questionnaire one) are as
follows: in the public drunkenness scenario the suspects are black female; in the
shoplifting scenario, the suspect is a white male; in the assault scenario, the suspect is a
black male while the suspect in the traffic offense scenario is a white female.

By changing or manipulating the race and gender of the suspects in each of the
four types of crimes, three “other types" of questionnaires containing identical facts and
circumstances but with changes only in the race and/or gender of the suspects are
obtained - see Figure 1 (Appendix D). Thus the total number of hypothetical offense
scenarios was 16. However, since each police officer in the sample responded to four
different hypothetical cases, the total number of responses analyzed were four times as
many as the number of responding officers. For example, as indicated in chart one, the
suspects in drunkenness, shoplifting, assault, and traffic scenarios in questionnaire
one are black female, white male, black male, and white female respectively. In

questionnaire two, everything is the same as in questionnaire one except that the

31

respective suspects in the offense scenarios as mentioned immediately above are new
white female, black male, white male, and black female respectively (see chart two).
The numbers and types of questionnaires completed and returned by police respondents
are shown in Table 14 (Appendix D).

As described above, a total of four types of questionnaires containing 16
"different" hypothetical offense episodes were used in this study. Figure 1 (Appendix
D) shows the complete distribution of hypothetical offense episodes, suspect demeanor,
race, and gender across the four types of questionnaires (one through four) that were
used for this study. Four response choices were provided at the end of each vignette and
officers were expected to respond by selecting only one answer from the four answers
provided. The response choices consisted of: ]= take no action; 2= question the suspect
and release; 3= issue misdemeanor summons or citation; 4= arrest. Officers' responses
to these choices constitute the dependent variable.

Further, each of the offense scenarios had another question in which the officer
was asked to indicate how confident the officer was of taking the action he/she
indicated under the offense scenario. Four response choices were provided at the end of
each of these confidence level questions: (1) very confident; (2) confident; (3) not sure;
(4) not confident; (5) not at all confident.

Following the four vignettes were ten questions designed to collect bio graphic
and other background information about the officers. The questions asked for
information on sex, race, age, marital status, rank, number of years in police work, level
of education, promotional potential, and number of traffic and non-traffic citations

made during the previous year. A complete copy of one type of questionnaire (type

32

one) is exhibited at the end as "Appendix A"

Each officer received only one type of questionnaire. Thus, the particular type of
questionnaire an officer received was determined by random assignment. However, the
questionnaires were arranged in such a way that equal numbers of each type of

questionnaire were handed out to the ofiicers.

The sample

The police department where the data for this study was collected is one of the
largest police departments in a Midwestern state. It is comprised of approximately 369
employees: 268 sworn ofiicers and 10] civilians. The department serves a community
of about 119,000 residents covering an area of about 33 square miles. Out of the 268
sworn officers, it was estimated that approximately 230 are patrol and field officers
who come into frequent contact with the public in the course of their routine duties.
Only full-time sworn field and patrol officers were targeted for this study. This means
that about 230 officers were targeted. Administrative staff were excluded because they
rarely carry out a significant number of arrests. They do not come into fi'equent contact
with citizens. In addition, civilian employees were also excluded for obvious reasons.

Police departments have a record of cooperating with researchers — at least those
who have used similar research methods employed by this study (survey research). For
example, both Thoresson-Rogers (1991), and Meyer (1981), yielded response rates of
more than 85 percent. In a 1994 study, this researcher had a return rate of 78 percent

(Ekwuaju, 1994). Due to the high degree of cooperation that has been received so far

33

fi'om this police department, added to all the measures and safeguards that have been
incorporated in the study, it was expected that enough responses that would be adequate
for a good analysis could be obtained. Approximately 250 questionnaires were
produced and this researcher met with the police officers just before shift roll calls.
Arrangements were made in such a way that this researcher would meet the officers
over a period of several days in December of 2002 on days that many of them were
scheduled for duty. The range of days enabled this researcher to meet with as many of
the police officers as possible. This researcher met with the officers during call sessions
at the beginning of each of each shift, briefly addressed them (emphasizing the
voluntary and confidential nature of the research), handed out the questionnaires to
those that agreed to take them, and collected them after completion. Of the 250
questionnaires distributed, about 132 were completed and returned. Each of the
questionnaires took an average of six minutes to complete. The number of police
officers who participated in this study and the number of types of questionnaire

completed are shown in Table 15 (Appendix D).

Research question

The research question addresses the effect of gender of the suspect on police
arrest decisions in non-felony offenses. It explores whether police process females
chivalrously or more punitively. Additionally, the effects of other independent
variables on police arrest decisions are also explored. These independent variables

include gender of police officer, race of the suspect and race of the police oflicer.

34

Specifically, the four hypotheses listed below are tested in this study:

Research hypothesis

A general hypothesis can be derived fiom the chivalry and labeling perspectives
discussed earlier regarding police treatment of males and females at arrest: Police
officers will arrest male suspects differently than female suspects for less serious non-

felony offenses (i.e., drunkenness, shoplifting, assault, and traffic offenses).

Null hypotheses

H0 1. There is no relationship between the arrest decisions of police officers in
misdemeanor offenses and the gender of suspects.

H0 2. There is no relationship between the arrest decisions of police officers in
misdemeanor offenses and the gender of police ofiicers.

H0 3. There is no relationship between the arrest decisions of police officers in
misdemeanor offenses and the race of the suspects.

H0 4. There is no relationship between the arrest decisions of police officers in

misdemeanor offenses and the race of police officers.

Variables
Oflicers’ responses to the four response alternatives provided to the four

hypothetical offense scenarios constitute the dependent variable (please refer to the

35

section titled “The Questionnaire” for a full discussion of the response options). This is
the way in which officers responded, or at least say they would respond to suspects.
Independent variables, chosen in view of the literature review consist of: type of
offense (public drunkenness, shoplifting, assault, and traffic offense), gender of suspect
(male, female), race of suspect (black/African American, white/Caucasian), and
demeanor of suspect (positive or civil, and negative or hostile). A complete list of
variables used including officer characteristics and their descriptions are annexed at the

end of this study and marked Appendix C.

Questionnaire coordination and control

The Chief of the department where data were collected gave his approval to
officially cooperate with this researcher. The chief was told in a letter tint: (l) the
department would not be identified by name in the study; (2) participation by all
officers was strictly voluntary; (3) responses by the officers would be kept anonymous;
and (4) the University Committee for Research Involving Human Subjects (UCRIHS)
would approve the questionnaire before the actual collection of data. The Chief gave
his approval subject to these conditions as itemized above. Subsequent to the above
approval, this researcher worked with one of his two Assistant Chiefs (Assistant Chief
of Field Services) to finalize the arrangements and other logistics for distribution and
collection of the questionnaires. It was agreed that this researcher would attend
“briefings” at the beginning of shifts. The department has three shifts each day and it

was planned in such a way that this researcher would attend all shifts in all the precincts

36

of the police department. The data were collected over a period of several days in
December of 2002. Table 15 (Appendix D) shows the number of officers and the types

of questionnaires completed for this study.

37

CHAPTER FOUR

ANALYSIS AND RESULTS

Demographic characteristic of police respondents

Table 1 shows the demographic characteristics of the police officers who
responded to the questionnaires. As indicated in that table, the majority of the
respondents are white (72.9%), male (86.8%), and married (59.7%). While the majority
of them (48.8%) have a 2-year degree, many of them (47.3%), have a four-year college
degree. The Afiican American and Hispanic composition of the police respondents are
10.9% and 9.3% respectively. Table 1 also shows tlnt most of them are patrol ofiiCers
(82.9%) and have worked less than ten years in police work as sworn police officers
(70.3% - mean experience in years = 8.78; std. dev. = 6.40). The majority of them
(48.4%) think they have an “average” promotional potential while 43.0% think they
have a “very good” promotional potential. Only 8.6% of them think they have “below
average” promotional potential. The age of the respondents ranges fiom 21 to 48 years,
with a mean age of 32.87 years (std. deviation = 6.86), and the majority (40.8%) are
between 31 and 40 years of age.

Lastly, Table 1 also indicates tint the majority of the police respondents issue
between 10 — 20 monthly traffic citations and the range of non-traffic citations every
month. The mean number of monthly traffic citations issued by these officers is 2.39
with a standard deviation of 1.16, while the mean number of monthly non-traffic

citations is 1.57 with a standard deviation of .58.

38

Figure 2 (Appendix D), shows the race and gender composition of all sworn
“police officers” of the department where the data for this study was collected. A brief
comparison with officers who responded to the questionnaires indicates that only about
20 percent of the department’s sworn officers failed to participate in this study. Not all
the demographic infernntion that was included in Table 1 was available in respect of
the whole department. The department stated that it would take nnny hours of manual
search to retrieve such information. Due to this time and labor constraints, the

department was unwilling to undertake such a manual search of their records.

39

Table 1: Characteristics of Police Respondents

 

 

Number Valid Number Valid
Percent Percent
Gender Rank
Male 1 12 86.8 Recruit 5 3.9
Female fl L2 Patl. Off. 107 82.9
129 100% Master PD 2 1.6
Sergeant 10 7.8
Above Sergeant 3 2.3
Ag; Other rank __2 1.6
21 to 30 47 37.6 129 100%
31 to 40 51 40.8
41 to 48 27 21.6
48 + L QQ_.Q Yrs. of exp.
126 100% Less 10 yrs. 90 70.3
10 to 20 30 23.4
Mean age (in years) = 32.8 21 to 25 _8 A;
Standard deviation = 6.86 128 100%
Mean experience. (in yrs) = 8.78
M2 Standard deviation = 6.40
White 94 72.9
Black 14 10.9
Hispanic 12 9.3 Education
Other _9_ _Z.Q_ H. sch. grad 1 0.8
129 100% 2 year degree 63 48.8
4 year degree 61 47.3
Grad. degree _;4_ A
Marital Status 129 100%
Married 77 59.7
Single 41 31.8
Separated 2 1.6 Monthly traffic citations
Divorced 9 7.0 Less than 10 28 21.9
Widowed _0 0.0 10 - 20 53 41.4
129 100% 21 ~30 26 20.3
31 - 40 10 7.8
41+ _11_ L6
Emmpfipnpl Pptentigl 128 100%
Very good 55 43.0
Average 62 48.4 Mean #, monthly traffic citations = 2.39
Below avg. A _LQ Standard deviation = 1.16
128 100%

40

Table 1 (cont’d).

 

Number Valid
Percent.
Monthfl non-trafiic citations
Less than 10 60 47.2
10 — 20 61 48.0
21 - 30 _6_ i7
127 100%

Mean, number of monthly non-traffic citations = 1.57

Std. dev. = .58

 

Missing cases excluded.

4l

Summary of responses of the emcers

Table 2 summarizes the responses of the officers to the public drunkenness
scenarios across the four types of questionnaires used in this study. Table 2 shows that
five officers (3.9% of all the responses to this scenario) said they would “take no
action” to the suspects in response to the inforrrntion provided in the public
drunkenness hypothetical episode. Of these five responses, three were to the
questionnaire with black female suspects while the other two were to black male
suspects. Twenty three officers (18.1%) indicated they would “question and release”
the suspects (eight in response to black female suspects; 12 to white female suspects; 1
to black male suspects, and the remaining two to white male suspects). Thirteen officers
(10.2%) responded that tiny would “issue misdemeanor citations.” Of these 13
responses, five were to black female suspects; two to white female suspects; four to
black male suspects and the other two were to white male suspects. The last and most
serious response option in the questionnaire was the arrest option. In the public
drunkenness scenarios under consideration, 86 officers (67.7%) said they would
“arrest” the suspects. Of these 86 arrest responses, 18 of them were to black female
suspects; 16 were to white female suspects; 27 were to black male suspects, while the
remaining 25 were to white nnle suspects. It is noted briefly that the most serious
disposition here (27 N or 21.3%) of the “arrest” responses were returned to the
questionnaire with black male suspects — this is however, not a “finding” of this study.

Table 3 summarizes the officers’ responses to all the shoplifting offense scenarios
across the four types of questionnaires used in this study. Table 3 shows that four

officers (3.1%) indicated that they would “take no action” toward the suspects in

42

 

response to the information provided in the shoplifting hypothetical episodes. Of these
four responses, one was to white male suspects; one to black male suspects while the
remaining two were to white female suspects. Seven officers (5.3%), indicated they
would “question and release” the suspects - (five in response to black rrnle suspects and
the remaining two to black female suspects). Three officers (2.3%) said they would
“issue misdemeanor citations” to the information provided in the shoplifting scenarios
under consideration. The three responses were to the questionnaire with black male
suspects. Table 3 also shows tint 117 officers (89.3%) said they would “arrest” the
suspects. The 117 responses were distributed as follows: 33 to white male suspects; 24
to black male suspects; 31 to white female suspects while the remaining 29 were to
black female suspects. Again, it is noted briefly that the most serious disposition here
(33 N or 25.2%) of “arrest” responses were returned to the questionnaire with white
male suspects - this is however, not a “finding” of this study.

The responses of the officers in the assault scenario are particularly interesting.
This scenario received the least amount of variability in the officers’ responses. As
Table 4 shows, six oflicers (4.6%) indicated they would “take no action” toward the
suspects. Of these six responses, one was to black male suspects; three to white male
suspects, while the rennining two were to black female suspects. Only one officer said
tint he or she would “question and release” the suspect in the assault scenario under
consideration. This response was to white male suspect. As shown in Table 4, none of
the officers indicated he or she would “issue a misdemeanor citation” to the suspects in
this scenario. However, one hundred and twenty three officers or 94.6% of all

respondents to this scenario indicated they would “arrest” the suspect. Of this number,

43

33 were to black male suspects; 28 to white male suspects; 31 to black female suspects
while the remaining 31 were to white female suspects. It is noted briefly that the most
serious disposition here (33 N or 25.4%) of “arrest” responses were returned to the
questionnaire with black male suspects — again, this is not a “finding” of this study.

In the traffic offense scenarios, Table 5 shows tint six officers (4.6%) indicated
they would “take no action” toward the suspects in response to the inforrrntion
provided in this scenario. Of the six responses, one was to a white female suspect; one
to a black female suspect; two to white male suspects while the remaining two were to
black male suspects. Seventeen officers (13.1%) said they would “question and
release” the suspects. The 17 responses were distributed as follows: 10 to black female
suspects while the remaining seven were to black male suspects. One hundred and six
officers (81.5%) said they would “issue misdemeanor citations.” Of these 106
responses, 32 were to white female suspects; 20 to black female suspects; 32 to white
male suspects while the remaining 22 were to black male suspects. Only one officer
indicated that he or she would “arrest” the suspect in this traffic offense scernrio under

consideration. That lone response was to black female suspects.

44

Table 2: Oficers’ responses to the four public drunkenness scenarios by suspect

race and gender (suspect demeanor is constant - negative).

 

 

OFFICERS’
RESPONSES SUSPECTS TOTAL
Black White Black White
Female Female Male Male
Take no
action 3 (2.4%)* - 2 (1.6%)* - 5 (3.9%)*
Question
and let go 8 (6.3%) 12 (9.4%) l (.8%) 2 (1.6%) 23 (18.1%)
Issue misd.
citation 5 (3.9%) 2 (1.6%) 4 (3.1%) 2 (1.6%) 13 ( 10.2%)

Arrest
suspect 18 (14.2%) 16 (12.6%)

TOTAL 34 (26.8%) 30 (23.6%)

27 (21.3%)

34 (26.8%)

25 (19.7%) 86 (67.7%)

29 (22.8%) 127 (100%)

 

* = Percentage of total responses to all four public drunkenness scenarios

45

Table 3: Oflicers’ responses to the four shoplifting scenarios by suspect race and

gender (suspect demeanor is constant - positive).

 

 

OFFICERS’
RESPONSES SUSPECTS TOTAL
White Black White Black
Male Male Fennle Female
Take no
action 1 (.8%)* 1 (8%)" 2 (1.5%)* - 4 (3.1%)*
Question
and let go - 5 (3.8%) 2 (1.5%) 7 (5.3%)
Issue misd.
citation - 3 (2.3%) - - 3 (2.3%)
Arrest
suspect 33 (25.2%) 24 (18.3%) 31 (23.7%) 29 (22.1%) 117 (89.3%)
TOTAL 34 (26.0%) 33 (25.2%) 33 (25.2%) 31 (23.7) 131 (100%)

 

* = Percentage of total responses to all four shoplifting scenarios

Table 4: Officers’ responses to the four assault scenarios by suspect race and

gender (suspect demeanor is constant — negative).

 

 

OFFICERS’
RESPONSES SUSPECTS TOTAL
Black White Black White
Male Male Female Fennle
Take no
Action 1 (.8%)* 3 (2.3%)* 2 (1.5%)* - 6 (4.6%)*
Question
and let go - 1 (.8%) - - 1 (.8%)
Issue misd.
citation - - - - -
Arrest
suspect 33 (25.4%) 28 (21.5%) 31 (23.8%) 31 (23.8%) 123 (94.6%)

TOTAL 34 (26.2%)

32 (24.6%)

33 (25.4%)

31 (23.8%) 130 (100%)

 

* = Percentage of total responses to all four assault scenarios

47

Table 5: Officers’ responses to the four traffic ofl'ense scenarios by suspect race

and gender (suspect demeanor is constant - positive).

 

 

OFFICERS’
RESPONSES SUSPECTS TOTAL
White Black White Black
Female Female Male Male
Take no
action 1 (.8%)"‘ 1 (.8%)* 2 (1.5%)* 2 (1.5%)* 6 (4.6%)*
Question
and let go - 10 (7.7%) - 7 (5.4%) 17 (13.1%)
Issue misd.
citation 32 (24.6%) 20 (15.4%) 32 (24.6%) 22 (16.9%) 106 (81.5%)
Arrest
suspect - 1 (.8%) - - 1 (.8%)
TOTAL 33 (25.4%) 32 (24.6%) 34 (26.2%) 31 (23.8%) 130 (100%)

 

* = Percentage of total responses to all four traffic offense scenarios.

48

Preliminary findings from a summary of officers’ responses to all ofl'ense episodes

Table 6 summarizes the distribution of officers’ responses to all the offense
scenarios in the four types of questionnaires. As indicated in that table, the public
drunkenness, shoplifting, and traffic offense scenarios each contain responses from all
the four response options: take no action; question and release the suspects; issue
misdemeanor citation; and arrest suspect. However, the traflic offense scenario has the
least percentage of respondents (.8%) who indicated they would take the most serious
action (arrest). This is in contrast to the other two scenarios (public drunkenness and
shoplifting) mentioned immediately above where the percentage of officers who
indicated they would arrest the suspects ranges from 67.7% to 94.6%.

The responses to the assault scenario raise some interesting issues. They differ
radically from the other three scenarios considered above in several unique respects.
The assault scenario contains the least amount of variation across the four response
options provided to the officers. The assault scenario is the only one that does not
contain responses from all the response options. The responses are not only “tightly”
distributed but they tilt heavily toward “arrest.” In fact, Table 6 indicates tint the
assault scenario has the highest percentage of officers (94.6%) who indicated they
would take the most serious action - “arrest” the suspects. Can this uniqueness be
linked to the nature of the offense and to the negative or hostile demeanor portrayed by
the suspects? It is difficult, and may involve some degree of speculation to offer a
widely acceptable ranking of the four offenses in order of seriousness. This notwith-
standing, assault seems the most serious of the four offenses. In the questionrnire, this

scenario was presented

49

Table 6: Summary of the distribution of ofiicers’ responses to all ofl'enses as

contained in all four types of questionnaires (suspect demeanor is constant).

 

 

RESPONSES TYPES OF OFFENSE
Drimkenness Shoplifting Assault Traffic
- demeanor + demeanor - demeanor + demeanor
% N % N % N %
Take no
action 3.9 4 3.1 6 4.6 6 4.6
Question
& let go 18.1 7 5.3 1 .8 17 13.]
Issue misd.
citations 10.2 3 2.3 - - 106 81.5
Arrest
suspect 67.7 1 1 89.3 12 94.6 1 _§
Total 100 131 100 130 100 130 100

 

50

Table 7: Oflicers’ levels of confidence in their decision-making — across all four

 

 

types of offenses
Levels of
Confidence Types of Offenses

Drunkenness Shoplifting Assault Trafiic

- demeanor + demeanor - demeanor + demeanor

N % N % N % N %

Very Conf. 76 59.4 97 73.5 118 89.4 98 74.8
Confident 50 39.1 35 26.5 14 10.6 31 23.7
Not Sure 2 1.6 - - - - 2 1.5
Not Conf. - - - - - - - -
NAA‘ Conf. _'_ _'__ ;_ _'_ _:__ _:__ _"_ _"
Total: 128 100 132 100 132 100 131 100

 

* NAA = Not at all confident

51

as involving some measure of bodily harm — an occurrence the average citizen and
indeed a typical police ofiicer might be expected to consider more serious than public
drunkenness, a minor traffic offense, or shoplifting. The assault scenario is the only
scenario with an identifiable “victim.” In addition, it may be observed that while the
combined percentages of arrest responses to the two offenses with a positive
demeanor, (that is, shoplifting and traffic offense) is 90.1% - a percentage which is
less than 100%; tint of the other two offenses with a negative demeanor (public
drunkenness and assault) is 162.3% - a figure which is greater than 100%.
Furthermore, one of the two offense scenarios with a positive demeanor is the traffic
episode. That scenario is also unique in having the least percentage of respondents
(.8%) who indicated that they would take the most serious action — arrest the suspects.
It therefore seems to me that these patterns of arrest responses suggest that a suspect’s
demeanor and the type of offense both are relevant variables in police arrest decisions.
When the offense is “serious” and the suspect’s demeanor is negative, police officers
are more likely to arrest than when the offense is relatively less serious and the
suspect’s demeanor is positive.

An examination of the respondents’ levels of confidence in making the arrest
decisions they indicated appears to strengthen the above argument. As described earlier,
each of the offense scenarios had another question in which the officer was asked to
indicate how confident he/she was in taking the action he or she had indicated under the
offense scenario. Table 7 summarizes the results of the officers’ levels of confidence in
all four offense scenarios. While a key “finding” from the table is that there is very little

lack of confidence in the officers’ arrest decisions, it is also seen that assault ind the

52

greatest percentage of officers (89.4%) with the highest level of confidence — “very
confident” in their arrest decisions.

The major finding fi'om the above analysis so far suggests that suspect demeanor
and the type of offense are relevant variables in police arrest decisions. When the
ofl'ense is “serious” and suspect demeanor is negative, almost every police officer
(94.6%) will arrest the suspect and the officer has the highest percentage of confidence
(89.4%) or feels “very confident” in nnking the arrest. From here, it is intended to
embark into firrther inquiries, this time utilizing more advanced statistical procedures to

test the specific hypotheses outlined earlier in this study.

Gender difl'erences in police arrest decisions

Impact of gender of the suspect
To analyze police responses to the questionnaires for gender differences in

police processing of suspects at arrest, a series of independent sample t-tests comparing
police arrest decisions on male and female suspects were executed for each of the four
offense scenarios. Police arrest decisions on all rrnles were also compared to their
decisions on all females. The effect of race was controlled by comparing white males to
white females and black nnles to black females. Suspect demeanor was held constant
in each of the four offenses as a result of the manner in which the scenarios were
depicted in the questionnaires. Type of offense was controlled or held constant by
nmning the test for each offense separately.

Although a simple comparison of means of police arrest decisions on male and

female suspects is easy, one is faced with an important question here. Does the

53

discrepancy in the means reflect a firndamental distinction or difference in the two
groups from which the means were computed, or can the difference be explained in
terms of chance? T-tests are very appropriate for comparing sample means of two
groups for statistical significance. T-tests tell us whether the two sample means reflect
two different populations or are drawn from the same population. Table 8 shows gender
differences in police arrest decisions controlling for type of offense, race and suspect
demeanor. As indicated, the means for the two groups (male and female suspects) are
numerically different. F or example, Table 8 indicates that for the public drunkenness
scenario, the mean of police arrest decisions for males is 3.71 while that of females is
3.12. However, the difference only indicates one of two possibilities. The higher mean
(nnles), may indicate that nnles are treated more harshly (arrested more) than females
(the rationale for this possible conclusion will be explained later in this section). If this
is the case, the discrepancy in the means reflects a real difference in police treatment at
arrest. On the other hand, the higher mean may simply be a chance happening in which
case the difference in the means is considered unimportant or trivial. To help decide
whether there is a real or statistically meaningfirl difference between the differences in
mean treatment of males and females, this researcher turns to the statistics of Matthew's
testing. In hypothesis testing, the null hypothesis (H0) is usually tint an observed
difference between two means is trivial. The rival hypothesis (H1) is tint the observed
difference is real. Thus null hypothesis one (H0 1) in this study is tint there is no
difference between the means of police arrest decisions for males and females. The

t-test significance test determines the probability that H0 is true. If the probability is

54

small, say 5 percent or less, H0 will be rejected in favor of H1 (rival hypothesis). This
is accomplished with Student’s Independent Samples t-tests.

The common alpha level set for statistical significance for such tests is usually .05
(Babbie, 1983: 417-426). According to Lurigio, Dantzker, Seng and Sirncore (1997:
124), “in hypothesis testing, you must select a chance probability for rejecting H0.

Although a number of such probabilities can and should be chosen, most statisticians

.11

select 5 percent (Cohen, 1990). This is called the level of significance. ” If the resulting

significance level is small enough (in this case .05 or less), the hypothesis tint the

WEI.‘W“—'.¢Cu.

means of the two populations samples are equal is rejected — which means that the null
hypothesis being tested will be rejected (Norusis, 1988).
Tests of significance are conducted as either one- or two-tailed tests. A one-tailed

test uses a cutoff score that is located in either the upper or the lower tail of the
t distribution. Two-tailed tests use cutoff scores that are located in both the upper tail
and the lower tail. The basic difference between a one-tailed and a two-tailed test is
illustrated by this example. Assume tint you conduct a one-tailed Most in which you
expect that group A performs better than group B. Given that you are conducting a one-
tailed test, you can reject H0 only if group A performs significantly better than group
B. However, if you conduct a two-tailed test, you can reject H0 if group A performs
significantly better tlnn group B or if group B performs significantly better than
group A. Given the above, Lurigio, et al. (1997:222) advised:

“We recommend two-tailed tests of significance even if there is a strong

expectation about the results. A one-tailed test should be conducted only in

rare situations in which results are expected to be in one direction and the

investigator does not want to or is not ethically bound to report results
opposite the expectation.”

55

According to Norusis (1993: 256), “a two-tailed test is used to detect a difference in

means between two populations regardless of the direction of the difference.” Against

the background of the foregoing, all the t-test procedures here involve two-tailed tests.
According to Norusis (1993), “the minimum specifications are:

0 One or more numeric test variables.
0 One numeric or short string grouping variable.
0 Group values for the grouping variable.”

The tests in this study meet the above specifications. A four-option response was

provided to the officers and a four-point scale was used to code the responses of the i ,

 

officers to the offense scernrios (See Appendices A and B for copies of the
questionnaire and coding guide respectively). The responses were ranked in order of
severity starting with the most lenient treatment (take no action) to the most severe
treatment (arrest). The responses consisted of: l = take no action; 2 = question and
release; 3 = issue misdemeanor citation; and 4 = arrest. The rrnxirnum mean police
response is therefore 4.00, indicating “arrest.” The minimum mean is 1.00 indicating
“take no action.” From the minimum mean of 1.00 (i.e., take no action), the higher the
mean, the more severe the police response, until the most severe police response which
is arrest (4.00).

Given the above, to test the four specific null hypotheses, a series of independent
sample t-tests were conducted. Null hypothesis one (H0 1) states that there is no
relationship between the arrest decisions of police officers and gender of suspects in
low-level offenses. Police arrest decisions on male suspects were compared to their
decisions on female suspects. The effect of race was controlled by also comparing

white males to white females and black males to black females. Suspect demeanor was

56

held constant in each of the four offense scenarios as a result of the manner in which
the scenarios were depicted in the questionnaires. Table 8 shows the results of gender
differences in police arrest decisions (See also Table 16, Appendix D for a detailed

SPSS Group Statistics printout).

57

Table 8: Gender differences in police arrest decisions controlling for type of

ofl’ense, and suspect race and demeanor

Whites Blacks All All
Males Pennies t. Males Females t. Males Pennies t.

 

Drunkenness
xresponse 3.79 3.13 -3.17* 3.64 3.11 -2.30* 3.71 3.12 -3.79*

Shoplrftmg
xresponse 3.91 3.81 .61 3.51 3.87 -1.98" 3.71 3.84 -1.06

Assault
xresponse 3.65 4.00 -2.04* 3.91 3.81 .61 3.78 3.90 -1.03

Traffic
xresponse 2.88 2.93 .55 2.64 2.65 .07 2.76 2.80 .33

 

* = p less than .05

58

As indicated in Table 8, in the public drunkenness scenario, males (mean
treatment = 3.71), when compared to females (mean treatment = 3.12), are treated
slightly more harshly by police. The same result holds when race is introduced as a
control variable and white males (mean treatment = 3.79) are compared to white
females (mean treatment = 3.13); and black nnles (mean treatment = 3.64) are
compared to black females (mean treatment = 3.11). A look at SPSS “Group Statistics”
- Table 16 (Appendix D) indicates that the standard deviations (tint is, variability
within groups), are .7055 for nnles and 1.0157 for ferrnles (.5593 and .9732 for white
males and white females respectively; .8121 and 1.0664 for black nnles and black
females respectively). This suggests tint there is less variability in police treatment of
males than females with or without race as a control variable. Police officers use their
discretion less for both white and black nnles when compared to their opposite sex. A
relevant question here is whether these differences in treatment between (not within)
the groups are significant or obtained by chance? Table 8 shows that the 1 values for
males, when compared to females, is -3.79; white males when compared to white
females = -3.17; and -2.30 for black males compared to black females. All three t
values are statistically significant (p = less than .05). All three t values are equal or
greater than +/- 1.96, which is necessary for the p value to reach statistical significance
at 95% confidence interval. This means in effect tint the police respondents treated
males slightly more harshly when compared to females irrespective of race. There is
therefore support for null hypothesis one (H0 1) to be rejected for the public

drunkenness offense.

59

In the shoplifting scenario, white males are treated a little bit harsher than white
ferrnles. However, the level of treatment here is slightly more severe (as indicated by
higher mean) than in the public drunkenness scenario considered above. This difference
in treatment is however not statistically significant. Table 8 also indicates that black
males are treated slightly more leniently than black females and as indicated in Table 8,
this difference in treatment is significant at .05 level When all males are compared to
all females, the males are treated slightly more leniently than females. However, the
difference in treatment is trivial and not statistically significant. Ho 1 is therefore not
rejected in the shoplifting offense because the difference in treatment between all males
and all females is not statistically significant.

In the assault scenario, white nnles are treated less leniemly than white females,
all of whom received the most severe treatment (arrest), as indicated by the most severe
mean score for white females - 4.000 (no variability within group treatment). This
difference in treatment is statistically significant at .05. Black nnles are treated slightly
harsher than black females but the difference is trivial. When all males are compared to
all females in the assault offense scenario under discussion, all males are treated
slightly more leniently than all females but the difference in treatment is trivial. Ho 1 is
therefore not rejected in the assault scenario.

In the traffic offense scenario, the nnles are consistently treated more leniently
with or without race as a control variable. However, none of these differences in police
processing at arrest was significant at .05 significance level. H0 1 is therefore not

rejected in the traffic offense scenario.

60

To summarize, the gender differences in the analyses of police arrest decisions,
results indicate that except in public drunkenness scenario, the gender of the suspect is

not directly related to police processing of suspects.

Impact of gender of the officer

The second null hypothesis (Ho 2), is concerned with the relationship between
police arrest decisions and the gender of the officer. In this analysis, all four offenses
were first computed into one offense. A comparison was then performed between the
mean arrest scores of male officers and those of female officers. As discussed earlier,
the maximum mean response to each offense is 4.00. Since the four offenses inve been
added or computed together, the maximum mean score is now 16.00. The closer the
mean score is to 16.00, the more severe the response of officers to the suspects. As
Table 9 indicates, the mean arrest score of male officers is 13.8868 while that of female
officers is 13.8000. The table also indicates that the standard deviation of male officers
is 2.0299 while tint of female officers is 1.6125. This suggests that there is more
variability in the arrest decisions of male officers when compared to female officers.
Male officers therefore appear to use more discretion than female officers. The t value
for the comparison is .158 while the significance value at the 95% confidence interval
is .874, a figure which is clearly more than .05. There is therefore no support for
rejecting Ho 2. The implication from this analysis is tint the gender of the police officer

is not directly connected with police arrest decisions.

61

Table 9: Independent sample t-test comparing arrest decisions of male and female

officers across all four offenses.

 

 

 

so; N M _n Std. dev. t y 2:
Male 108 13.8868 2.0299

.158 119 .874
Female 15 13.8000 1.6125

 

62

Race differences in police arrest decisions
Impact of race of the suspect

The third null hypothesis (H0 3) is concerned with the relationship between police
arrest decisions and the race of the suspects. To test this hypothesis, a series of
independent t tests comparing police arrest decisions on white and black suspects was
conducted for each of the four offenses. Police decisions for all whites were compared
to their decisions for all blacks for each of the four offenses. To control for the effect
of gender, white males were compared with black males and white females with black
ferrnles. Suspect demeanor was held constant in each offense as a result of the manner
in which the scenario was presented in the questionnaire. The findings are summarized
in Table 10.

As indicated, there are some differences in police treatment of white suspects
when compared to black suspects in the public drunkenness scenario - both with and
without gender as a control variable. When all whites are compared to all blacks, in the
public drunkenness scenario, Table 10 indicates tint whites are treated more harshly.
Table 17 (Appendix D) shows that the standard deviation for whites is .8575 while that
of blacks is .9776. These two figures indicate variability within groups and suggests
that officers use more discretion in processing white suspects than black suspects.
However, all these differences are trivial since none of them is significant at .05
significance level. There is therefore little support in rejecting H0 3 for public
drunkenness both with and without gender as a control variable.

In the shoplifting scenario, white suspects were treated slightly more harshly

when compared with black suspects. However, when gender is introduced as a control

63

variable, two contrary trends are observed. While police response to white females is
less severe than the response to black females (3.8] vs. 3. 87), the response to white
nnles is more severe than the response to black males (3.9] vs. 3.51). However, only
the difference in police response between white and black nnles is statistically
significant at the .05 significance level. The other differences are not statistically
significant. HO 3 is therefore not rejected in the shoplifting scenario.

In the assault scenario, white suspects were treated slightly more leniently when
compared with black suspects (3.82 vs. 3.86). However, when gender is introduced as a
control variable, again as in the shoplifting scenario, two contrary trends are observed.
While police response to white ferrnles is more severe than the response to black
females (4.00 vs. 3.81), the response to white males is less severe than the response to
black males (3.65 vs. 3.91). However, these differences in treatment are trivial and
none is statistically significant at the .05 significance level. H0 3 is therefore not
rejected in the assault scenario.

In the traffic scenario, whites consistently received more severe dispositions than
blacks both with and without gender as a control variable. The differences in treatment
are statistically significant except in the treatment between white and black male
suspects. Ho 3 is therefore rejected for the traffic offense variable.

Summarizing the findings on race differences in police arrest decisions, Table 10
indicates that except as found in traffic offenses, race of the suspect is not a relevant
variable. Overall therefore, race of the suspect is not directly related to police arrest

decisions.

64

Table 10: Race difl’erences in police arrest decisions controlling for type of offense

and suspect gender and demeanor

Males Females All All
White Black t. White Black t. Whites Black t.

Drunkenness
xresponse 3.79 3.64 -.81 3.13 3.11 -.06 3.45 3.38 -.45

Shophftmg
xresponse 3.91 3.51 2.27" 3.81 3.87 -.33 3.86 3.68 1.5

Assault
xresponse 3.65 3.91 1.38 4.00 3.81 -1.39 3.82 3.86 .35

Traffic
xresponse 2.88 2.64 1.75 2.93 2.65 233* 2.91 2.65 2.88“

 

* = p. lessthan .05

65

Impact of race of the officer

The fourth null hypothesis (H0 4) is similar to the second null hypothesis
discussed above except tint H0 4 deals with officer race (H0 2 deals with officer
gender). As in H0 2, all four offenses were first computed into a single offense and t-
tests comparing the mean arrest scores of white officers with black officers were
executed. As discussed earlier, the maximum mean response to each offense is 4.00.
Since the four offenses have been added or computed together, the nnxirnum mean
score is now 16.00. The closer the mean score is to 16.00, the more severe the response
ofofficers to the suspects. The results are shown in Table 11. As indicated in that table,
the mean arrest score of white officers is 14.2022 (with a standard deviation of 1.4157),
while that of black officers is 12.6154 (std. dev. = 2.1031). This suggests that compared
to black officers, white officers are likely to make more arrests but process suspects
with less discretion. Table 11 also indicates that the difference between the mean arrest
scores of white and black officers is statistically significant at .05 level. HO 4 is
therefore rejected in this comparison between the differences in arrest treatment by
white and black police officers.

Three other analyses involving the impact of race of the officer in police arrest
decisions were also performed. Table 12 shows t-test comparisons between the mean
arrest scores of white and Hispanic officers; Table 13 exhibits the comparison between
black and Hispanic oflicers, while Table 14 shows the comparison between white
officers and the other minority emcers (blacks, Hispanics and others) combined. There
are some differences in the disposition of the groups but in respect to the first-two

comparisons mentioned above, the differences are trivial. There is therefore little

66

support in rejecting H0 4 in these comparisons - that is differences between white and
Hispanic officers and black and Hispanic officers. A8 briefly mentioned above, Table
14 shows the results of an independent sample t-test between the arrest disposition of
white officers and other races (blacks, Hispanics and others - hereinafter called
“minority officers”). As indicated in that table, the mean arrest score of white officers
is 14.2022 while that of minority officers is 12.9688. The standard deviation (variability
within group) of white officers is 1.4157 while tint of minority officers is 2.8791. This
suggests that white officers use lesser discretion in processing suspects at arrest than
minority officers. Table 14 also indicates tint the comparison between the mean scores
of white officers and minority officers is statistically significant at .05 level of
significance. There is therefore support for H0 4 to be rejected for this comparison (that
is, between white and minority officers).

Summarizing, the implication from the last four arnlyses described above is tint
there are mixed findings as to the impact of the race of the officer in police arrest
decisions. White officers are likely to arrest more than Afi'ican American officers and
minority officers, and this difference is statistically significant. However, when distinct
racial minority groups are introduced as a control variable, results indicate that race of
the officer is not a relevant variable in police arrest decisions. Based on the main
comparison between white and black officers above, there seems to be available
evidence to support the rejection of HO 4. Thus, white officers are likely to effect more

arrests than black officers and minority officers.

67

Table 11: Independent sample t-test comparing arrest decisions of black and

white oflicers across all four oflenses.

 

 

Race N Meg; Stuev. t (LI E
White 89 14.2022 1.4157

3.528 100 .001
Black 13 12.6154 2.1031

 

68

Table 12: Independent sample t-test comparing arrest decisions of white and

Hispanic officers across all four offenses.

 

 

 

gee N M _p Std. dev. t y
White 89 14.2022 1.4157

1 .242 97 .21 7
Hispanic 10 13.5000 3.3747

 

69

Table 13: Independent sample t-test comparing arrest decisions of black and

Hispanic emcers across all four offenses.

 

 

 

Race N M _p Std. dev. t Jd_._f
Black 13 12.6154 2.1031

-.773 21 .448
Hispanic 10 13.5000 3.3747

 

70

Table 14: Independent sample t-test comparing arrest decisions of white and

minority oflicers across all four offenses.

 

 

Race N Mea_ Std. dev. t d.f E
White 89 14.2022 1.4157

3.1 36 1 19 .002
Minority" 32 12.9688 2.8791

 

* Minority = Afi'ican American, Hispanic and Other racial group officers.

7]

Summary of findings

On the basis of this research, there is some justification to suggest that the major
variables in determining how police officers will respond are the nature of the offense
and the manner in which the offender behaves when confionted by the officer. Gender
of the suspect is overall not directly related to police arrest decisions. In effect
therefore, neither the chivalry nor the labeling perspectives as summarized earlier
above was supported by the results of this study. Gender of the oflicer is not directly
related to police arrest decisions. Race of the suspect is overall not directly related to
reported police arrest decisions. Race of the officer appears a relevant variable that
interacts with other variables in police arrest decisions. lastly, results of this research
indicate that police officers have a high degree of confidence in the arrest decisions

they make.

72

CHAPTER FIVE

DISCUSSION AND CONCLUSIONS

As reported earlier, a “preliminary finding” of this study is that when the offense
is “serious” and the suspect demeanor is negative, police officers are more likely to
arrest than when the offense is relatively less serious and the suspect demeanor is
positive. There seems to be unanimity on the impact of the seriousness of the offense
on police arrest decisions. Perhaps offense seriousness may be the most important and
reoccurring variable on police arrest decisions (Sherrrnn, 1980; Riksheim & Chermak,
1993). The first part of this finding (the impact of seriousness of the offense), was
therefore not only expected but also consistent with the literature on police arrest
decisions. However, research findings on the effects of certain individual characteristics
of the suspect are either almost evenly divided or unresolved. Nevertheless, among
individual characteristics, there is much controversy on such variables such as race and
gender. On the other hand, the impact of suspect demeanor, even though it is an
individual characteristic, has received relatively more support than race and gender.
Meyer (1981, 1982); Stone (1985); Visher (1983); Corley et al. (1989); Riksheim &
Chermak (1993); and Sloan (1991) are just a few of the many studies that report similar
finding as to the relevance of demeanor of the suspect on police arrest decisions. The
second part of my finding is therefore also consistent with a majority of the literature.

The nnjor variables in determining how police officers will respond are the nature
of the offense and the manner in which the offender behaves when confronted by the

oflicer. It is generalized that the combined effects of these two variables (that is, type of

73

offense and suspect demeanor) influenced the respondents to decide to arrest 95% of all
the suspects in the assault scenario. As stated earlier, all the suspects in the assault
scenario exhibited a hostile or negative demeanor. As to the nature of the offense, it is
submitted that assault involving bodily harm or injury is more serious and is, in fact,
punished more severely by most criminal statutes than drunkenness, shoplifting, and
traffic offenses. The effect of demeanor is, however, significantly more important for
some kinds of crimes than for others. This explains why drunkenness, in which all the
suspects displayed a hostile demeanor, were not all arrested but still had a higher arrest
percentage than traffic offense in which the suspects displayed a cooperative demeanor.
This suggests that for more serious crimes, suspect demeanor is not very influential.
This position is supported by all prior research reviewed for this study (see Sherman,
1980; Riksheim and Chermak, 1993).

The results of the arnlyses conducted in this study on gender differences on police
arrest decisions indicate tint except in the drunkenness offense, there is no statistically
significant differences between police treatment of males and their treatment of females
in the rest of the three offenses considered — see Table 8. This notwithstanding, there
are still some substantive (though statistically not significant) gender differences in
police decisions worth discussing. Contrary to expectations, females consistently
received harsher dispositions than nnles in shoplifting, assault, and traffic offense
scenarios. This trend is not consistent with the direction of the findings reported by the
majority of studies tint found gender differences in police treatment at arrest. These

studies generally support the chivalry perspective as discussed above under “statement

74

of the problem.” Even though this comparison is not statistically significant, how could
this trend be explained?

It has been argued that agencies actively “sexualize” offenses by reclassifying
specific violations as sex-role or sexual violations (Chesney-Lind, 1973). The tendency
is to punish female behavior more severely when it falls within these categories. The
offenses involved here are shoplifting, assault and traffic offenses and none is an
offense of a sexual nature. Can Chesney-Lind’s argument be extended to cover these
offenses? Families, and indeed society as a whole, have historically found reasons to
generally punish women more severely when sexual violations are involved. Unless a
high leap can be made and justified, it is difficult to rationalize the need for extending
Chesney-Lind’s argument to cover the offenses under discussion here.

On the other hand, labeling theorists suggest that women are likely to be dealt
with in a more severe manner tlnn their male counterparts. Labeling theorists suggest
that individuals with lower status and less power are more likely to have their
criminality detected, labeled, and sanctioned because these individuals do not have
resources to manipulate the system to their benefit (Corley et al., 1989; Curran, 1983).
This “finding” of harsher treatment of females than males is therefore consistent with
studies that support the labeling perspective (Chesney-Lind, 1973; 1977; Curran, 1983).
However, as mentioned above, the finding of a harsher treatment of females than nnles
for the three offenses mentioned above is not statistically significant and therefore not a
frrm finding of this study. This position is supported by at least one study which failed

to find a statistical significance. Visher (1983:22), reported that “the overall arrest

75

percentages for males and females (20% versus 16%) are not significantly different and
chivalrous treatment of female suspects is not apparent.”

Gender analyses in this study also indicate that when race is introduced as a
control variable — white males compared to white fennles and black nnles compared to
black females — the consistent pattern of harsher treatment of females reported above
did not hold. For example, overall, black females received harsher treatment tinn black
males and as seen in Table 8, the differences in treatment in the shoplifting scenario is
significant. This finding of harsher treatment of black females over black nnles is
unsupported by a nnjority of the literature, which suggests significantly more lenient
police treatment of female over male suspects (Pollark, 1950; Haskell & Yablonsky,
1973; DeFleur, 1975; Price, 1977; Pastor, 1978; Sealock and Simpson, 1998).
However, some studies have suggested that black women do not receive the preferential
treatment that their white counterparts get (see Black, 1980; Smith & Visher, 1980;
Datesman and Scarpitti, 1980; Roberts, 2000). According to these studies, black women
are more independent than many white women because they fiequently occupy the role
of head of households, and females in black households usually have equal or greater
status than nnles members. Gender expectations for black males and females are less
differentiated than for white males and females, with females in black households
usually having equal or greater status than male members. Therefore, black females
may be less inclined than white females to display traditional gender behaviors in their
encounters with police thereby losing the preferential treatment enjoyed by their white
counterparts. If this is so, the implication is tint in the criminal justice system, black

females rrny not receive preferential treatment and may in fact sometimes be treated

76

more harshly than their male counterparts. This position is supported by one study of
comt decisions, which found that black ferrnles, relative to white females, were given
more severe dispositions than black males relative to white males (Datesman and
Scarpitti, 1980).

Finding tint the gender of the officer does not influence police arrest decision is
consistent with recent (post-1980) research on the issue. Research in the 19808 on the
effects of officer gender ins clarified previous (prior tol980) research cited by Sherman
(1980). Summarizing pro-1980 position, Sherman (1980: 73-74) reported mixed
findings, citing four studies that found female officers nnde fewer arrests than male
oficers and one study that reported no difference (see Block and Anderson, 1974;
Sherman, 1975; Sichel et al., 1977; Forst, Lucianovic, and Cox, 1977). However,
research from the 19808 indicate that officer gender had no effect on arrest behavior
(Worden, 1989: 701). It is noted here that patterns of arrest identical to those reported
prior to 1980 studies were actually found but the difference in the arrest decisions of
the officers are not statistically significant. As reported earlier, the pattern found by this
study is that male officers use more variability in their disposition options and are likely
to make more arrests than female officers. However, the differences here are trivial. In
an attempt to reconcile the conflict between prior and post-1980 research on this issue,
it is important to point out here that this study did not control for type of offense, race,
or demeanor of the suspects in this particular analysis (impact of gender of the officer
in police arrest decisions). Furthermore, this study had a small number of female
officers (N = 15 = 12.9%). Did this small number of fennle officers affect the results

of the present arnlysis under discussion? The gender composition of the police

77

department studied fo Hows the national pattern of gender composition of most local
police departments with 100 or more sworn officers (U .S. Department of Justice, 2000:
53). Therefore, most other police departments will invariably produce same small
representative sample of female officers. Again, due to differences in agency policies, it
was not advisable to combine officers fi'om different departments. Only the mean arrest
scores of the officers and not their number or the number of arrests they made were
compared. It is therefore submitted tint this difference in the number of female officers
my not have affected the results of this analysis.

The finding of this study on gender differences in police arrest decisions shows
tint overall, the gender of either the suspect or the police officer, is not directly related
to police arrest decisions in the offenses considered. As stated by Sloan (1991: 119),

“Indeed, gender may not be a significant factor in police arrest decisions but

instead, arrest decisions are based on other factors such as the seriousness of
the offense or the demeanor of the suspect.”

Visher (1983), Meyer (1981, 1982), and Stone (1985), all report that gender in and of
itself is not a significant predictor of arrest decisions. The results of this study may
point to a basic agreement with the findings of these other more seasoned researchers
mentioned above although different methodologies have been used.

Results of race differences in police arrest decisions reported in this study indicate
that except in the traffic scenario, there is no statistically significant difference between
police treatment of white suspects from that of blacks in the rest of the three offenses
considered. It seems therefore, that except as reported in the traffic scenario, race is not
directly related to police arrest decisions. However, in the traflic scenario, contrary to
expectations and to reports of earlier research (Friedrich, 1977; Powell, 1990), whites

are treated more harshly than blacks. This is a surprising result.

78

The analyses on racial differences conducted in this study shows that when gender
is introduced as a control variable, the above results hold except in one interesting
instance. In the shoplifting scenario, white males were significantly treated more
harshly than black nnles. This is another surprising result and runs against the results
reported previously. Powell (1990) for example, found tint “race of [the] offender is a
variable that strongly influences their [polices’] decision making process” (p. 15). He
reported that whites are treated more leniently than blacks. Friedrich (1977) concluded
in his study tint racial prejudice leads to higher arrest and ticketing rates for blacks and
has a complex impact on the policeman’s manner toward blacks.

Given this, how does one explain the main finding of a lack of significance of
race in police arrest decisions? It is necessary to point out tint some of the respondents
reported in the comment section of the questionnaire that they have clear, aggressive
and continuing race relations programs in the department. The details of those programs
were not nnde available to this researcher. However, it appears reasornble to speculate
tint it is possible tint these programs might have contributed to this positive finding as
to a lack of effect of race of the suspect on a police arrest decision. It is also possible
that these programs have nothing to do with this positive finding and that race is simply
not a factor in police arrest decisions, at least, in the jurisdiction studied.

The finding that white officers are likely to arrest more suspects than black
officers and minority officers (black, Hispanic and other racial groups combined) is
supported by prior-1980 literature. Summarizing the pre-1980 position, Sherman (1980:
74-75, 76) cited research suggesting that an oflicer’s race and attitudes influence police

arrest behavior. According to him, “Theorists of many persuasions have argued tint

79

black police emces behave differently from white oflicers.” (p. 74). black oflicers are
more aggressive and make more arrests than white officers (Friedrich, 1977; Sherman,
1980). However, research from the 19808 contradicted the above findings. Analysis of
the PSS data indicated that race of officer had no effect on police arrest behavior
(Worden, 1989: 701) and rennined non-significant across department types, that is,
legalistic, service, militaristic, or fiaternal (Smith and Klein, 1983: 87—89).

Again, as in the effect of the impact of officer gender as discussed above, it is
pointed out tint this study had a small non-white number of police officers and did not
control for type of offense, gender, or demeanor of the suspect in this particular
analysis. However, based on the reasons adduced earlier above, the same conclusion
was reached (tint is, the small number of black and other minority officers most

probably did not affect the results).

Cautionary remarks

A few cautionary remarks on factors that may have affected the results of this
study are warranted. The issue as to whether what respondents indicated they would do
in hypothetical crime episodes and whether it corresponds to how they would actually
beinve in the course of their daily routine duties is crucial to the validity of this study
and also to most studies with similar research designs. Against this background, every
possible effort was made to obtain unbiased responses and this researcher believes that
the responses are unbiased. Due to the importance attached to this issue, it is considered

that a little detail is necessary.

80

Vignettes were selected as a valid method for obtaining original empirical data to
determine how police respond, or at least say they will respond to a variety of people
and situations. The use of hypothetical cases permits the systematic manipulation of the
relevant variables. As stated by the authors of one study,

". . .such a stimulus would more closely approximate a real-life decision-
making or judgment-making situation. Furthermore, by holding the stimulus
constant over a heterogeneous respondent population, the survey researcher
gains a degree of uniformity and control over the stimulus situation .. .

[T]he vignette technique makes possible an analysis of the effects on

people's judgments while systematically varying the characteristics used on
the situation description"

(Alexander & Becker, 1978: 93). It should be pointed out here however that vignettes
may sometimes be inflexible and respondents may sometimes be biased, or at times lie
outright. Another major problem with vignettes i8 tint responses may differ from how
officers will respond when actually faced with similar situations in real life.

Each of the questionnaires used in this study was prefaced with a letter thanking
the ofiicers for participating. Included in tint letter of consent were assurances that
responses would remain strictly and absolutely confidential. Further, the directions
requested that any officer who had any concerns about this research was free to contact
the dissertation clnir whose name, address, telephone number, and e-mail address were
listed therein. The consent letter (see Appendix A) was signed by both the dissertation
chair and this researcher and was subsequently approved by the University Committee
on Research Involving Human Subject (UCRIHS). The measures outlined above were,
among other things, intended to create an atmosphere conducive for the officers not
only to volunteer information, but also to give honest and unbiased responses.

The Chief of the department where the data for this study were collected gave his

approval to ofiicially cooperate with this researcher. The chief was told in writing that:

81

__
. ~‘..

:'u.f\.'~l‘

 

“‘—.€..—- . .

(1) the department would not be identified by name in the study; (2) participation by all
officers was strictly voluntary; (3) responses by the officers would be kept anonymous;
and (4) the University Committee for Research Involving Hurrnn Subjects (U CRIHS)
would approve my questionnaire before data was actually collected. The chief gave his
approval subject to these conditions as itemized above. It is necessary to add here that
this researcher fulfilled both the “letter and spirit” of the above conditions.

The questionnaires were constructed to minimize attempts to adjust responses
artificially on the basis of race, gender, demeanor, type of offense and seriousness of
offense. However while officers certainly perceived variation by race and gender across
the episodes, demeanor and type of crime were also varied. Thus, it may be difficult to
equalize deliberately the severity of responses based on race, gender and demeanor of
the suspect.

Only full-time sworn field and patrol officers (about 230) were targeted for
this study. Administrative staff were excluded because they rarely carry out any
meaningful number of arrests and do not come into frequent contact with the
citizens. In addition, civilian employees were also excluded for obvious reasons.

As mentioned earlier (see the section on methodology), the data used in this
study were limited to police practices in a relatively large police department in a
Midwestern state. While findings technically indicate police practices in the
department sampled, they are suggestive in respect to other police departments and
should not be considered definitive of the general beinvior of all police officers.

Again, while the total number of scenarios responded to by the police

respondents was approximately 528 (132 officers multiplied by the number of

82

scenarios in each questionnaire, minus missing data), the number of “cases” in each
cell was much smaller (528 divided by 16 = 33). In addition, the percentage of
female respondents (12.9%), as well tint of “minority officers” (blacks = 10.9%;
Hispanics = 9.3%; others = 7.0%) were small and this might have affected some of
the statistical analyses. However, there is not much one can do here since this
mirrors the composition of most police departments nationwide. Furthermore, t-

tests are appropriate and in fact actually designed for arnlyses with such a number

 

of cases. This notwithstanding, this study may be better described in generic terms,

that is, “police” rather than in terms of race/gender of police officers. E
A few of the officers, especially women and racial minorities, were

concerned about the confidentiality of the specific biographic data requested. A

couple of officers for example noted that they were the only one in their race,

gender, or role assignment group and felt concerned about the anonymity of their

responses. They were reassured on the absolute confidentiality with which all

research subjects and nnterials are treated by both this researcher and Michigan

State University.
The generalizability of this study nny also be affected by the fact tint the

extent to which the attitudes and characteristics of officers in the same department

who did not participate are different fi'om those who did is not known (that is, there

may be a self-selection problem). However, this may not be a serious impediment

since not more than about 20% of the eligible officers did not participate in the

study.

83

The above cautions notwithstanding, on the basis of this research, there is
some justification for believing that: The nnjor variables in determining how
police officers will respond are the nature of the offense and the manner in which
the offender behaves when confronted by the officer. Gender of the suspect is
overall not directly related to police arrest decisions as neither the chivalry nor the
labeling perspective was supported. Gender of the officer is not related to police
arrest decisions. Race of the suspect is overall not directly related to police arrest
decisions. Race of the officer appears a relevant variable tint interacts with other
variables in police arrest decisions. Lastly, this analysis indicates that police officers

have a high degree of confidence in the arrest decisions they make.

Implications for official arrest statistics

Do these findings have any implications for official indicators of male/ female
criminality? It has been suggested tint the levels of criminality included in official
arrest statistics, such as the Uniform Crime Reports (U .C.R) under-represent the
involvement of female crimirnl activity. This conclusion was reached based on the
finding that females receive preferential treatment by police during arrest (V isher,
1983). Such an implication is not supported by the findings of this study. If any
differences exist in officially recorded levels of male/female crime levels, such

differences may not be significant.

84

Policy implications

The demographic characteristics of the respondents clearly shows that certain
groups — women and non-whites in particular, appear to be underrepresented in the
police department studied. As mentioned earlier, this is also the pattern of gender
and race composition of most police departments across the nation (U .S.
Department of Justice, 2000). The policy implication is for some changes to be
uncle in the recruitment policy of police departments. Such changes will provide for
a system whereby more qualified women and non-whites are given opportunities to
serve in our police forces. The present official position of President Bush on
aflirrrntive action may not help nntters here.

The trivial relationship between the variables studied and police arrest
decisions was linked partly to aggressive educational programs in the department
studied. If this is correct, similar programs are highly recommended for other police
jurisdictions. This study did not however investigate the nature and extent of these
programs.

Apart fi'om re-focusing attention on the issues of gender and race disparities at
arrest, and the utility of policy implications offered above, this study is also usefirl
and significant in other respects. F or example, findings suggest that gender disparity
at arrest may have existed in earlier times but is no longer with us any more
(Sherman, 1980). If this is correct, it implies that meaningful research efforts should
be diverted to other neglected areas.

Further, the argument is presented that unlike the procedure adopted by almost

all prior research, relatively smaller cities, not big cities, offer more logical sites for

85

studying gender differences at arrest. Another unique and special aspect of this
study worthy of mention is tint it offers an insight into the degree of levels of
confidence officers have in their arrest decisions. This study found a high degree of
confidence in officer arrest decisions. This implies tint these patterns of responses
are likely to be repeated by the officers. Departmental policies and training
materials can then be based or guided by the patterns of police arrest decisions as

reported by this study.

Directions for future research

For the firture, findings as to the effect of officer race on arrest decisions,
which are clearly inconsistent with post-1980 research merits further investigation.
As mentioned before, this study did not control for type of offense, race, and
suspect demeanor in arnlyzing the effect of officer race on arrest decisions. It is
suggested tint other research techniques (e.g., field observation) be incorporated
into future studies to strengthen research validity. Furthermore, following the
procedure adopted by this study, it is recommended tint in future research it be
considered to incorporate Hispanic officers in the arnlyses. While a few studies
have done this, such considerations were merely peripheral and incidental. The
percentage of Hispanic officers are continuously growing (U .S. Department of
Justice, 2000). They constitute a 9.3 percentage of respondents in this study while
black oflicers represent 10.9 percent of the respondents.

Race and gender-based differences at arrest are behavioral attitudes tint are

capable of changing with time, place, departmental policy and even the prevailing

86

civil rights environment. It is possible that patterns of arrest reported in this study
are limited to the period and department sampled. If this is so, future studies
utilizing appropriate sampling techniques and involving many police jurisdictions,
and conducted over a period of time, may offer additional insights into the issue.
Finally, both theoretically and empirically, social psychologists have reported
for years that there is at least a moderate relationship between people’s attitudes and
their behavior (see Sloan, 1991). Against the background of the foregoing, if
criminologists are to gain a clearer understanding of the nature and extent of ferrnle
crime and how the criminal justice system responds to it, efforts must be made to
examine not only the statistics on ferrnle crime but also how the attitudes of
criminal justice personnel affect their behavior. Only then can a more complete
picture of the criminal justice system’s response to female crime begin to emerge

(see Sloan, 1991).

87

“uni“.nn-Jr‘

APPENDICES

88

 

APPENDIX A

THE QUESTIONNAIRE (Type 1)

Please read each of the following incidents carefully and circle only the answer that
comes closest to how you would respond, given the inforrrntion provided.

Case # 1:

While on routine patrol duty one day, you observe two Afi'ican American women
arguing loudly at a shopping center parking lot. As you approach, you realize that the
two women are certainly drunk and arguing over a bottle of brandy. When you attempt
to question them, they become hostile and tell you to mind your own business. Then
they begin to curse you for not being out catching the real criminals.

(a). Given the above incident, which one of the following comes closest to how you
would respond?

Take no action

Question the suspects and let them go.
Issue a misdemeanor summons or citation.
Arrest them.

99’5”!"

(b). How confident are you that you would take the action you indicated in (a) above?

1. Very confident.

2. Confident.

3. Not sure

4. Net confident

5. Not at all confident.

89

Case # 2:

At a local discount department store, you notice a Caucasian man trying on a new pair
of shoes and then walking into another section of the store wearing the new shoes. You
next observe him dropping a radio into his shopping bag. As he walks toward the front
of the store, he puts two candy bars in his pocket. He then looks at his watch and walks
out of the door without paying for the new shoes, the radio or the candy bars. When you
stop him, he becomes very embarrassed, apologizes, and starts crying.

(a). Given the above incident, which one of the following comes closest to how you
would respond?

1. Take no action

2. Question the suspect and let her go.

3. Issue a misdemeanor summons or citation.
4. Arrest her.

(b). How confident are you that you would take the action you indicated in (a) above?

1. Very confident.

2. Confident

3. Not sure

4. Not confident

5. Not at all confident

90

 

Case #3:

At 1: 00 am, police receive a report of a loud and heated argument in an apartment
complex. When you arrive, an African American man comes to the door and asks you
to leave. He is very angry and states that slightly hitting his cousin with a stick is a
private family dispute. The man is uncooperative and refuses to answer questions,
while his cousin lies on the floor. A check at the hospital finds his cousin to be in
satisfactory condition.

(a). Given the above incident, which one of the following comes closest to how you
would respond?

1. Take no action.

2. Question the suspect and her go.

3. Issue a misdemeanor summons or citation.
4. Arrest her.

(b). How confident are you that you would take the action you indicated in (a) above?

1. Very confident.

2. Confident

3. Not sure

4. Not confident

5. Not at all confident

91

Case #4:

During routine patrol duty, you observe a Caucasian worrnn in a 2003 model Lexus
300 series driving 50 M.P.H. along a street with a posted limit of 35 M.P.H. The car
screeches to a halt at a red light and then two blocks later goes through a yellow light
coming very close to a car making a right turn. At this point, you stop the driver. The
worrnn is very polite and states she just purchased the car and was trying it out. A
routine check reveals only one previous violation for speeding.

(a). Given the above incident, which one of the following comes closest to how you
would respond?

1. Take no action

2. Question the suspect and let him go.

3. Issue a misdemeanor summons or citation.
4. Arrest him.

(b). How confident are you that you would take the action you indicated in (a) above?

1. Very confident.

2. Confident

3. Not sure

4. Not confident

5. Not at all confident

92

Please supply some background information about yourself by circling the appropriate
response or filing in the blank.

6. Sex:
1 .Male
2.Ferrnle

7. Race:

1.White

2.Black

3.Hi8panic

3. Other (please specify):

 

 

8. Age:
Year of birth: 19 ...........

9. Marital status:
1 .Married

2.Single, never married
3.Separated
4.Divorced

5.Widowed

10. Present rank:

1.Recruit

2.Patrol officer

3.Master police officer

4.Sergeant

5.Above sergeant

6.0ther: specify ............................................

11. Length of time in police work as a sworn officer

Specify: yearn).

 

93

12. Highest grade completed at school:
1.High school graduate

2.Two year degree

3.Four year degree

4.Graduate degree

13. How do you perceive your promotional potential in the police department?
1.Very good

2.Average

3.Below average

14. About how many traffic citations do you usually write in a one-month
period?

1. 0 - 9

2.10 — 20

3.21 — 30

4.31 — 40

5.41 — 50

15. About how nnny non-traffic misdemeanor arrests do you usually make in a
one-month period?

1.0 — 9

2.10 -— 20

3.21 — 30

4.31 - 40

5.41 - 50

16. What is your present role assignment in the department?
Briefly describe:

Comments: Please feel flee to add any comments you like. Thank you for your
assistance.

94

 

CONSENT FORM

Prospective Participants
Police Midwest Survey
The United States of America

Dear Participating Officer:

Having obtained the necessary approvals to distribute this confidential
questionnaire, we hereby invite you to participate in this research if you so desire.

Please do not write your name or provide any information that may identify you.
Your individual responses will be reported in aggregate form only. All information
you provide will remain in possession of the researchers and will be held in the
strictest confidence. Your privacy will be protected to the maximum extent
allowable by law.

The attached questionnaire contains four series of scenarios that an officer may
encounter while on duty. Please read each scenario carefully and circle the response
that best describes your response to the proposed scenario. If a protocol that
describes your potential action is not included, please feel flee to write in your own.
The estimated time required to complete this survey is five minutes.

The individual researchers and Michigan State University want to ensure that all
research participants are treated fairly and their rights protected. If you have any
questions regarding this particular study, please feel flee to contact the supervising
research investigator Dr. Charles J. Corley, at 522 Baker Hall, East Lansing, MI
48824. You may also reach Dr. Corley via telephone at (517) 353-5225 or by e-rrnil
at: corley@msu.edu. Furthermore, if you have any questions or concerns regarding
your rights as a study participant, you may contact Ashir Kumar, M.D., Chair of the
University Committee on Research Involving Hurrnn Subjects (UCRIHS) by
phone: (517) 355-2180, fax: (517) 432-4503, ennil: ucrihs@msu.edu or regular
nnil: 202 Olds Hall, East Lansing, MI 48824.

 

Participation in this project is entirely voluntary. You may choose not to participate
at all or decline to answer certain questions. Indication of your consent is
acknowledged upon receipt of your completed questionnaire. Please feel flee to
keep this introductory letter for your records.

Thank you for your participation.
Charles J. Corley, PhD., Alex 0. Ekwuaju
Associate Professor and Dissertation Chair Doctoral Candidate

95

 

APPENDIX B: SPSS CODING GUIDE

VARIABLE AND CODES NAME (WIDTH)

1. Case Identification CASEID (3)
Code exact (001-200+)

2. Questionnaire type QUEST (1)
Code exact (1 — 4)

3. Public Drunkenness response PDRUNK (1)
1 = Take no action
2 = Question and let them go
3 = Issue misd. summons
4 = Arrest them
9 = Missing

4. Confidence level (PDRUNK) CONDRUNK (1)
1 = Very Confident
2 = Confident
3 = Not sure
4 = Not confident
5 = Not at all confident
9 = Missing

5. Shoplifting response SHOPLIFT (1)
l = Take no action
2 = Question and let him go
3 = Issue a misd. Summons.
4 = Arrest him
9 = Missing

6. Confidence level (SHOPLIFT) CONFSHOP (l)
l = Very Confident
2 = Confident
3 = Not sure
4 = Not confident
5 = Not at all confident
9 = Missing

96

W

1-3

7. Assault Response ASSAULT (1)
1 = Take no action
2 = Question and let him go
3 = Issue a misd. Summons
4 = Arrest him
9 = Missing

8. Confident level (ASSAULT) CONFASS (1)
1 = Very Confident
2 = Confident
3 = Not sure
4 = Not confident
5 = Not at all confident
9 = Missing

9. Trafiic Offense response TOFFENSE (1)
1 = Take no action
2 = Question and let her go
3 = Issue a misd. summons
4 = Arrest her
9 = Missing

10. Confidence level (TOFFENSE) CONFTO (1)
1 = Very Confident
2 = Confident
3 = Not sure
4 = Not confident
5 = Not at all confident
9 = Missing

11. Officer Sex OFSEX (1)
1 = Male
2 = Female
9 = Missing

12. Officer Race OFILACE (1)
1 = White
2 = Black
3 = Hispanic
4 = Other (please specify): ...................................
9 = Missing

97

 

13. Officer Age OFAGE (2) 15-16
Code exact: Year of birth: 19 ..........
99 = Missing

14. Officer Marital Status OFMSTAT (1) 17
1 = Married
2 = Single, never married
3 = Separated
4 = Divorced
5 = Widowed
9 = Missing

neck

15. Officer Present Rank OFRANK (1) 18
1 = Recruit i-
2 = Patrol Officer
3 = Master Police Officer
4 = Sergeant
5 = Above Sergeant

6 = Other: specify ......................................
9 = Missing

 

16. Officer yrs of experience OF EXP (2) 19-20
Code exact (in years): Specify .................... years
99 = Missing

17. Officer Education (highest) OFEDU (1) 21
1 = High School Graduate
2 = 2 year degree
3 = 4 year degree
4 = Graduate degree
5 = Missing

18. Oflicer Promo. potential OFPROMO (1) 22
1 = Very good
2 = Average
3 = Below average
9 = Missing

 

98

19. # of mthly traffic citations MT CITAT (2)
1= 0- 9
2 =10— 20
3 = 21 —- 30
4 = 31 -40
5 = 40 - 50
99=Missing

20. # of mthly non-traffic citations MNTCITAT (2)
l=0—9
2= 10—20
3 =21-30
4=31—40
5 = 41- 50
99=Missing

21. What is your present role assignment in the department?
Briefly describe:

22. Comments: Please feel free to add any comments you like.

Thank you for your assistance.

99

23-24

25-26

Qualitative

Qualitative

ALPHABETICAL LIST OF VARIABLES

APPENDIX C

 

 

VARIABLE NAME VARIABLE DESCRIPTION VARIABLE LABEL

Dependent Variables

Assault Police responses to the assault Assault
offense scenarios.

Public Drunkenness Police responses to the public Pdunk
drunkenness offense scenarios

Shoplifting Police responses to the shop- Shoplift
lifting offense scenarios

Traffic Police responses to the traffic Toffense
offense scenarios.

Independent Variables

Case Identification Case number Caseid

Confidence Level, Police responses to assault Confass

Assault confidence level question.

Confidence Level, Police responses to public Condrunk

Public Drunkenness drunkenness confidence level
question.

Confidence Level, Police responses to shoplifting Confshep

Shoplifting confidence level question

Confidence level, Police responses to trafiic confi- Confto

Traffic dence level question.

Number of monthly Range of traffic citations issued Mtcitat

Traffic citations by the police officer per month

Number of non-traffic Range of non-traffic citations Mntcitat

Monthly citations issues by the officer per month.

100

 

 

 

Officer Age

Officer Education

Office marital status

Officer Present Rank

Office promotional
Potential

Oflicer Race
Officer Sex

Officer years of exp-
erience

Questionnaire

Age of the police officer

Highest education attained by
the police officer

Marital status of the police officer

Present rank of the police officer

Promotional potential as perceived
by the police officer

Race of the police officer
Gender of the police officer

Highest number of years as a
sworn police oflicers

Type of questionnaire (1 through
4)

101

Ofage
Ofedu

Ofrnstat
Ofi-ank

Ofpromo

Ofrace
Ofsex

Ofexp

Quest

 

 

APPENDIX D

SPSS SOFTWARE PRINTOUT AND OTHER TABLES/ FIGURES

Fgg' ure 1: Distribution of hypothetical offense episodes, demeanor, race, and
gender of the suspect across the four types of questionnaires.

 

Offenses :> Public Shoplifting Assault Traffic

Drunjkenness Offens_e
Demeanor :> Negative Positive Negative Positive
Quest. One = BF WM BM WF
Quest. Two = WF BM WM BF
Quest. Three = BM WF BF WM
Quest. Four = WM BF WF BM

CODES:

BF = Black Female
WF = White Female
BM = Black Male
WM = White Male
Quest . = Questionnaire

102

Figure 2: Race and gender composition of all “police officers” in the department

where data was collected.

 

Native Asian/ African Hispanic Caucasian
American Pacific Islander American
Males 3 5 22 12 95
Females 0 0 7 4 29
TOTAL 3 5 29 16 124

103

Table 15: Number of police officers and types of questionnaire completed and

returned.

 

 

Type of Questionnaire N completed Percent“
Quest 1 34 25.8
Quest 2 33 25.0
Quest 3 34 25.8
Quest 4 31 23.5
TOTAL = 132 100%

 

 

 

 

 

* = from SPSS output

104

Table 16: Gender differences in police arrest decisions: SPSS Group Statistics.

 

 

 

in

 

 

 

 

 

 

 

Gender Groups # of scenarios Mean Std. dev
Pdrunk: Quest 1 (Black Female) 34 3.1176 1.0664
Quest 3 (Black Male) 34 3.6471 .8121
Quest 2 (White Female) 30 3.1333 .9732
Quest 4 (White Male) 29 3.7931 .5593
All Males 63 3.7143 .7055
All Females 64 3.1250 1.0157
Toffense: Quest 1 (White Female) 33 2.9394 .3482
Quest 3 (White Male) 34 2.8824 .4777
Quest 2 (Black Female) 32 2.6563 .6016
Quest 4 (Black Male) 31 2.6452 .6082
All Males 65 2.7692 .5525
All Females 65 2.8000 .5062
Shoplift: Quest 1 (White Male) 34 3.9118 .5145
Quest 3 (White Female) 33 3.8182 .7269
Quest 2 (Black Male) 33 3.5152 .8704
Quest 4 (Black Female) 31 3.8710 .4995
All Males 67 3.7164 .7346
All Females 64 3.8438 .6228
Assault: Quest 1 (Black Male) 34 3.9118 .5145
Quest 3 (Black Female) 33 3.8182 .7269
Quest 2 (White Male) 32 3.6563 .9370
Quest 4 (White Female) 31 4.000 .0000
All Males 66 3.7879 .7548
All Females 64 3.9063 .5261

 

 

105

 

Table 17 : Race difi'erences in police arrest decisions: SPSS Group Statistics.

 

 

 

 

 

 

 

Racial Groups # of scenarios Mean Std. dev
Pdrunk: Quest 1 (Black Fennle) 34 3.1176 1.0664
Quest 2 (White Female) 30 3.1333 .9732
Quest 3 (Black Male) 34 3.6471 .8121
Quest 4 (White Male) 29 3.7931 .5593
All Whites 59 3.4576 .8575
All Blacks 68 3.3824 .9776
Toffense: Quest 1 (White Female) 33 2.9394 .3482
Quest 2 (Black Female) 32 2.6563 .6016
Quest 3 (White Male) 34 2.8824 .4777
Quest 4 (Black Male) 31 2.6452 .6082
All Whites 67 2.9104 .4167
All Blacks 63 2.6508 .60000
Shoplift: Quest 1 (White Male) 34 3.9118 .5145
Quest 2 (Black Male) 33 3.5152 .8704
Quest 3 (White Female) 33 3.8182 .7269
Quest 4 (Black Female) 31 3.8710 .4995
All Whites 67 3.8657 .6251
All Blacks 64 3.6875 .7319
Assault: Quest 1 (Black Male) 34 3.9118 .5148
Quest 2 (White Male) 32 3.6563 .9370
Quest 3 (Black Female) 33 3.8182 .7269
Quest 4 (White Female) 31 4.0000 .0000
All Whites 63 3.8254 .6849
All Blacks 67 3.8657 .6251

 

 

 

 

 

106

 

BIBLIOGRAPHY

Alexander, C. S., & Becker, H. J. (1978). The use of vignettes in survey research
Public opinion quarterly, 42: 93-104.

Alschuler, A. W. (1978). Sentencing reform and prosecutorial power: a critique of
recent proposals for ‘fixed’ and ‘presumptive’ sentencing. University of
Pennsylvania law Review, 126, p. 550.

Anderson, E. A. (1976). The chivalrous treatment of the female ofl‘ender in the arms of
the criminal justice system: a review of the literature. Social Problems, 23: 350-
357.

Babbie, E. (1983). The practice of social research 94"1 ed). California: Wordsworth
Publishing.

Bayley, D. & Mendelsehn, H. (1969). Minorities and the police. New York: Free Press.

Bernstein, I. Cardascia, J., & Ross, C. (1977). Defendant’s sex and criminal court
decisions, In R Alverez (ed). Discrimination in Organization.

Bittner, E. (1970). The function of police in modern society Washington, DC: National
Institute of Mental Health.

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

Black, D. (1971). The social organization of arrest. Stanford Law Review, 23(6), 1087-
1111.

Black, D. (1980). Dispute settlement by the police. In Black, D. (ed). The manners and
customs of the police. New York: Academic.

Bloch, P. B., & Anderson, D. (1974). Policewomen on patrol: Final Report.
Washington, DC: The Police Foundation.

Bozza, C. M. (1973). Motivations guiding policemen in the arrest process. Journal of
Police Science and Administration. 1(4): 469-176.

Chesney-Lind, M. (1973). Judicial Enforcement of the female sex roles: the family
court and the female delinquent. 8 Issues in Criminology 51 (1973).

Chesney-Lind, M. (1977). Judicial paternalism and the female status offender: training
women to know their place. Crime and Delinquency, 4, 122.

107

Chesney-Lind, M. (1978). Chivalry reexamined: women and the criminal justice
system. In L. Bowker (ed), Women, crime and the criminal justice system.
Lexington, MA: Lexington Books.

Corley, C. J., Cemkovick, S., & Giordano, P. (1989). Sex and the likelihood of
sanction. The journal of criminal law & criminology, (1989 ), 540-556.

Criminal Justice Collective of Northern Arizona University (2000). Investigating
drflerence: human and cultural relations in criminal justice. Boston, MA:
Allyn and Bacon.

Crites, L. (1976). Women Offenders: Myth vs. Reality. In L. Crites (ed).

Crites, L. (1978). Women in the criminal court. In W. Hepperle & L Crites (eds).
Women in the courts. Williamsburg, VA: National Center for State Courts.

Curran, (1983). Judicial discretion and defendants’ sex. Criminology, 21, 41

Datesman, S. & Scarpitti, F. (1980). Unequal protection for males and females in the
juvenile court. In S. Datesman & F. Scarpitti (eds). Women, crime, and justice.
New York: Oxford.

Davis, K. C. (1975). Police discretion St. Paul, Minn: West.

DeFleur, L. (1975). Biasing influence on drug arrest records: implication for deviance
research. American Sociological Review, 40(2), 88-103.

Ekwuaju, O. A. (1994). The impact of gender on police arrest decisions. Unpublished
Masters Degree Thesis - University of North Carolina at Charlotte.

Feeley, M. M. (1979). Perspectives on plea bargaining. Law and Society Review, 13(2),
199-209.

Federal Bureau of Investigation (1 993). Crime in the United States, I 992. Washington,
DC: United States Government Printing Office.

Federal Bureau of Investigation (1998). Crimes in the United States, I 997.
Washington, DC: United States Government Printing Oflice.

Feinman, C. (1982). Sex role stereotypes and justice for women In B. R. Price and R
J. Sokoloff (eds). The Criminal justice system and women: Offenders, victims,
workers. New York: Clark Boardrnan.

F inn, M. A., Stalans, L. J. (1997). The influence of gender and mental state on police

decisions in domestic assault cases. Criminal Justice and Behavior, 24(2), 157-
1 76.

108

 

Flowers, R. B. (1990). Minority and criminality. New York: Praeger Publishers.

Forst, B., J. Lucianovic, and S. J. Cox (1977). What happens after arrest? A court
perspective of police operations in the District of Columbia Washington, DC:
Institute for Law and Social Research.

Friedrich, R. (1977). The impact of organizational, individual, and situational factors on
police behavior. Unpublished PhD. dissertation, University of Michigan, Ann
Arbor. Dissertation Abstracts International, 38m 6908A

Goldstein, J. (1993). Police discretion not to invoke the criminal process: low visibility
decisions in the administration of justice. In G. F. Cole (ed). Criminal Justice:
Law and Politics. Belmont, California: Wadsworth.

Groves, R. M. (1996). How do we know what we think they think is really what they
think? In Schwarz, N. and Sudman, S. (eds.). Questions about questions:
inquiries into the cognitive bases of surveys. San Francisco: Jossey-Bass.

Harris, (1977). Sex and theories of deviance: toward a functional theory of deviant
transcripts. 42 Zm. Soc. Rev. 3, 13 (1977).

Harris, R, Huenke, C., Redriguez-Labarca, J. O’Connell, J. (1998). Disproportionate
representation of minority juveniles at arrest: an examination of I 994 charging
patterns by race. Dover, Delawre: Delaware Statistical Analysis Center.

Haskell, R., & Yablonsky, L. (1973). Crime and Delinquency. Chicago, Ill: Rand
McNally.

Healey, J. F., Boli, J .. Babbie, E. R, & Halley, F. (1999). Exploring social issues: using
SPSS for windows 95 versions 7. 5, 8.0, or higher. Thousand Oaks, Californai:
Pine Forge Press.

Hedderson, J., & Fisher, M. (1993). SPSS made simple (2"4 ed). Belmont, California.
Wadsworth.

Hollinger, R. C. (1984). Race, occupational status, and pro-active police arrest for
drinking and driving. J. Crim. Justice 12: 173-83.

Hutchinson, E. O. (1990). The mugging of black America. Chicago, Illinois: African
American Images.

Johnson, W., Peterson, R., & Wells, L. e. (1977). Arrest probabilities for marijuana
users as indicators of selective law enforcement. American Journal of
Sociology, 83(11), 568-599.

109

 

Klein, D. (1973). The etiology of female crime: A review of the literature. Issues in
Criminology, 8: 11 7-135.

Klinger, D. A. (1994). Demeanor or crime? Why “hostile” citizens are more likely to be
arrested. Criminology, 32: 475-493.

Klinger, D. A. (1996). More on demeanor and arrest in Dade county. Criminology
34(1), 61-81.

Krohn, M. D.; Curry, J. P.; and Nelson-Kilger, S. (1983). Is chivalry dead? An analysis
of changes in police dispositions of males and females. Criminology 21: 417-
437.

Kruttschitt, C. (1981). Social status and sentences of female oflendem. Law and Society
Review, 15(2), 247—265.

LaF ave, W. R. (1965). Arrest: The decision to take a suspect into custody. Boston:
Little Brown.

Lord, L. K. (1986). A comparison of male and female peace officers: stereotypic
perception of women and women peace officers. Journal of Police Science and
Administration, 14(2), 83-97.

Lundman, R. J. (1974). Routine arrest practices: a commonwealth perspective. ISocial
Problems, 22(10), 127-141.

Lundman, R. J. (1996b). Demeanor and arrest: additional evidence from previously
unpublished data. Journal of Research n Crime and Delinquency 33(3), 306-
323.

Lundman, R. J. (1996). Extralegal variables and arrest. Journal of research in Crime
and Delinquency, 33(3), 349-353.

Lurigio, A. J., Dantzker, M. L., Seng, M. J., & Sinacore, J. M. (1997). Criminal justice
statistics: A practical approach Butterworth-Heinemann: Boston, MA.

Maanen, J. V. (1973). Observations on the making of policemen. Human
Organization, 32(4), 407-418.

Mann, C. (1989). Minority and female: a criminal justice double bind. Social Justice,
16(4): 95-114.

Mann, C. (1995). Women of color and criminal justice system. In B. Price and N.

Sokoloff (eds). The criminal justice system and women: offenders, victims, and
workers, 2"d ed. New York: Clark Boardman, pp. 118-135.

110

 

Martin, S. (1989). Police officers on the move: a report on the status of women in
policing. In R. G. Dunham and G. P. Alpert (eds). Critical Issues in policing.
Prospect Heights, Ill: Waveland Press.

Merlo, A. V., & Benekos, P. J. (2000). What ’s wrong with the criminal justice system:
ideology, politics and the media. Cincinnati, OH: Anderson.

Meyers, A. R; Heeren, T., and Hingson, R. (1989). Discretional leniency in police
enforcement of laws against drinking and driving: two examples from the state
of Maine, USA. J. Crim. Just. 1 7: 1 79-186.

Moulds, E. F. (1980). Chivalry and paternalism: disparities of treatment in the criminal
justice system In S. Datesman and F. Scarpitti (eds). Women, Crime ad Justice.
New York: Oxford.

Moyer, I. L. (1981). Demeanor, sex, and race in police processing. Journal of Criminal
Justice, Vol. 9, pp. 235 -246.

Moyer, I. L. (1982). Police responses to women offenders in a southern city. Journal of
Police Science Administration 10:376—83.

Muraskin, R. & Roberts, A. R. (2002). Visions for change: crime and justice in the
twenty-first century. Upper Saddle River, New Jersey: Prentice Hall.

National Advisory Commission on Criminal Justice Standards and Goals (1973).
Police. Washington, DC: US. Government Printing Office.

Nees, H. H. (1986). Police use of discretion: a comparison of community, system, and
oflicer expectations. (Unpublished Doctoral Dissertation, University of
Colorado at Denver). In Dissertation Abstracts International, 49, 626A.

Neuman, W. L. (2000). Social research methods: qualitative and quantitative
approaches (4:1. ed). Boston, MA: Allyn and Bacon.

Neuman, W. L., & Wiegand, B. (2000). Criminal justice research methods: qualitative
& quantitative approaches. Boston, MA: Allyn and Bacon.

Niederhofl‘er, A. (1967). Behind the shild Garden City, New York: Double-day Press.
Norusis, M. J. (1988). SPSS/PC + V2.0 Base Manual. Chicago, Ill: SPSS Inc.

Norusis, M. J. (1993). SPSS for windows: Base system users ’ guide, release 6.0.
Chicago, Il: SPSS Inc.

Ogawa, B. K. (1999). Color of justice: culturally sensitive treatment of minority crime
victims (2“ ed). Boston, MA: Allyn and Bacon.

lll

 

Pastor, P. (1978). Mobilization in public drunkenness control: a comparison of legal
and medical approaches. Social Problems, 25(Spring), 373-384.

Pollak, O. (1950). The criminality of women. New York: Barnes.

Powell, D. D. (1990). A study of police discretion in six southern cities. Journal of
police science and administration, 17(1), 1-7.

Price, R. R (1977). The forgotten female ofi‘ender. Crime and Delinquency, 23 pp.
101-108.

Riksheim, E. C., Chermak, S. M. (1993). Causes of police behavior revisited. Journal
of Criminal Justice, Vol. 21, pp. 353-382.

Roberts, D. (2000). Shattered bonds: the color of child welfare. New York: Basic
Books

Rubinstein, J. (1973). City police. New York: Farrar, Straus & Giroux.

Russell, K. K. (1998). The color of crime: racial hoaxes, white fear, black
protectionism, police harassment, and other macroaggressions. New York:
New York University Press.

Schur,E. (1983). Labeling women deviant. Philadelphia: Temple University Press.

Sealock, M. D., Simpson, S. S. (1998). Juvenile offender type-script’s. Justice
Quarterly 15(3), 427-457.

Sherman, L. W. (1975). An evaluation of policewomen on patrol in a suburban police
department. Journal of Police Science and Administration 3 (4): 434-438.

Sherman, L. W. (1980). Causes of police behavior: the current state of quantitative
research. Journal of Research in Crime and Delinquency 1 7(1), 69-100.

Sichel, J. (1978). Expectations which have shaped women ’s role in policing.
Unpublished paper presented to the national Institute on Crime and
Delinquency: Miami, Florida (June, 1978).

Sichel, J ., et a1. (1977). Women on patrol: a pilot study of police performance in New
York City. Washington, D. C.: Government Printing Office.

Sloan, J. J. (1991). The ‘new’ female criminal revisited: police oflicer attitudes about
female criminality. American Journal of Police, 10(4), 105-130.

112

 

Smart, C. (1976). Women, crime and criminology: a feminist critique. London:
Routledge and Kegan Paul.

Smith, D. A. (1982). Invoking the law: determinants of police arrest decisions.

Unpublished Doctoral Dissertation, Indiana University. In Dissertation
Abstracts International, 43, 2797A.

Smith, D. A. (1984). The organizational context of legal control. Criminology 22: 19-
38.

Smith, D. A (1986). The neighborhood context of police belmvior. In A. J. Reiss, Jr.

and M. Tonry (eds.) Crime and justice: 0 review of research. Chicago:
University of Chicago Press.

Smith, D. A. (1987). Police response to interpersonal violence: defining the parameters
of legal control. Soc. Forces 65: 767-82.

Smith D. A, & Visher, C. A. (1980). Sex and deviance/criminality: an empirical review
of the quantitative literature. American Sociological Review, 45(8), 691-701.

Smith, D. A, and Visher, C. A. (1981). Street level justice: situational determinants of
police arrest decisions. Soc. Prob. 29: 167-77.

Smith, D. A & Klein, J. R. (1984). Police control of interpersonal disputes. Social
Prob. 31: 468-481.

Stockdale, E., & Casale, S. (1992). Criminal justice under stress. London: Blackstone
Press Limited.

Stone, H. (1985). The changing pattern of female crime: a police perspective. In I. L.

Moyer (ed), The changing roles of women in the criminal justice system.
Prospect Heights, IL: Waveland Press.

Sykes, R. 13., & Brent, E. E. (1983). Policing: a social behaviorist perspective. New
Jersey: Rutgers University Press.

Thoresson-Rogers (1991). The impact of an audience on police arrest decisions.
Unpublished Master’s degree thesis, University of North Carolina at Charlotte.

US. Department of Justice: Bureau of Justice Statistics (2000). Source book of
criminal justice statistics. Washington, DC: USGPO

Van Wormer, K. S., & Bartollas, C. (2000). Women and the criminal justice system.
Boston, MA: Allyn and Bacon

113

Visher, C. A (1983). Gender, police arrest decisions, and notions of chivalry.
Criminology, 2(1), 5-28.

Waaland, W. B. (1985). Police decision making in wife abuse: the impact oflegal and
extra-legal factors. Law and Human Behavior 9: 355-366.

Walker, S. (1998). Sense and nonsense about crime and drugs: a policy guide (4“ ed).
Belmont, CA: West/Wadsworth.

Walker, D., Spohn, S., & De Lone, M. (2000). The color of justice: race, ethnicity, and
crime in America. Belmont, CA: Wadsworth Thomson Learning.

Walklate, S. (2001). Gender, crime and criminal justice. Portland, Oregon: Willan
Publishers.

Walsh, W. F. (1986). Patrol officer arrest rates: a study of the social organization of
police work. Justice Quarterly 2:271-290.

Worden, R. E. (1989). Situational and attitudinal explanations of police behavior: a
theoretical reappraisal and empirical assessment. Law & Soc R. 23: 667-707.

Worden, R E., and Pollitz, A A (1984). Police arrests in domestic disturbances: a
further look. Law andSoc R. 18: 105-119.

Worden, R E., Shepard, R. L. (1996). Demeanor, crime and police behavior: a
reexamination of the police services study data. Criminology, 34(1), 83-105.

Worden, R E., Shepard, R L. & Mastrofski, S. D. (1996). On the meaning and
measurement of suspects’ demeanor toward the police: a comment on

“demeanor and arrest” Journal of Research in Crime and Delinquency 33(3),
324-332.

114