WI IIIan' II'III fi‘u 3 ”III I’JII“”I‘IIIZ“ "

 
 
   
  
  
 

 

1 W1 IIIII. 1 1,1111 I 3 - ‘15:," . '
. 'I ‘f1'IIIIIII13‘III' 3I._I I11 1IIII3'I1‘II 'r: . .1. 3‘: 111
3 131% II "I”“IIII l - ~-' 1 131‘“ I II Im .‘I.:NI$W II! ' Ms'mu III
1 . | "1 1|, H I 1&4 “I 3 .I 7- II' _'1|I «I 11 )bIgIr I. 11.33”“ V l, 133-511" g1 1.5g ‘ '4
I'I- IIII'III' l‘III. III IIICIIVIIII'IL'I'IIIJIIII ,.II‘1II IIIIII I‘M II I 'IIIII III‘IIIIEIII 11111111313131; L331 I ‘3 . 11312:,11319111‘1‘3
I I. It: . .1 .r 1 r'."!_11‘1|”,1 . I. I; I 113,111I ,‘ .\ ‘, . 31-2.).11I 121...“
9 .

 

     
     

I I‘I' II‘I'1L lq‘J‘ ‘0 L 5‘ \
3""\'II l H! ‘1211' ': A f I,
IJL‘umvux' 'III1 ”I I’II'IIIIII‘II‘ N1I$ . H II,1ffl fiIyIt?€$§§I”n3
{v

I3 3..I' 'IIII1LI' 3 ’- (€11.11.3,Il"1 "31-1 $151,154,111”.
WIUWIJINIWIIII'3fiIII "“““1“
III I I 'III“II'IIIIIIIIIIIIIII ,I.;I.3III3I‘I““' :3 II

  
 
   
  

11.11" 11

I
(“131111, 431:1
I? 3131‘“ '1‘

1. 1...,1 %

 

'I

    
 

 

 

 
  
 
  

 

     
    
  
  

 

   

 

       
  

 
   

    
    

3:43

   

 

151:“ U n

   

 
 

.
<1

 

"II: . , 13., . I: II'I::::II-Iifi‘1Is..I I23.
11I,.H‘I.1111' I1IIIIII’: 'iI-II- 4 111I ”1111;111:153!

-1 'L3I'II I{.tI I‘»I.I.u 11IQ“11»:, II1'IIIII- Ii 11 I, 1 ‘ ’Tn' ‘1‘?! w1§$ 1:: 1111, ...
" 'III’II'IIIIIns-IIIIIIHI IIIIIIIIIIIIIII III;III III 'IIII II'-IIIIIIII I {Ing‘IIIII IIIIIIIIII . WWI?” JI
'74 uflam I I WAIVQnfiIbI ::Iu IQNIII: "1%., “* ‘fififigf a

I. vII3r1.IWIBI<n;qfiIr.khI . WI1 1.1231I1I I: -~ 23 31% .a:§: 3g

~ "3: “13911qu ‘311‘3‘3‘ '."I-I;I'1I"' 'r'IIIIfi'III ,IIL1II1.II,;3I.I 111111113. ~ -.,;“5': x {:5'

'51 : _11?1_1.1£II«IIIII:‘.1I ,IzuII‘ -3311‘1I1.131111 " II'B‘II 11'":'1;‘.1I'12111‘§= I 1.:g
2:: z-II'lzafI;1=III~§‘3I1 31': 313-11, IIII’JIIIIIIIT‘I.’ '3! MI IIIUI II‘I‘I. '3
I 1:: , I: 13,13,“;113111111121111Iv11-1111:1'1 11151 1‘I'I; "~ 31 II113:1. 91111 1
1'; ' 11.111111134111I‘1II '1 \1111f1I31113'I‘31111 ,1111111111111511‘1411111111 I
2.1- ””‘i‘Iz’I'I III‘I. II=1I?3~.I§ ,=_-,K';'I;.. .!

.1: 1' 11‘1Jf‘i 1| $11 1113111 can ”13.14.:1XI'II‘M - 11V .11" 1...,1 51:3, . .

=I1I11‘31111I-1t‘4‘3 :- «15151111131 1 I‘ - , 111‘

."'. I
«1’ '

     
  

.—
‘L
. or:

I
.4
'0

  
   
  
   

  
  
 

 
   

f

I

3., W’s

#3:..132»
"Ht. H.

..-

‘W :90
L
figm.‘

In 1,,
111,14 :1 :1-1-11 1

 

'..-

..

   

u

“'47.! OJ"-;"
. v ' I -

I. 3L“; ‘ I
c—wUH-C’

 
 
 
   

 
 
 

. ‘11
3‘.- ' 3.".
1' 9,1‘ 12‘
‘1; "a'
:d- I3,”
:' :.

   
 
 

3:
.:—3

:' .H
a.15111,—,1 I
“3111 I

111 1:‘ r‘ 1;;

     
   

 
 

'1:: 1W ‘

  
     
  
    
  

  
  

IA. ..
_.-, ,_,
_,_ ... .1
~...—..v 3e .. . I : '-
‘ ... A. AAI . : . . .- n -
r ' I J
... - '_ ...: I ‘ : A A
3-_- .,, ..K:.. _
- . .. ...“ .
. 3- :.. "< . --

   
 
  

    
 
 
 
  

   
 

 

 
   

 
  
   
    
    

  
 
 

 

 
    
  
     
   

   
 
 
      

    
     
       

       

 

I I :'::j 13:52.13“ ' 3! SI.
151 III! MEI): III'III 1'13? II F
III""_:§.-'II “113113, MI 1:151
vawmwtm:I
"IIIII '3II‘.IIII"‘II ‘I‘IIII' . i-IIII I
'1MI,11 , I11 1,..1,1.
' ‘II‘I'IIIII'IIII-‘I‘I-I'"I1III'IIJIIII’ III' . 3
-IhquuIarsJJ, IIIIII I” r:
-*t11 I 'IIIIMIHQRIILh, ' WInhI1 ‘1IC
III III -I2I 1:31;: :I: III. I
"h”‘7 ' -I“I3Wm9
3-. JQCI {.1 1:1 I -3{%
III ‘3 1.111111". 5| III I. :I3 . .', _ ;
11I ' LI 131:1}I11I1I1 III! II- If}! ' 3f." 3 II;E1I I ‘II'I

“up
. .
h‘I
a...

“Hit, t,
9

“V‘. 0 -,3
...- .

 

.2“
:-
....—

"L‘. .;»,
..-”... —
....

.5 ' I.
- =;.._ , -v
p‘- :— _‘.‘ 2'
' “a N .

\ .,..'.

_,_..

..
::.._
{I

.
-—.~_.
2;:
.— ..... .

     

- 3 ~ I'I’IIIIIII‘E’II'I‘“
I' ‘2'. 11' ' , ;§.'1,.3;3.f”

'7.
Ci"; .
i .

II?- 1 - 3 '
I '1. 1‘16“.
’11?“ 15.551".

..._.
...--

<-.

 

 

 

 

 

 

 

 

 

 

 

 

llTl'llllll/lllllllllHill/till

3 1293 01707 4281

This is to certify that the

thesis entitled
A CROSS-SECTIONAL ANALYSIS OF MURDER CLEARANCES

IN DETROIT: 1994-1995
presented by
SEAN PATRICK VARANO

has been accepted towards fulfillment
of the requirements for

_MASIERS_ degree in LRIMINAL. JUSTICE

Major professoi

Date MAY 5, 1998

 

0-7639 MS U is an Affirmative Action/Equal Opportunity Institution

 

 

 

i

l
r
l
i

 

LIBRARY

Michigan State
Universlty

 

 

PLACE IN RETURN BOX
to remove this checkout from your record.
TO AVOID FINE-3 return on or before date due.

 

MTE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1/” WWW.“

A CROSS-SECTIONAL ANALYSIS OF MURDER CLEARANCES IN DETROIT:
1994-1995

By

Sean Patrick Varano

A THESIS

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

MASTER OF SCIENCE

School of Criminal Justice

1998

ABSTRACT

A CROSS-SECTIONAL ANALYSIS OF MURDER CLEARANCES IN DETROIT:
1994-1995

By

Sean Patrick Varano

Homicide clearance rates have been steadily decreasing over the past thirty years.
Researchers have hypothesized that changes in the nature of crime caused by sociological
phenomena such as the drug industry are at least partly responsible for lower clearance
rates. The current research entails an analysis of two-hundred homicide incidents that
occurred in the city of Detroit between 1994 and 1995. Data indicate similar findings
between the victim and offender demographic information and previous research.
African-American males are overrepresented among both homicide victims and
offenders, and homicide victims tend to be younger than homicide offenders. Where the
victim-offender relationship was known, most incidents involved victims and offenders
who knew each other prior to the incident. When victim-offender relationship
information was excluded from the multivariate analysis, victim’s violent criminal
history, incident location, time reported, victim culpability, and drug-motivated

circumstances developed as the most significant predictors of homicide clearance.

To my family, especially my parents, whose loving support and encouragement are the
sole reason this work was completed. Dad, you are perpetually in our thoughts and
prayers...

iii

ACKNOWLEDGMENTS

I am deeply grateful to my thesis committee for their unending support and
encouragement throughout this project. Specific acknowledgments go to Dr. Tim Bynum
for giving me the opportunity to be part of the team that was responsible for collecting the
Detroit data. Without this opportunity, none of this would have been possible. To Dr.
Christina DeJong for always being available when I needed to discuss an idea, and reading
the multiple drafts of the manuscript. To Dr. Michael Reisig for also always being available
to run ideas by, and for not being too upset at the length of the proposal.

I am also grateful the Justice Research and Statistics Association and the National
Institute of Justice for making this data available. The quick and thorough data entry by
JRSA staff made the analysis process substantially more convenient.

Acknowledgments are made to the Homicide Section of the Detroit Police
Department. Their open door policy not only made the data collection processes smooth, but
the willingness of the detectives to speak to us gave a unique insight into homicide
investigations in Detroit. Special thanks are extended to Sergeant Chester Kicinski and
Officer Paul Kersanty who always helped us feel at home.

Finally, to my friends and fellow students at Michigan State who have made graduate
school one of the best experiences of my life. Especially to Tracy O’Connell, Scott Kunkel,

Brian Carroll and Marcus Mizanin who put up with me during the process of writing this

iv

thesis. They all were extremely patient during my seemingly endless frustration and
aggravation.

Words to not begin to express my appreciation to all of you!

TABLE OF CONTENTS

LIST OF TABLES ..................................................... viii
LIST OF FIGURES ..................................................... ix
CHAPTER 1
INTRODUCTION ................................................. 1
CHAPTER 2
NATURE AND PATTERNS IN CRINflNAL HOMICIDE ................. 8
Patterns in Race and Sex Characteristics ......................... 10
Age Patterns ............................................... 12
Weapons and Methods of Injury Infliction ....................... 15
Temporal Patterns .......................................... 16
Patterns in Alcohol Influences ................................. 17
Previous Record ............................................ 18
Motives .................................................. 20
Victim/Offender Relationship ................................. 22
CHAPTER 3
CORRELATES OF POLICE CLEARANCE RATES .................... 27
Victim Characteristics ....................................... 29
Weapon .................................................. 3 1
Previous Criminal Record .................................... 32
Felony Circumstances ....................................... 33
Victim-Offender Relationship ................................. 34
Temporal Impacts .......................................... 34
Spatial Characteristics ....................................... 35
Physical Evidence .......................................... 35
Presence of Witnesses ...................................... . 37

vi

CHAPTER 4

METHODOLOGY ............................................... 39
Location .................................................. 41
Time Frame ............................................... 43
Sampling Frame ............................................ 44
Information Sources ......................................... 47
Data Instrument ............................................ 47
Variables ................................................. 48
Statistical Techniques ....................................... 55
CHAPTER 5
DESCRIPTIVE AND INFERENTIAL RESULTS ....................... 58
CHAPTER 6
CONCLUSIONS ................................................. 76
Limitations of Findings ...................................... 82
Implications ............................................... 82
APPENDIX A ......................................................... 87
APPENDIX B ......................................................... 90
REFERENCES ........................................................ 99

vii

LIST OF TABLES

Table 1 - Variable coding and preliminary descriptive statistics .................... 51
Table 2 - Additional descriptive statistics for sample ........................... 59
Table 3 - Percentage of cases cleared by arrest, by categories of independent variables. 65

Table 4 - Logistic Regression coefficients, significance levels, and odds ratio. All
variables included in model. .............................................. 70

Table 5 - Logistic Regression coefficients, significance levels, and odds ratio. Victim-
ofi‘ender Relationship information excluded from model ........................ 72

viii

LIST OF FIGURES

Figure 1 - National Clearance Rates ................................ Appendix A
Figure 2 - Population Race Characteristics ........................... Appendix A
Figure 3 - Population Sex Characteristics ............................ Appendix A
Figure 4 - Population Educational Characteristics ...................... Appendix A
Figure 5 - Mean Incomes ......................................... Appendix A

ix

Chapter 1
INTRODUCTION

Fear of crime, particularly violent crime, has been a fundamental political,
economic, and social concern in recent decades. While it may be argued that interest in
rising crime rates has always been high in the United States, the phenomenon has been
pushed into the spotlight since the 1960's. The national interest has manifested itself in
several different ways, most of which involve periodic times of relatively substantial
levels of federal, state and local resources to attack a gamut of social issues. The past
fifty years have been an era of such initiatives, including welfare programs and their
modification, tens of billions of dollars for law enforcement assistance,‘ and numerous

other projects that were both directly, and indirectly, developed to address these concerns.

The present research will focus on criminal homicides, and will attempt to
determine what situational variables affect the ability of police to solve these crimes.
Thus, the primary research aim will be to determine what factors, when present, are of the
greatest importance to making an arrest for a particular murder offense. A secondary, yet
significant aim focuses, on the nature of criminal homicides. The current analysis was

influenced by Marvin Wolfgang’s (1958) landmark analysis of homicides in Philadelphia

 

' Assistance to local and state law enforcement agencies was begun as a large scale project in the 1960’s
with the federal govemment’s Law Enforcement Assistance Agency. This program funneled large scale
resources into local and state law enforcement efforts. The funds were primarily used for equipment,
training, personnel, and education for law enforcement officers (Samaha, 1991). More recently, with the
Violent Crime Control Act of 1994, the federal government has allocated nearly $10 billion for equipment,
personnel, and overtime. In addition, the 1994 Crime Bill also made provisions for training projects,
research, and the development of innovative strategies for dealing with crime issues (United States
Department of Justice, 1994).

2

from 1948-1952. Findings from the Detroit sample will be compared with previous

research to determine how the nature of homicide has changed.

The concurrent development of increasing criminal homicide rates and decreasing
clearance rates (see Figure l in Appendix A) that occurred during the period from 1971 to
1990 sparked considerable interest among academicians and practioners alike.2 From
1971 to 1990 there was a 33 percent increase in the number of known homicides that
coincided with a 174 percent increase in the number of uncleared homicides (Cardarelli
and Cavanagh, 1992, p. 2). There are three primary reasons why this should raise
concern. First, the fact that offenders are not caught for nearly one-third of all homicide
cases raises questions about safety and security, especially when it is considered that
some of these incidents probably involve repeat offenders. Second, since crime rates and
clearance rates are seen as measures of police effectiveness, evaluations of police
operations are often negatively affected, which subsequently affects police morale.
Finally, the criminal incident itself tends to have a traumatic impact on victims and their
families that is only made worse when the offender is not arrested (Riedel and Rinehart,

1996,p.84)

The literature review will address several topics that are essential to the
understanding of homicide clearances. The first section of the literature review will be

used to define terms and concepts that are necessary to gain an understanding of the

 

2 While a more detailed discussion will be presented on the definition of a "cleared" case, it will refer to
those cases in which a suspect was arrested for the crime in question.

3

research at hand, and will include discussions on the legal classifications of homicides,
and definitions of case clearances. Definitions of clearance rates will be supplemented
with information on how they are developed, and how they are used by the criminal
justice system will then be discussed. The first section of the literature review will
present findings on patterns that exist in the homicide incident. It will address several
facets of the victim—offender interaction, and other general incident characteristics.
Finally, findings from previous research will be used to determine what victim and
situational characteristics have been shown to affect case clearance. Offender
characteristics will not be discussed in the analyses because as logic implies, offender
information is not readily available for most open cases. Exceptions to the rule will occur
in those cases in which a suspect has been identified, yet has not been taken into custody.
Such cases are technically still uncleared cases. Finally, a bivariate and multivariate
statistical analysis will be conducted to determine the relationship between particular

situational and victim characteristic variables and final status.

The current research has both theoretical and practical implications to academics
and practioners alike. The discipline of sociology, and more recently the field of criminal
justice, have invested a substantial amount of time and energy in determining what
patterns actually comprise criminal incidents, and secondly how these patterns relate to
the causal factors of crime. It is only by understanding the nature of crime, in this case
criminal homicide, that theorists and/or public policy makers can attempt to develop

prevention or intervention programs that effectively deal with the underlying causal

4

factors. Accordingly, Chapter 2 presents findings from previous research about the
characteristics of criminal homicide in the city of Detroit. Notwithstanding the objective
value of establishing patterns in crime, Chapter 2 is also intrinsically related to Chapter 3
in that it begins to explain the effect that victim and situational characteristics have on
homicide clearances. In determining how the nature of the crime itself might have
changed, if at all, it allows the reader to conceptualize a relationship between any

apparent changes and their relationship to decreasing homicide clearance rates.

This research is potentially valuable to police managers or executives who are
considering the implementation of a case screening model that allocates resources more
effectively. vaindling police department budgets in conjunction with more competition
for the limited resources that are available have prompted managers to develop strategies
that identify and give priority to cases that have higher probability of successful clearance
(Greenberg, Elliot, Draft, and Procter, 1977). Although it seems apparent that the serious
nature of criminal homicide requires a substantial amount of investigative effort to all
cases, it is still possible to develop procedures that focus attention to cases having a

higher probability of clearance.

Chapter 2

LITERATURE REVIEW

The scope of this research project is limited to criminal homicide and focuses on
determining what factors are relevant to the clearances of these crimes. Research
focusing on one type of crime is pivotal to gathering a more accurate understanding of the
dynamics that occur during the interaction between the participants of that crime. Albert
Morris (1955) argued the "serious investigator" errors by studying crime as a whole, as it
would be more beneficial to study the "patterns, similarities, and repetitions" of specific
crimes, so that we may develop "effective control, prevention, and treatmen " efforts (p.
3). Other researchers have supported this contention and point out that research on one
particular type of crime, including victim-offender characteristics, is relatively rare
(Wolfgang, 195 8, p. 4). For these reasons the scope of this research has been limited to

one crime, namely criminal homicide.

For purposes of this research, clearance refers to those instances where an
offender has been identified an_d arrested or detained3 for a particular crime. This is an
important definition because the term "cleared" often has different formal and informal
operational meanings. For crimes such as burglary and robbery some agencies determine
a case to be cleared if detectives are able to match the modus operandi of a suspect in

custody with that of other crimes (Cordner, 1989). Other agencies do not consider a case

 

3 In many jurisdictions juveniles cannot legally be arrested, and are technically "detained" when they are
taken into police custody for violations of the law.

6

cleared unless at least one perpetrator is taken into custody, which is consistent with the
definition used here (Bottomley and Pease, 1986) . Still others argue that a case is not
cleared unless all suspects are taken into custody (Greenwood, Chaiken, and Petersilia,

l 977) .

One scenario presents itself as an exception to the above definition. It was
previously stated that the offender must be taken into custody in order for the case to be
considered cleared. Most departments allow for the status of "exceptional clearance."
There are instances, however, where an offender has been identified, yet it is not possible
to apprehend the suspect. Examples are when a victim refused to cooperate (generally
not an issue in homicides), the suspect was prosecuted by another jurisdiction, the suspect
died, or the suspect was already in custody (Cawley, Miron, Araugo, Wasserman,
Mannello, and Huffman, 1977; Stephens, 1996). Many police departments include these
cases in calculations as cleared cases because exclusion underestimates police

effectiveness (Rinehart, 1994).

Clearance rates are generally used as a measure of police effectiveness in police
operations. The reader should exercise caution when interpreting these rates because it is
widely accepted that official statistics on known offenses ofien substantially under
represent victimization rates as reported by the National Crime Survey (N CS). Of all
crimes reported to police, it is logical to assume that the number of known homicides are

more representative of actual victimization rates versus other crimes such as burglary,

7

larceny, or assault. One factor influencing the reliability of homicide victimization rates,
however, could be the manner in which departments determine whether the crime is
actually a criminal homicide versus a crime that is justified by citizen, natural causes, or
suicide. Misclassifications have the potential to inflate or deflate both homicide and

clearance rates.

Police managers have historically used clearance and other arrest-based statistics
as a measure of police effectiveness. There are two primary reasons for using clearance
rates. First, arrests are easy to count. Second, because the police have direct control over
their production. A result of the apparent overemphasis on arrests and clearance rates is
that the arrest itself often becomes the objective rather than solving the crime problem the

arrest is supposed to address (Bayley, 1996; Stephens, 1996).

A principal of the Total Quality Management (TQM) model suggests that to

measure quality, managers should focus on outcomes not outputs. Output measures are

 

measures that merely count the frequency of particular activities/events (e. g., number of
meetings attended by personnel, number of arrests). Conversely, outcome measures are
used to determine an organization’s success in achieving a particular organizationally
defined goals (e.g., decrease in juvenile crime in a particular geographical location), and
usually involves a predefined method of evaluating effectiveness. Outcome measures
differ from output measures in that personnel activities do not become goals, yet the

effect of personnel activities are. Policing has often used output measures such as

"simplistic tallies of arrests, citations, field interrogations, and clearances to measure
the adequacy of police performance on both the individual and agency level" (Hoover,
1996, p. 17) even though it is commonly held that these alone do not measure how well
police respond to citizens. Police executives may be reluctant to use outcome measures
such as the number of cases in which charges are filed, or number of convictions that
result from an arrest because it is often argued that they have no control over whether a
prosecutor decides to file charges, if a grand jury decides to indict, or if a jury finds a
suspect guilty. Police can reasonably argue that even in ideal circumstances, these

outcome measures would not be an effective and reliable measure of performance.

There is growing criticism directed toward the use traditional measures of police
effectiveness such as clearance rates (Bayley, 1996; Cook, 1979; Hoover, 1996). One
possible reason is that effectiveness of the criminal justice system can often be mediated
by the choice behavior of criminals. There is a relationship between the choice behaviors
of criminal, such as the level of caution used by a burglar in choosing their modus
operandi. So, it is argued clearance measures are merely a reflection of the relationship
between police strategies and personal choices of criminals, and not some abstract

measurement of police effectiveness (Cook, 1979).
Nature and Patterns in Criminal Homicide

In determining the impact that victim and situational characteristics have on the

clearance of homicide cases it is first important to understand the patterns that exist in the

9

typical homicide incident. This is important because many researchers have argued that
crime is a social phenomenon which often follows predictable patterns in both victim and
situational characteristics (Block, 1976; Block and Christakos, 1995; Blumstein and
Heinz HI, 1995; Daly and Margo, 1982; Lee, 1995; Rojeck and Williams, 1993; Segall
and Wilson, 1993; Wolfgang, 1958), behavior (Bernard, 1993; Levi, 1980; Luckenbill,

1977; Messner and Tardiff, 1985), and victim/situational characteristics.

The recognition of these patterns and interactions has resulted in the development
of sociological theory to explain the reasons for their existence. The causal factors of
crime, and subsequently the nature of crime itself, have traditionally been explained using
one of two paradigms. The positivistic school argues the propensity to commit crime is
largely determined by forces outside the control of the individual. Conversely, the
deterministic school argues criminals are differentiated from noncriminals in the extent to
which the individual restraints himself/herself from committing crime (Gottfredson and
Hirschi, 1990). The deterministic school argues the individual is responsible for
committing crime, versus being encouraged to commit crime by other factors such as

economic inequality.

The patterns that exist in criminal homicide can be explained using nearly every
criminological theory. Although multiple criminological theories are relevant, one of the
most applicable theories to this study is the "routine activities" theory. The theory posits

the nature of the victim-offender relationship is the most important factor in determining

10

the participant’s socio-demographic characteristics (e.g., age, sex, race, marital status, and
employment status) and the temporal characteristics (e.g., time of day, day of week, and
time of year) of the incident. The structure of the participant’s daily activities dictates the
characteristics of the crime. For example, crimes occurring in the home are more likely
to involve family members because interaction between family members is greatest in

this location (Messner and Tardiff, 1985).

The nature of homicide has remained rather steady over time while homicide rates
have fluctuated, thus characterizing homicide as both a stable and changing phenomenon
(Block, 1976). Generally, homicides have been classified as intraracial (Walker, Spohn,
and DeLone, 1996), mostly between close acquaintances (Maxfield, 1989), and
committed by those who have essentially no criminal career (Wolfgang, 1958). The next
section of chapter two will present previous findings about the patterns that exist in

homicide incidents.
Patterns in Race and Sex Characteristics

Marvin Wolfgang (1958) was one of the first researchers to analyze patterns in a
particular type of crime. He thoroughly scrutinized characteristics of the victim, offender,
and situation, presenting a descriptive analysis of these characteristics. Wolfgang’s study
focused on age, sex, race, and previous record differences of both the victim and offender,
and also situational characteristics such as weapons used for inflicting death, temporal

patterns, spatial patterns, alcohol use, motives, and victim/offender relationship. In

ll

addition, Wolfgang’s research also examined the effect of multiple victim/offender
characteristics on situational characteristics (e.g., the weapon of choice for black males,

versus white males).

The most striking feature of race and sex distributions in criminal homicide is the
extent that percentages for victimization and offending for blacks exceed those of whites,
and the extent that rates for males far exceed the rates for women (Wolfgang, 195 8). The
Philadelphia study revealed that while Blacks comprised only 18 percent of the US.
population in 1950, they accounted for 73 percent of homicide victims and 75 percent of
homicide offenders. Black males were 36.9 times more likely to be homicide victims,
and 41.7 times more likely to be homicide offenders than white males (Wolfgang, 1958,

pp. 33-34).

Since the 1950’s, other researchers have arrived at similar findings. Substantial
support has been found for the conclusion that African American males face a much
greater risk of death by homicide than do whites. More recently, it was determined that
the homicide victimization rate for African American males was 72 per 100,000, while

the rate was 9.3 for white males (Walker, Spohn, and DeLone, 1996, p. 34).

12

Age Patterns

A significant‘ relationship has generally been found between criminal homicide
offending rates and age, however this relationship is relative to the offender or victim age
and race. Black male offenders tended to be younger than white male offenders. The
highest offending category for black males was the 15-19 group (92.5 per 100,000), and
the highest category for white males was the 20-24 group (8.2 per 100,000) (Wolfgang,
195 8). The data reveal that the offending rates for white males are also proportionately

much smaller.

Offending patterns women are different from that of men as they tend to commit
much less crime. White female offenders tend to be younger than black females, and they
proportionately commit much less homicide.s The highest offending rate for black

females was in the 25-29 age group (22.3 per 100,000).

Patterns in victimization rates tend to be consistent with offending rates.
However homicide victims generally are slightly older than offenders (Wolfgang, 1958;
Segall and Wilson, 1993). The Philadelphia data revealed that victimization rates were
highest for blacks in every age group, with the highest being for black males in the 35-39
group (69.4 per 100,000). Rates for white males were highest in the 30-34 group, and

this rate of 4.0 is again substantially lower than that of black males.

 

‘ "Significant" will refer to those instances when a test for significance was conducted, and a value for p is
less than .05 (Wolfgang, 1958, p. 18).
5 The highest category for white females was the 15-19 group, and the offending rate was 1.2 per 100,000.

13

It appears that a more recent trend has been for homicide offenders and victims to
be younger than was previously the case. Analyses of Chicago homicides from 1970-
1993 revealed that while similar patterns were present, the findings differed in relation to
the age groups that experienced the highest offending rates. Offending rates for
Chicago’s 15-24 age category were consistent with the 20-24 category in Wolfgang’s
(1958) research until the 15-24 category’s rates dramatically increased after 1985 (Block

and Christakos, 1995)

It is not clear what cultural or structural changes have affected these patterns in
offending and victimization levels, but Decker (1996) suggested that the crack cocaine
epidemic that hit many major cities in the 1980’s has changed the dynamics of homicide.
This epidemic has created instrumental motives (e. g., robbery and rape)6 within primary
relationships (e.g., family), and has diminished the traditional protection that such
relationships previously provided. Primary relationships were thought to shield family
members from crimes having instrumental motives. Evidence suggests this insularly
dynamic is no longer as prevalent. Case narratives collected by Decker (1996) also
suggest that "homicide networks" might actually facilitate the contagious spread of

violence (p. 441).

 

6Incidents involving expressive motives are those where the homicide incident itself is the primary
motivation. Instrumental motives are present when the primary reason for the criminal act is to achieve
some other ends, for example the presumed financial reward from a robbery. While the two motives can
overlap, the primary motivation for the incident is emphasized.

14

Several connections have been established between drug-markets and violent
crime. Drug markets have altered the levels and type of lethal violence in the United
States (Block and Christakos, 1995; Blumstein, 1995; Blumstein and Heinz III, 1995;
Decker, 1996; Rosenfeld, 1989). Goldstein (1985) identified a tripartite framework that
encapsulates the major subcategories of drug-related crimes. Drugs and violence are
related in three ways: psycho-pharmacologically, economically compulsive, and
systemically. The psycho-pharmacological model suggests short or long-term use of
specific substances may result in persons becoming susceptible to violent behavior due to
a prolonged state of excitability and/or irrationality. Economically compulsive crimes are
those where the offender’s primary motivation is to obtain money to support his/her drug
habit. Finally, the systemic model suggests violent behavior is a natural outcome of the
normal process of the sale, distribution, or use of narcotics. Violence is a means to
streamline the drug business, and is often used to settle disputes over territory between
rival drug dealers, enforce normative codes, retaliate (e.g., selling phony drugs, failing to

pay one’s drug debt), or to eliminate informers (Goldstein, 1989, pp. 24-30).

Another possible explanation is that changes in population demographics are
linked to changes in homicide patterns. Since the 1950's, there has been an increase in
the proportion of young people in the population, especially among 15-24 year olds who

commit a disproportionate amount of crime (Markides and Tracy, 1977).

15

Weapons and Methods of Injury Infliction

There are mixed findings about the importance that type of weapon plays in the
homicide situation. It has been argued that the presence of a suitable weapon was the
most significant factor in the homicide incident - - a revolver, a shotgun, a rifle, a hunting
knife, a butcher knife, a heavy bottle, or a club (Topping, 1952). Others argue that the
lack of a suitable weapon at the homicide scene does not always have a protective effect.
In certain situations, participant(s) often leave the scene, acquire a firearm, and return to

the setting (Lundsgaarde, 1977).

Depending on the motivation of the offender (instrumental versus expressive), the
typical homicide is usually quick, brutal, and the weapon is usually simple and
commonplace (Wolfgang, 1958). It is not altogether clear what Wolfgang (1958) meant
by "commonplace," but it can be inferred that he proposed a relationship between types of
weapons and cultural orientation. It has been argued that historical circumstances, such
as frontier violence, racial segregation, and arming of the populace, are important to the
study of American crime (Block, 1981). These particular historical and/or cultural
orientations might relate to what has been referred to as the "subculture of violence."

The "subculture of violence" thesis argues that cultural differences make presence of
weapons, violence, and use of violence to solve conflict, as more or less acceptable
(Wolfgang, 1958). Wolfgang (1958) cites the Royal Commission on Capital

Punishment’s following statement as further support for this idea: "Irish immigrants in

16

the United States used to get into all sorts of fights, but their homicide rate was always
low because they fought with their fists and brass knuckles, and not with knives or

revolvers" (p. 80).

A significant relationship between race, sex and weapon preference has been
found, and these preferences are probably due to socialization that certain groups have to
a specific kind of weapon (Wolfgang, 1958). For every age and race category, "pistol,
revolver" had the highest proportion of involvement in criminal homicide incidents,
followed by "knife." This pattern remains when analyzed for the offender, as well as the
victim.

Temporal Patterns

Criminologists have long assumed a relationship between crime, especially
violent crime, and season (Tennenbaum and Fink, 1994). Emile Durkheim (1951) was
the first to study this from the approach of determining the social structural factors which
influence this phenomenon, and argued that behaviors such as suicide and homicide are
actually a result of temporal factors, versus individual choice. Some have of
hypothesized that crimes against the person are highest in the summer because people are
frequently in contact with each other, and the consumption of alcohol is high (Sutherland
and Cressey, 1978). The Philadelphia data did not reveal significant associations.
However, most homicides did occur during the months of May, June, July and August

(36 percent), which are followed closely by the months of March, April, September, and

17

October (35 percent) (Wolfgang, 195 8, Table 8). Other studies have made similar
findings, but added that homicide is cyclical in nature and that the patterns are sometimes

masked by temporary trends or influences (Tennenbaum and Fink, 1994).

Temporal influences are not only apparent when analyzed from the most deadly
months, but significant relationships have also been found to exist between days of the
week, as well as time of the day. The Philadelphia data reveals that for both races and
sexes, Saturday is the most deadly day of the week (31.8 percent), followed by Friday (17

percent) and Sunday (16.8 percent) (Wolfgang, 1958, Table 8).
Patterns in Alcohol Influences

Although the social and economic costs of alcohol use and abuse are well
documented, fifty years after the abolitionist era the exact role of alcohol in our culture is
not clearly defined. There is a well established relationship between the use of alcohol
and criminal activity, yet this phenomenon appears to be particular to the United States
(Parker, 1995). Since there is not a significant difference in the genetic pool between the
United States and other countries, there must be social structural conditions that influence

this relationship between alcohol and crime (Ibid.).

Although the alcohol/crime nexus has been investigated from multiple
viewpoints, Wolfgang’s Philadelphia study is still the classic piece of literature that
addresses alcohol and its relationship to violent behavior (Ibid.). Wolfgang (1958) found

that alcohol was present in 63.6 percent of all criminal homicide situations, 43.5 percent

18

of which involved the presence of alcohol in both the victim and offender (p. 136, Table
14). The Philadelphia data revealed significant relationships between race, sex, and
alcohol. Alcohol was present on average in 57 percent of all homicide incidents, and was

most prevalent among victims and offenders who were black males.

It has been argued certain features of alcohol justify an investigation of this topic
separate from that of other substances such as drugs. Drug-related violence is usually a
direct result of the sale and distribution of drugs, not the use of drugs per se. Alcohol
differs in that the mere use of alcohol is associated with violence. It is argued that this
difference requires a unique conceptual understanding of drug-related violence, and the

result that this violence has on the victim-offender interaction (Parker, 1995, p. xiii).

Analyses of 689 of the homicides from St. Louis that occurred between the years
1985-1989 revealed over one-quarter were reported to be drug-related. Drug-related was
defined as any drug involvement in the homicide incident, including drug use by either
the victim or the offender the day of the incident. Drug-related homicides tended to
involve a larger percentage of blacks, males, guns and acquaintance relationships than

non-drug-related homicides (Rosenfeld, 1989, p. 12).
Previous Record

Information on victim’s and offender’s previous criminal record has primarily
been gathered from one of two sources, either court/commitment records or arrest

records. The source of this information is important because the data sources do not

19

necessarily reflect the same information. Court/commitment records not only reflect the
decision of the police to arrest, but also reflect an evaluation of guilt, and/or a court’s
decision to incarcerate. Researchers often incorrectly refer to an offender’s or victim’s
"criminal history" when discussing the number of arrests attributable to that individual.
This information only reflects arrest records and is not necessarily a reflection of criminal
history (Wolfgang, 1958). Wolfgang’s (1958) criticisms appear to be too critical in
nature because he follows a very legalistic approach. Arrest records are one of the mostly
widely available pieces of information that measure previous criminal records. When
interpreted correctly, they provide an accurate picture of a victim or offender’s prior
criminal history. For these reasons it is argued here that arrest records are the most

accurate source of information about the victim and offender’s previous criminal record.

In general, research prior to 1958 found that most offenders had no previous
record of any kind before the criminal homicide incident.7 These findings were
contradicted by the Philadelphia data that revealed offenders (and probably victims) in
criminal homicide appear to have previous arrest records in greater proportion than
elsewhere reported in the literature (Wolfgang, 195 8). Results indicate a significant
relationship between previous arrest record, and the race and sex of the offender and

victim. Sixty-eight percent of blacks offenders had a previous criminal record, compared

 

7 Lewis Lawes’ found that 90 percent of those committed to Sing Sing for murder had no previous record.
Lawes, Dublin, and Bunzel found that 75 percent of those admitted on a charge of homicide were classified
as first time offenders, however, they also warn that available data might not adequately evaluate a
person’s criminal career (as cited in Wolfgang, 1958, pp. 169-170).

20

to 53 percent of white offenders. F ifty-four percent of blacks victims had previous

records, compared to 28 percent for whites (Wolfgang, 195 8, Table 19).

A significant relationship also exists for sex. Sixty-eight percent of male
offenders and 54 percent of male victims had a previous record, compared to 48 percent
of female offenders and 25 percent of female victims (Wolfgang, 1958, Table 19). Race
and sex differences not only have a significant impact on previous record, but also on the
type of crime previously committed. The Philadelphia data revealed that 66 percent of
offenders, and 54 percent of victims had a previous record for an offense against the
person. In addition, 34 percent of offenders, and 46 percent of victims had previous

records for crimes against property (Wolfgang, 1958, pp. 178-179).
Motives

Researchers have continually cited the motivation for criminal homicides as
important to understanding of the criminal event, and that it is this characteristic that
distinguishes criminal homicide from other crimes (Wolfgang, 1958). It is difficult to
undertake a discussion on the role of motives without also presenting a discussion on the
intimate connection between motives and the victim-offender relationship. While this
discussion will touch on this connection, a more detailed discussion on the

victim/offender relationship will follow.

Criminal motives, especially motives for criminal homicide, have long been

dichotomized into two categories. The names or labels associated with those two

21

groupings have varied slightly, yet the fundamental distinction has remained rather
consistent. The distinctions have been referred to as "felony" versus "nonfelony" (Levi,
1980; Luckenbill, 1977-1978; Wolfgang, 1958), "instrumental" versus "expressive"
(Block and Christakos, 1995; Decker, 1995; Decker, 1996), or "instrumental" versus
"impulsive" (Block, 1981). Whatever label is assigned to the terms, felony/instrumental
homicides are those that result during the commission of another crime, and generally
occur as a result of the offender attempting to achieve some level of personal gain (e.g.,
when a store owner is killed during the commission of a robbery). Conversely,
nonfelony/expressive homicides are those incidents that involve an expression of some

immediate or long term outrage, yet bring no personal gain to the offender.8

Previous research indicates the most significant factor influencing the motive of
the incident is the victim-offender relationship. Expressive motives are most prevalent in
homicides involving family, friends, or other acquaintances, and instrumental homicides

are most prevalent between strangers (Decker, 1995).

A distinction has also been make between "normal" and "deviant" homicides
(Decker, 1996). "Normal" homicides are those fit into the generally accepted

conceptualization of "acceptable" homicides, those where the motivation is consistent

 

' It is generally accepted that expressive homicides do not include an aspect of personal gain. However it
could be argued that the mere confrontation brings some level of personal gain to the offender and/or
victim. Meaning, the confrontation itself could be a mode of relieving some level of stress or emotion
within either party. The type of personal gain suggested as part of an instrumental crime generally refers to
material or financial gain, as in the case of robbery, or the sexual gratification in the case of a rape. In
expressive homicides, the goal is the violence itself, other motives are secondary (Block and Christakos,
1995, p. 28).

22

with the victim-offender relationship. Examples would be expressive violence between
family, fiiends, or acquaintances, or instrumental violence between strangers. "Deviant"
homicides are those incidents that do not fit into this conceptualization. Examples of
deviant homicides would be instrumental violence between intimates, and expressive

violence between strangers (Decker, 1995).

Notwithstanding the victim/offender relationship, previous research has found that
the nonfelony/expressive homicide is most common. While it is not completely clear if
Wolfgang’s (195 8) classifications of homicides are consistent with that of other research
cited here,9 the Philadelphia data reveal that only 7.9 percent of offenders had
instrumental motives.lo Similarly, Maxfield (1989) found that 11 percent of homicides
were instrumental, with robbery being the most common motive (p. 674). Thus,

expressive homicides are much more prevalent than instrumental homicides.
Victim/Offender Relationship
Criminal homicide has been characterized as:

"[T]he most personalized crime in our society. Because motives do not
exist in a vacuum, the subject-object, doer-sufferer relationship is of prime
importance in this particular crime... .[H]omicide is a dynamic relationship

 

9 Wolfgang (1958) made many more distinctions between motives then are present in any other piece of
literature. He has distinguished between, among others, general altercations, domestic quarrels, jealousy,
and robbery. It seems apparent to this author that this classification more accurately reflects homicide
circumstances rather than motives. For example, it could be argued that it is completely possible to have a
domestic quarrel which is motivated by some instrumental motive. While Wolfgang (1958) might have
been aware of this problem and coded the most obvious motive, this limitation is not addressed in his
research.

'° The only motive that is clearly instrumental in nature included by Wolfgang (1958) is robbery.

23

between two or more persons caught up in a life drama where they operate
in a direct, interactional relationship" (Wolfgang, 1958, p. 203).

Relationships like motives, are often dichotomized by researchers. Relationship
categories are usually specified as either nonstranger or stranger (Block, 1981; Riedel,
1981; Decker, 1995), or primary or secondary (Smith and Parker, 1980; Parker, 1989).
Wolfgang’s (1 95 8) categorization of victim-offender relationship included 11 categories
that can be grouped into the nonstranger/primary or stranger/secondary dichotomy. "
Realizing the limitations caused by the dichotomous categories, Decker (1995) and others

added the third category of friends/acquaintances.

Sixty-five percent of all victim/offender relationships in the Philadelphia sample
fell in the primary contacts categories, and 35 percent of the incidents involved strangers.
This distinction is even more pronounced when considering the significant impact that
sex and race have on the victirn-offender relationship. Categories that involve primary
group contacts, when combined, constitute 59 percent of all relationships among males,
but 84 percent among females. When specified to both sex and race, incidents involving
white males tend to be between strangers, while the incidents involving black males tend

to be among nonstrangers (Wolfgang, 1958, pp. 206-210).

These findings are consistent with a study that found 17 percent of homicide

incidents involved strangers, while 47 percent involved either acquaintances or

 

" Wolfgang’s (1958) categories include: (a)close friend; (b)family relationship; (c)acquaintance;
(d)stranger; (e)paramour, mistress, prostitute; (f)sex rival; (g)enemy; (h)paramour of offender’s mate;
(i)felon or police officer; (j)innocent bystander; (k)homosexual partner (p. 204).

24

spouses/lovers (Maxfield, 1989, pg. 673). Readers should exercise caution when
analyzing relationship data because when no offender has been identified, the information
is usually missing (Maxfield, 1975). Dealing with missing data, especially data that
might be systematically missing, is a serious problem in the social sciences and is
particularly problematic with homicide data sources. It has been argued that incidents
with a substantial portion of missing offender data tend to involve strangers (Williams

and Flewelling, 1987).

Historically, the family structure was thought to protect individuals from
instrumental crime. Recent trends indicate that deviant homicide, as conceptualized by
Decker (1996), is on the increase. The insularity protection of primary relationships has
decreased (Decker, 1996). It has been argued that drugs, especially the crack epidemic, in
conjunction with the breakdown of the family structure (Hacker, 1992; Decker, 1996),

have tremendously altered this naditional protection.
Conclusions

Prior research supports the conclusion that generally, criminal homicide follows
specific patterns, which on the whole, have been relatively stable and consistent over

time. The most significant results are:

0 males and blacks have consistently been the highest proportion of both offenders
and victims of criminal homicide;

0 both victims and offenders tend to be relatively young, however, offenders tend to
be slightly younger than their victims;

25

0 alcohol has been shown, at least in the United States, to have a substantial impact
on the nature of criminal homicide;

O the motives for most offenders tend to be more expressive, and;
0 generally, victims and offenders tend to maintain some level of acquaintanceship
before the criminal incident.

These findings are important because by understanding the nature of crime in any
given population, the reader is better able to discern the relationship between these
patterns and criminological theory. Routine activities theory is an appropriate model for
conceptualizing the relationship between victim-offender sociodemographic
characteristics and the nature of the crime. The theory is supported by the results
presented above as stranger homicides tend to involve males, especially white males.
Since the socioeconomic status of white males is generally higher than that of black
males, it would follow that the social mobility of white males is greater. Higher incomes
provide more flexibility "for the dispersion of activities across physical and social space
(social mobility), while low incomes serve to constrict activities around the immediate
household" (Messner and Tardiff, 1985, p. 245-246). This increased social mobility
would bring white males in greater contact with strangers, thus increasing the chances
that they would be involved in fatal assaults involving strangers. Since the social
mobility of women, especially when Wolfgang’s research was conducted, is substantially
lower than that for men it would stand to reason that they would more often be killed by a

family member or acquaintance. This conclusion is again supported by the data.

26

Finally, by comparing these findings with data obtained from Detroit, it allows the
reader to discern how the nature of homicide has changed. One can then hypothesize a

relationship between any changes or similarities and homicide clearances.

Chapter 3
CORRELATES OF POLICE CLEARANCE RATES

Research on correlates of criminal homicide clearance rates have been taken from
numerous perspectives, including the impact of situational factors, police organizational
factors, and social structural influences. Clearance rates are commonly used as a measure
of police productivity. However, numerous theorists and practitioners have argued that
this rate has limited utility (Bayley, 1996; Cook, 1979; Stephens, 1996). A negative
relationship has been reported between crime rates and clearance rates. While there is
controversy as to the usefulness of the clearance rate figure, the present research will not
question the legitimacy of using it as a measure of effectiveness. Since the figure is
commonly cited, and is often used both practically and strategically in police planning

and operations, however erroneously, its utility will be assumed.

Most previous research pertaining to the clearance of crimes focused on crimes
other than homicide such as burglary and robbery, and these findings have subsequently
been extrapolated and applied to criminal homicide. An examination of the relationship
between victim characteristics and homicide clearance was conducted to determine if
particular variables including victim sex, race, age, or previous criminal record are
correlated with case clearance. Situational variables such as the presence and type of

weapon, felony circumstance,12 and victim-offender relationship, which have previously

 

'2 Felony versus nonfelony homicide: An assessment was made to determine if the homicide event took
place during the commission of a concurrent felony act such as a robbery, rape, burglary, or arson.

27

28

shown to be significant correlates, are included in the analysis to determine their
relationship to case clearance. In addition, the relationship between spatial/temporal
characteristics (season, time of day were, and physical location) will also be assessed.
Subsequently, this research determines if the presence of physical evidence and witnesses

at the scene have an appreciable effect on case clearance.

Previous research has also demonstrated that seriousness of the offense is often
related to clearance of crimes (Bynum et al., 1982;13 Riedel and Rinehart, 1994; Waegel,
1982). However since this study is limited to one type of crime, and since criminal
homicide is generally considered to be the most serious of all crimes, it would be nearly
impossible to define the idea in such as way as to achieve a significant level of
variability. Although it is anticipated that this variable would have limited value in this
instance, the relationship might be more significant in cases where the crime was

particularly heinous.

Finally, this research assesses the relationship between victim culpability and case
clearance. Previous research suggests victims often play some active role in their own

demise (Levi, 1980; McDonald, 1976; Wolfgang, 1958; Wolfgang, 1967). While victim

 

'3 Consistent with other studies, Bynum et a1. analde the impact that particular characteristics had on
court decisions and outcomes. It can be argued that these studies are relevant for several reasons. First, the
decision to prosecute, and the subsequent disposition of the case, are contingent upon police discretion. It
is argued that these factors, while important at later stages of the crime processing system, are originally
deemed relevant or irrelevant at the commencement of the process, the investigation. These decisions
often set the stage for future determinations by representatives of the criminal justice system (Bynum et al,
1982; Gaines, 1983)

29

culpability is difficult to quantify, it will be assessed according to particular situational

variables present.
Victim Characteristics

Victim traits seem to be more useful in explaining police decisions to investigate
versus offender characteristics (Bynum et al., 1982). These influences can be broken
down into two categories, physical (e.g., victim age, sex, and race) and situational (e.g.,
victim-offender relationship, and the interaction between the two in the criminal incident)
(Denno and Cramer, 1976). For property offenses, none of the personal characteristics of
the victim were shown to be significantly related to the level of investigative effort except
for victim’s income, which was marginally significant (Bynum et al., 1982). Personal
offenses differed in that none of the victim’s personal characteristics have been shown to
be significantly related to investigative effort. A nonlinear relationship has been
observed between victim age and investigative effort, indicating cases involving the very

old and very young were more intensely investigated (Bynum et al., 1982; Boris, 1979).

When a suspect was not immediately identified for burglary cases, social
characteristics of the victim, particularly the victim’s class position and race, had a
decisive impact on the particular handling strategy adopted (W aegel, 1982). These
strategies primarily refer to the investigative techniques adopted, such as canvassing a

neighborhood for information, or processing a scene for latent prints. The use or non-use

30

of particular strategies is important because they can have a tremendous impact on the

likelihood of a follow-up investigation being conducted.“

While the impact of victim characteristics is not clear, the majority of evidence
suggests the impact of these variables is negligible at best. Furthermore, the affect of
victim characteristics appears to be specified to property crime versus personal crime
suggesting that burglary victims from higher socio-economic statuses, for example, are
perceived to be deserving of more intensive investigative effort (Bynum et al., 1982).
The just world theory is one possible explanation for this relationship. The theory argues
people get what they deserve, either rewards or punishments. In instances where it is
perceived that this natural balance is violated, meaning an individual received an
undeserved "punishment," efforts are required to rectify the apparent injustice. Based on
this theory, in certain situations police may view victims as more or less deserving of
police attention (McDonald, 1976). Others have argued that police often use some sense
of perceived victim vulnerability to determine the appropriate level of effort (e. g., some
police may view women as inherently more vulnerable than men, subsequently more

deserving of intensive effort) (Bynum et al., 1982).

 

" Many departments, whether informally or formally, use "case screening" to identify those cases that are
most likely to be cleared with a moderate amount of investigative effort. Waegel (1982) argues that
detectives use informal "interpretive schemes" to screen out low solvability cases in order balance the
practical problems of deadlines and making arrests (p. 453). "Part of the detective’s interpretation of a case
is based upon his understanding of the victim’s lifestyle, racial or ethnic group, class position, and possible
clout or connections..." (p. 454). Case stereotypes for particular crimes such as rape, assault, and homicide
provide "recipes" for handling cases(Waege1, 1982, p. 464).

31

Weapon

In an attempt to develop an effective case screening model, Gaines, Lewis, and
Swanagin (1983) found that of the 11 variables included in a model, presence and type of
weapon was one of five variables considered to have a significant impact on case
clearance." The relationship between the presence and type of weapon and case
clearance has been supported by some researchers (Cardarelli, 1993; Cardarelli and
Cavanagh, 1992; Rinehart, 1994), and others found little support for the relationship
(Riedel and Rinehart, 1996). In addition, Oakland Police Department’s case screening
model includes weapon used as one of thirteen significant factors (Greenberg and

Wassherman, 1979, p. 36).

One factor influencing the usefulness of weapon information is in the ability of
the department to systematically analyze and search records pertaining to weapons. The
problem might be similar to that latent fingerprints left at crime scenes, which will be
discussed in greater depth later. Most police departments rarely conduct "cold searches"
on latent prints. Latent prints are usually only used to match an identified suspect to a
crime scene (Chaiken, Greenwood, and Petersilia, 1977; Seamon, 1990). Similarly, the
value of information provided by ballistic evidence is greatly diminished by not

conducting "cold searches" on ballistic findings.

 

'5 For this model, the following factors and weights were found to be significant: suspect associate named
(508); clothing evidence only (335); clothing match (670); weapon evidence only (348); weapon match
(696); property (473); license (431) (Table 4, p. 27). The "scores" are added up and if the total exceeds
1000, then the case is determined to be solvable.

32

Others have found that the type of weapon to be associated with the social
distance between the victim and the offender. As the social distance became greater, the
use of personal weapons decreased, and the use of guns increased (Rinehart, 1994). This
information is valuable as it could assist investigators in limiting the field of likely
suspects. Even with the inconsistencies, it appears as if there is limited support for the
argument that presence and type of a weapon are related to case clearance, and the

relationship is negative when the weapon is a gun (Rinehart, 1994).
Previous Criminal Record

Surprisingly, there is little previous research that analyzed the relationship
between the previous criminal record of the victim or the offender and case clearance.
There might be numerous explanations for this, one of which might be that the data is not
readily available because the offender’s prior criminal record is not known in instances
where an offender has not been identified. Prior criminal record has, however, been
shown to be significantly related to overall decisions made by the criminal justice system.
The findings suggest that the fewer violent and property crime convictions a victim has,
the greater the odds of prosecution, and the more aggravated assault convictions an

offender has, the harsher the case disposition (Boris, 1979).

Like presence and type of weapon, it seems reasonable to assume that the
previous record of the offender could play a more prominent role in the solvability of

crimes if more cold searches were done on latent prints collected at crime scenes. As

33

previously stated, numerous studies found that many homicide offenders have previous
criminal records (see Chapter 2). A study conducted in Philadelphia revealed that useable
latent prints were lifted at 17.2 percent of all burglary scenes. Of those, 25 percent that
were fed into a system such as AF ISl6 were identified and matched to a suspect (Seamon,
1990, p. 49). Thus, it is plausible to argue that the usefulness of the offender’s criminal

record is related to the availability of automated information systems.
Felony Circumstances

The presence of felony circumstances has been shown to be clearly associated
with case clearance. Some argue this is the case because felony related incidents tend to
occur in settings where they are likely to be observed by others (Riedel and Rinehart,
1996). It has also been argued that felony circumstances plays such an important role in
case clearance because they are so intimately related to the victim-offender relationship
(Riedel and Rinehart, 1996). As previously stated, family and acquaintance relationships
tend to offer protection to victims fi'om felony-related homicides, although this traditional

protection does not seem as strong (Decker, 1996).

In nonfelony homicides, the proximity of the victim-offender relationship usually
means that people in the victim’s life such as family and friends can lend assistance to the

investigation although they might not have directly witnessed the crime. Taking this into

 

" Automated Fingerprint Identification System.

34

consideration, it was found that when relationship was removed, felony circumstances

developed as the most important predictor of case clearance (Rinehart, 1994) .
Victim-Offender Relationship

Similar to felony circumstances, there is substantial support for the significant
relationship between victim-offender relationship and clearance. It has been argued
detectives use a combination of victim-offender prior relationship and information
available from the crime scene to develop an interpretative scheme of the case, that is the
subjective framework that detectives use to assess the importance of a case (W aegel,
1982). Victim-offender relationship has consistently demonstrated to be significant to the
clearance of crimes (Bottomley and Pease, 1986; Rinehart, 1994). Homicides involving
families were had a strong positive relationship to homicide clearance, and homicides

involving strangers were had a strong negative relationship to clearance (Rinehart, 1994).

Limitations of the data used by Riedel and Rinehart (1996) resulted in the
researchers being unable to test this relationship because a large portion of open cases
used in the analysis were missing information on the victim-offender relationship.
Homicide data, especially offender information, is often missing from data files because

the information is usually submitted at the early part of the investigation.
Temporal Impacts

Month of incident does not appear to be correlated with case clearance. It does

appear, however, to be related to the victim-offender relationship. Stranger homicides are

35

more common in the summer, while family/acquaintance homicides were more common
in the winter. This is probably because there is greater interaction between strangers
during summer months. One researcher also found support for a marginal relationship

between the time of the incident and case clearance (Rinehart, 1994).
Spatial Characteristics

The result of research on the relationship between spatial characteristics, such as
whether the incident occurred in private or public, is mixed. Rinehart (1994) found that
no significant relationship with case clearance existed in any of the individual eleven
years from 1981-1991. Others, however, have found that if the crime occurs in public,
there is a greater likelihood that witnesses will be present, and/or the police will discover

the crime through strategies such as preventative patrol (Skogan and Antunes, 1979).
Physical Evidence

There is a general consensus that the presence of physical evidence is more
valuable in building a case against a suspect versus solving crimes. In their review of the
literature pertaining to the usefulness of physical evidence, Horvath and Meesig (1996)

made two primary conclusions:

"Because... [detective] training in and knowledge of physical evidence and
scientific analyses are limited, [they] tend to focus on the human aspects of
investigations - primarily interviews and interrogations over which they have the
most control. For these reasons, physical, scientifically analyzable evidence has
little inherent meaning to detectives, and it plays a very subordinate role in the
great majority of criminal investigations" (p. 965).

36

Although many researchers have found that latent prints can be raised from a
substantial number of crimes scenes (Horvath and Meesig, 1996; Seamon, 1990), only 1
to 2 percent of all burglaries are solved from the collection of latent prints (Chaiken et al,
1977). Even though the value of physical evidence such as latent prints is not clear,
several police departrnents‘7 use the presence of significant physical evidence as one of
twelve solvability factors used in a case screening model (Greenberg and Wassherman,

1970)

It appears as if a substantial amount of physical evidence is simply not collected
at crime scenes, even when crime scene technicians are sent (Chaiken etal., 1990). It has
been found that the collection and documentation of evidence is related to investigative
effort. It has been argued that personal characteristics of the victim might lead to a more
or less intensive initial investigation, and it is the intensity of the initial investigation that

usually leads to the discovery of physical evidence (Bynum et al., 1982).

Although the literature indicates that the presence of physical evidence has little to
no impact on the clearance of crimes, it does not appear as if this is not due to the
inherent quality of physical evidence. As stated above, physical evidence is simply not
collected at crime scenes. Even when the evidence is collected, it is rarely systematically

assessed. The lack of automated information systems in most departments makes cold

 

'7 For example, Rochester, NY, Oakland, CA and Montgomery County, MD.

37

searches of latent prints collected at crime scenes nearly impossible (Chaiken et al, 1977;
Eck, 1983; Horvath and Meesig, 1996; Seamon, 1990). Since, many offenders in
criminal homicides often have previous records, it is reasonable to assume that this

success rate could be higher.
Presence of Witnesses

There is general consensus that a relationship exists between the presence of
witnesses and case clearance. Third parties who provide information are one of several
key factors in the process of clearing murders (Riedel, 1995). Witnesses can take one of
several roles in the homicide incident. Witnesses tend to play one of five roles in
criminal incidents, depending on their level of involvement: surrogates; guardians;

facilitators; precipitators; or bystanders (Decker, 1995).

A "surrogate" is a witness who was either the initial target or offender at the
beginning of the incident, though in the final outcome was neither killed nor did the
killing. A "guardian" is an audience member who exerts some effort to intervene in the
situation. The "facilitator" is a witness who, in many cases, produced the circumstances
surrounding the incident, but who did not directly create the motive for the killing. The
"precipitator" differs from the " facilitator" in that they provide the motive through overt
encouragement. Finally, the "bystander" is not involved in the criminal incident at all

and generally is a mere observer (Decker, 1995).

38

Witnesses tend to provide the most pertinent information to investigations (Riedel
and Rinehart, 1994; Rinehart, 1994), and their presence tends to be the best predictor of
case clearance (Bloch and Bell, 1976; Bynum et al., 1982; Bottomley and Pease, 1986;
Greenwood et al., 1977). One study found that nearly 50 percent of the homicide cases
were solved due to information provided by a victim or witnesses (Greenwood et al.,

1 977)

Conclusions

Evidence on the relationship between victim/situational characteristics and the
clearance of crimes is mixed. While there are inconsistencies in the findings just
presented, there is also a thread of agreement. First, there is limited support for the
argument that victim demographic and social characteristics affect the outcome of an
investigation. The evidence seems most compelling for property crimes (Bynum et al.,

1982), yet there is some evidence that the relationship also exists for personal crimes
(Rinehart, 1994).

Although the effect of the victim or offender’s previous criminal record is not
clear, both have been found to relate to the dismsition of criminal cases (Boris, 1979).
Weapon also appears to be related to case clearance as cases that involve more personal
weapons (e.g., knife, hands/feet) tend to be solved more readily, and cases involving guns

more fi'equently remain unsolved. Felony circumstance emerges as one of the best

39

predictors of case clearance as nonfelony crimes are cleared more often than felony
related crimes. Victim-offender relationship is also a consistent and substantial predictor
of case clearance because crimes involving family or acquaintances are solved more often

than those involving strangers.

Findings have been presented that support the contention that case clearance is
related to temporal and spatial variables. Crimes that occur during the winter months are
marginally more likely to be solved, but this is likely to be related to factors such as the
victim-offender relationship since crimes during the summer months tend to involve
strangers. Also, since crimes that occur in public tend to have more witnesses, there is

support that this is marginally related to case clearance.

Of all the variables included in this review, the relationship between the presence
of physical evidence and case clearance is the most unclear. Most researchers have
argued that the presence and type of physical evidence ha little impact on case clearance
(Chaiken, 1977; Horvath and Meesig, 1996). This is disparate from the popular
impression held by most people who often believe "detective’s work...involves sleuthing
out clues in sordid underworld dens, [involves] deductive reasoning based on speedy and
accurate lab work and intuition, and nearly always [results in the apprehension of the]
criminal" (Eek, 1992, p. 19). The reality appears to be that physical evidence is rarely

collected, and when it is collected, it is usually only utilized to build a case against a

40

suspect once they have already been identified (Chaiken, 1977; Horvath and Meesig,

1996).

Finally, another consistent predictor of case clearance is the presence of witnesses
at the scene. These participants tend to supply a substantial portion of the information

which is utilized by investigators.

Chapter 4
METHODOLOGY

There are two primary purposes for this research. The first is to understand the
nature of homicide, and compare the results from the Detroit sample with those of
previous research. The second is to investigate the relationship between
victim/situational characteristics and the clearance of homicides. The analyses employ a
variety of statistical techniques to assess the relationships between variables. Descriptive
statistics are compared to the existing literature, and are then followed by bivariate and
multivariate analyses that assess the relationship between the victim/situational

characteristics and case clearance.
Location

The data were collected fi'om Detroit, Michigan during the summer of 1997, and
were part of a larger national study.18 There are several advantages, and disadvantages to
using information from one site. First, limiting the scope of the analysis to one site
reduces the random error caused by variations in available information sources, and how
incidents are documented between sites. In addition, there is less concern about
variations in social structural variables that could be correlated with clearance. One could

expect little variation in these variables to take place over a short period (Rinehart, 1994).

 

" The data from Detroit, MI were made available by the National Institute of Justice under a grant awarded
to the Justice Research and Statistics Association, Washington, DC. Sites included in the national study are
Detroit, MI; Milwaukee, WI; Los Angeles, CA; Baltimore, MD.

41

42

A major disadvantage to limiting the scope of this study to one site is with the issue of
generalizability. Generalizability refers to the ability to apply findings fi‘om one study to
different times and populations, and serves as a fundamental principle of social science
(Maxfield and Babbie, 1995, pp. 266-267). The purpose of social science is to apply
findings from one site to another site that has similar characteristics. It is possible that
local characteristics that do not exist in other populations, or exist in varying degrees,
make these findings applicable only to this one time and place, adding little to the overall
understanding of homicide clearances. Data obtained from the United States Census
Bureau suggest that the population of Detroit is not representative of the United States,
thus limiting generalizability. Detroit’s 1990 population of 1,027,974 has been relatively
stable in recent years, however, Figures 2 and 3 in Appendix A suggest that Detroit
differs from the United States and Michigan in both race and sex composition. The
figures suggest that Detroit’s population has a higher concentration of both women and
minorities than either the United States or Michigan. The higher concentration of
minorities in Detroit also indicates higher proverty levels in the city because minorities
tend to have lower socio-economic statuses. Figures 4 and 5 in Appendix A indicate that
this relationship does appear to be present. They indicate that Detroit’s residents are

more concentrated among the lower educational levels,19 and that on average, the mean

 

'9 Educational levels are for all adults over 25 years of age. Michigan had 5,842,642 residents over 25, and
Detroit consisted of 612,078 residents over 25 (United States Census Bureau, 1990).

43

household income level is lower (see Figures 1-4 in Appendix A). These characteristics
suggest that Detroit is not representative of the general population of either the United

States or Michigan.
Time Frame
Data were collected fiom homicides that occurred in the city of Detroit during
the years 1994 and 1995. While this short time period does not allow the reader to make

any longitudinal inferences, it provides a clear picture of the relationship between
solvability factors and clearance rates at one point in time. Homicide clearance
information for Detroit was available up to 1996, yet the scope of this research will be
limited to the calendar years of 1994 and 1995. Using the calendar years 1994 and 1995
are an advantage to this study because it can be argued that if a case was likely to be
cleared by arrest, the arrest would most probably occur in this two to three year follow-up
period. This window of opportunity will provides detectives ample time to prepare and
complete case files, and to prepare them for prosecution. This study will also control for
the most common criticism of the Supplemental Homicide Report (SI-IR) data. The
usefulness of SHR data has been questioned because the data are generally submitted at
the early part of an investigation and are rarely updated at a later date (Maxfield, 1989).

By using the police department’s case file as the primary source of information, the

44

present research uses the most current data that exists on each case included in this

analysis.
Sampling Frame

The Homicide Unit of the Detroit Police Departments has limited access to
automated information systems. Most of the information about detectives activities and
criminal incidents are manually stored. Automated information systems are limited to
criminal records and property management. This lack of automated information required
research staff to manually collect necessary data, and to verify these sources against

actual case documentation.

The Homicide Unit handles all incidents that involve suspicious deaths that occur
in the city of Detroit. A synopsis of each case, including the incident type (e.g., murder,
justified by police, justified by citizen), victim name, suspect name(s) (if available), and
status was recorded on an index card and filed. A member of the research team went
through each card and noted the internal case number and statuses of all cases that were
ruled to be founded criminal homicides. These classifications were usually a result of
information available from the initial investigation, and did not take into account changes
in case classification that could have possibly happened at a later date. This was not a
reliable source of information as departmental personnel had concerns that information

such as case status (open versus closed) was not systematically updated.

45

Due to these concerns, research staff concluded that it would be necessary to
verify case statuses in those instances where the case was recorded as open. It was
reasoned that there was little chance of a case changing fi'om closed to open, so it was
unnecessary to verify these cases. Research staff pulled the files for all open cases to
determine if an arrest had been made for the crime, and numerous misclassifications were
found. Additionally, it should be noted that Detroit initiates an investigation into each
homicide m. The Homicide Unit creates an internal case number for each victim. If
multiple homicide victims were present in a single incident, more than one internal case
number is created which ultimately refers to the same incident. These case numbers were
the basis for our sample selection. If multiple victims from the same incident fell into the
sample, the first victim selected was chosen and the others were excluded. This occurred
six times, and the population was resampled for six additional cases. The research design
called for the recording of demographic information for all victims who were involved in
the incidents that fell into the sample. The detailed situational variables were gathered
for the primary victim only. Thus, information from each individual crime incident was

captured only one time.

The total population of homicide cases for both 1994 and 1995 were combined,
making probability of selection equal across years. A proportional stratified sampling
technique was used to increase the sample’s representativeness in relation to the

proportion of cleared and uncleared cases. This would better ensure that the sample

46

chosen represented the actual population fiom which it is was drawn. Based on this
information, a mean clearance rate of 72 percent was established for the two combined
years. This clearance rate is slightly higher than the national average, probably because
the data was collected two to three years after the incidents occurred. Once a list was
created of all open and closed criminal homicides20 that occurred in the city, the number
of open and closed21 cases where chosen in proportion to their existence in the entire
population. Thus, 144 open cases, and 66 closed cases were chosen.22 The sample
represented nearly one-fifth of the total population of 1010 total homicide victims that

occurred during the 1994 and 1995 calendar years.

 

” It is probable that this sample included crimes involving less legal intent such as manslaughter. The data
focused on the original classification made by the department, and did not include other factors such as
charges filed by the prosecutor, and/or court disposition.

2' From this point forward, the term open will be used interchangeably with not cleared, and closed will
refer to cleared.

’2 When research staff verified case status by determining if an arrest was made, a very technical approach
was taken. Stafl‘ analyzed the cases to determine solely if an arrest was made, but did not attempt to
determine what the arrest was for. This could have entered error into the sample. There was a chance that
a person was arrested during the investigation process who was neither implicated in nor charged with the
crime. Common examples are those instances in which a hostile witness was arrested because he/she
refused to show up for questioning. If it was determined that a case was misclassified during the full
analysis of the incident, the file was left in the sample to limit the amount of systematic error. However,
the correct information such as actual case status was recorded. In a very small number of cases, although
an offender had been identified, no offender information was coded. This discrepancy only occurred in
those cases that were clearly in error as researchers did not attempt to question case documentation on a
systematic basis. An end result is that the proportion of open and closed cases is not completely consistent
with their proportion in the actual population.

47

Information Sources

An analysis was conducted on homicide files maintained by the Detroit Police
Department. Information used to investigate these crimes was gathered from a variety of
sources. Reviewers were instructed to read each file thoroughly because the department
does not use many forms that systematically gather pertinent information. One of the
primary sources of information for this study were Preliminary Complaint Reports (PCR)
filed by all officers who respond to a scene or call. PCR’s are the standard police report
officers complete when they respond to the scene of a reported crime, or take any official
police action. These reports include information such as the time a call was received, the
time the officers arrived at the scene, precinct where incident occurred, summary of the
incident, complainant identifying information, description of the scene, witnesses at the
scene, and suspect information. PCR’s also function as the starting point for all criminal
investigations. In addition, witness statements, Coroner’s reports, criminal history
reports, newspaper articles, and informal case documentation notes (memos, hand-written
notes pertaining to interviews, internal notes to detectives) were among the primary

sources of information.
Data Instrument

The data collection instrument was developed by the co-principal investigators of

each site. The instrument included questions about the victim, offender, location,

48

circumstances surrounding the incident, investigative effort, and presence of physical
evidence, among other attributes. Most questions were closed-ended and listed responses
that were exhaustive and mutually exclusive. The majority of the closed-ended questions
were either nominal or ordinal level responses. Where applicable, responses to these
closed-ended questions were supplemented by additional information, for example, the
caliber and exact use of a gun. Where closed-ended questions were not feasible (e. g., a
detailed list of the physical evidence collected at the scene), open-ended responses were

present (see Appendix B for specific questions that were used for the analyses).
Variables

The bivariate and multivariate analyses assessed the effect of numerous
independent variables on the dependent variable (e. g, whether the case was closed). Most
of the variables were nominal level. They include: victim sex, victim race, victim’s
previous criminal record, season of crime, shift of crime, spatial characteristics, the
presence of physical evidence, felony circumstance, drug circumstance, weapon, victim-
offender relationship, and the presence of witnesses at the scene. Victim culpability was

included as an ordinal level variable, and victim age has been coded as an interval level.

49

Case Status

The dependent variable, case closed, was measured by determining if an arrest
had been made for an incident. Case closed was operationalized as a binary variable (see

Table 1).
Victim Demographic Characteristics

Victim race consisted of two responses -- white and nonwhite. The nonwhite
category included those victims who were Black, American Indian or Alaskan Native,
and Asian or Pacific Islander. Due to the small sample size, it was necessary to create
these categories in order to have the appropriate numbers in each category for statistical

analyses.

For the bivariate analysis, an ordinal variable was created for victim age category
(see Table l for categories). Victim age was determined by finding the difference in
years between the victim’s date of birth and the date the incident occurred. The age
categories appeared to be conceptually sound. For the multivariate analysis, victim age

was used in its interval level form.

Three dummy variables were created to assess the relationship between victim’s
previous criminal record and case clearance. This study measured the affect of previous
criminal for drug crimes, violent crimes, and property crimes independently (see Table l

for coding).

50

Weapon was categorized into gun and no gun. Previous research has generally
categorized weapon into four specific categories: gun, knife/other cutting instrument,
blunt object, and personal weapons (hands/feet) (Riedel and Rinehart, 1996; Rinehart,
1994). Although it was more desirable to be specific as possible with the type of weapon
used, the sample size prohibited the use of these categories. It is also possible that in
some instances multiple types of weapons were used to inflict injury to the victim. The
this analysis focused on the weapon that was primarily responsible for the fatal injuries.
This information was ascertained from a combination of the description of the wounds,

witness statements, and the Coroner’s report.

Felony circumstance was dummy coded from multiple questions on the data
collection instrument (see "Felony Circumstances" in Appendix B). The instrument did
not directly ask if the criminal incident was felony related, so an operational definition
was created from the available data. The indicators of felony status are not exhaustive in
nature, and it is possible that all felony circumstances might not have been captured.
However, the indicators used cover the most common types of felony circumstances --
burglary, robbery, and sexual offenses. In addition, the question that determined if the
victim was sexually assaulted does not attempt to determine if the sexual assault was the
motile for the crime. The question merely records if a sexual assault took place, and this

applies felony status even if the sexual assault was incidental to the crime.

51

Table 1. Variable coding and preliminary descriptive statistics' (N=200).

11mm Coding Values N Mean SQ,

Case Closed Yes 1 138 .69 .46
No 0 62

Victim age categories 17 or younger 1 15 1.00
18-25 2 50
26-35 3 42
Over 35 4 57

Gender of victim Male 1 177 .89 .32
Female 0 23

Race of victim White 1 27 .14 .34
Nonwhite 0 173

Victim has previous drug record Yes 1 66 .33 .47
No 0 134

Victim has previous violent record Yes 1 81 .41 .49
No 0 l 19

Victim has previous property record Yes 1 85 .43 .50
No 0 115

Weapon Gun 1 160 .80 .40
No Gun 0 40

Incident involved felony circumstances Yes 1 39 .20 .40
No 0 161

Incident involved drug circumstances Yes 1 57 .29 .46
No 0 138

Victim-offender relationship Family 1 26 .71
Acquaintance 2 62
Stranger 3 36

Time incident report Midnight 1 51 .29 .45
Other 0 126

Spatial characteristics Private 1 100 .5 1 .50
Public 0 98

Physical evidence present at scene Yes 1 152 .78 .42
No 0 43

Witnesses present at scene Yes 1 85 .84 .37
No 0 105

Victim Culpability Group Low 1 73 .78
Medium 2 76
High 3 5 1

 

' Missing data are excluded.

52

This research also assessed the relationship between drug circumstance and case
clearance. Previous research has indicated that the drug epidemic, especially the
introduction of crack cocaine, in our nation’s inner cities has changed the nature of crime
(Decker, 1996). The data collection instrument included a question that asked if the
homicide was drug-related. The incident was determined to be drug-related PM if the

sale, distribution, or use of narcotics was the precipitating factor in the incident.

Missing victim-offender relationship information posed a problem, as it did in
previous research (see Riedel and Rinehart, 1996). For familial relationships, the data
collection instrument collects a more specific level, such as sibling, parent, wife, or
husband. However, for purposes of this paper they will all be combined into one group.
There have been variations about how victim-offender relationship was coded in previous
studies. Some researchers dichotomized victim-offender relationship into stranger or
nonstranger groups (see Block, 1981 and Riedel, 1981), and others have argued that more
precise levels of measurement allow the reader to better understand how these
relationships affect other characteristics of homicide (Decker, 1995). More recent studies
have argued for the addition of the third category of acquaintances (Decker, 1995; Riedel
and Przybylski, 1993; Riedel and Rinehart, 1996; Rinehart, 1994). For this analysis, the
victim-offender relationship was coded as family, acquaintance, and stranger (see Table

1).

53

Season and shift during which the incident was reported are the two temporal
characteristics included. This analysis distinguished the relationship between crimes that
were reported between September and February, versus those that occurred between
March and August (see Table 1). These groups were created because mean clearance
rates for the months were very similar. Shift was operationalized to measure the
relationship between crimes that were reported during the midnight shift (12 am. - 8
am), and those reported on any other shift (see Table l). Shift was originally broken
into three different groups. The final analysis included midnight shift versus non-
midnight shift groups because the mean clearance rates for the other two shifts were

clearly similar, and significantly different from the first.

Spatial characteristic of the criminal incident was operationalized into private
versus public setting. Due to the limited support for the supposition that crimes occurring
in public are more likely to be solved, a positive relationship with the dependent variable
was hypothesized. Previous studies that used SHR or Victim Level Murder data (VLM)
were unable to make this determination because the information was not readily

available.

Previous research found victims often precipitated the violence that lead to their
final demise (Levi, 1980; Wolfgang, 1958; Wolfgang, 1967), yet none of these

researchers have appropriately quantified victim culpability. While it was difficult to

54

quantify the interaction that takes place between the victim and the offender, indicators
contained in the instrument made it possible to measure general victim culpability
characteristics. It has been argued detectives often consider factors such as the victim’s
lifestyle and social characteristics as important in their interpretation of the level of

necessary investigative effort (Waegel, 1982).

The data collection instrument provided information concerning the victim’s
lifestyle that were used to determine their level of culpability, and a dummy variable
called "victim culpability score" was created fi'om these indicators (see "Victim
Culpability in Appendix B). Like all other indicators, information fi'om detective notes,
witness statements, crime scene photographs, evidence technician reports, and Coroner
reports were used to obtain the details. For the bivariate analysis, the variable was
grouped into three categories at the 33rd and 66"1 percentiles (see Table 1). The ratio
measure of culpability was used for the multivariate analyses. Since a review of the
literature did not yield an operational definition for this concept, comparisons are not
possible. It is hypothesized that as the victim culpability score goes up, case clearance

will decrease.

Evidence technician reports were the primary source for information about the
presence of physical evidence (see "Physical Evidence in Appendix B, Table l for

coding). The reader should exercise caution when interpreting the findings associated

55

with this variable since the analysis did not attempt to determine the usefulness of the
physical evidence collected, merely its presence. A positive relationship is hypothesized

between this variable and case closure.

Finally, the analysis will determine if a relationship exists between presence of
witnesses and clearance (see Table 1). Similar to the caution about interpreting the
results of information pertaining to the presence of physical evidence, this information
does not indicate if the witnesses provided any valuable information. It is hypothesized

that a positive correlation will exist between the two variables and the dependent variable.
STATISTICAL TECHNIQUES
Descriptive Summary

The first section of Chapter 5 will presents a descriptive summary of the
population selected for the sample, including information on victims and offenders by
sex, race, and age. In addition, information on gun crimes, felony circumstances, victim-
offender relationships, temporal characteristics, spatial characteristics, physical evidence
present at the scene, and presence of witnesses will be furnished. This will allow for a
descriptive analysis of the nature of homicide in the city of Detroit for the years 1994-

1995, and will make comparison to previous findings possible.

The first level of analysis will be a bivariate analysis between each of the

independent variables and the dependent variable "case closed." A table will be

56

presented that identifies each categories mean clearance rates for each independent
variable. The bivariate analysis will include only categorical and interval level variables.
Analysis of variance will be used to determine if a significant relationship exists between
each independent variable and clearance. The analysis of variance will be used to
determine the mean clearance rate for each independent variable category, and determine
if the mean clearance rates for the subcategories are significantly different from each

other.

After the findings from the bivariate analysis have been presented, multivariate
analysis will be conducted. Logistic regression will be used to measure the relationship
between each independent variable and the dependent variable, while controlling for all
other variables in the model. Logistic regression is preferred over ordinary least squares
regression when the dependent variable is dichotomous. A true linear relationship cannot
exist with a dichotomous dependent variable, and the subsequent s-shaped regression line
violations one of the major assumptions of linear regression (Bachman and Paternoster,

1997, pp. 567-572).

Two models will be constructed, one with all of the independent variables, and the
other excluding victim-offender relationship information. Victim-offender relationship

will be excluded from one model because of missing data. The missing data results in

57

substantially more cases being excluded from the analysis due to missing data. Due to

the small sample size, it is useful to include as many cases as possible.

Chapter 5
RESULTS

The city of Detroit achieved an overall clearance rate of 69 percent during the
period of 1994-1995; this figure exceeds the national average of 64 percent (United States
Department of Justice, 1995). The two to three year difference from when the crimes
occurred to when the data were collected could be at least partly responsible for the
increased clearance rate by providing more time to identify and apprehend a suspect.
Most crime data submitted to the Federal Bureau of Investigation is done so on an annual
basis, and this information is used to calculate clearance rates. If a case was cleared
during another calendar year, the clearance statistic was applied to the year it was cleared.
Thus, clearance rates do not necessarily reflect how many crimes that occurred during a
given year were cleared by arrest, but merely how many crimes were cleared by arrest
during the period. Collecting the data 2 to 3 years after the initial criminal incident
allowed the researchers to determine a more accurate clearance rate for criminal

homicides that occurred during the calendar years 1994-1995.

Table 2 suggests that both victim and offender demographic characteristics found
in the Detroit sample are generally consistent with the findings of previous research. The
typical homicide victim can be characterized as a younger black male who was killed
with a gun. This representation of young black males among homicide victims is slightly

higher than previous research. The age patterns found in the present sample was

58

59

Table 2. Additional descriptive statistics for sample (N =200)

Age
Median Victim Age 28 years
Missing Victim Age 15.5%
Median Offender Age 24 years
Missing Offender Age 36.5%
§Q§
Male Victim 88.5%
Male Offender 88.5%
Missing Offender Sex 18.5%
Lacs.
Nonwhite Victims 86.5%
Nonwhite Offenders 75.0%
Missing Offender Race 18.5%
thnginmtim
Victims w/Drug Record 33.0%
Victims wNiolent Record 40.5%
Victim w/Property Record 42.5%
Gun Incidents 80.0%
Felony Circumstances 19.5%
Drug Circumstances 28.5%
Missing Drug Circum. 2.5%
Victim-Offender Relationship
Stranger 18.0%
Acquaintance 31.0%
Family 13.0%
Missing Relationship 38.0%

Environmental Characteristics

Midnight Shift (0001-0800)
Missing Time

Season of Incident
September-February
Missing Season

Weekday of Incident
Weekend (F ri-Sun)
Missing Weekday

Spatial Characteristics
Private
Missing Spatial

Evidence
Physical Evidence Present

Missing Evidence

Witnesses
Witnesses Present At Scene

Victim Qulpabilifl Scores
Low (0-1)
Medium (2-5)
High (6-13)

28.8%
11.5%

42.5%
5.0%
45.0%
2.5%
50.0%
1.0%

78.0%
2.5%

84.0%

36.5%
38.0%
25.5%

 

 

60

consistent with previous research. The distribution for age was positively skewed, and the

median victim age was 28 years.

Studies that attempt to systematically analyze the relationship between
victim/offender characteristics and case solvability suffer from one primary shortcoming;
offender information is often missing. Of all the offender demographic variables included
in this analysis, offender age was most often missing. The reasons seems to be because
offender race and sex were more easily obtained from sources such as witness testimony,
but if the offender was neither positively identified nor apprehended, exact age was difficult
to establish. In the instances where offender information was known, parallel findings were
evident between victim and offender characteristics. Similar to victims, homicide offenders
also tended to be younger black males. As expected from previous findings, homicide

offenders tended to be slightly younger than homicide victims (24 years versus 28 years).

The reader should exercise caution when interpreting these results because the racial
composition of Detroit is not representative of either the United States or Michigan (see
Figures 2 and 3 in Appendix A). The Detroit homicide data revealed victimization and
offending levels closely mirror the actual proportion of blacks in the total population (Table
2). This is a significant factor because blacks represented only 12 percent of the total US.
population, and 14 percent of Michigan’s total population in 1990 (United States Census

Bureau, 1990). Thus, blacks are significantly overrepresented in homicide victimization

 

61

and offending rates in both the United States and Michigan, while only slightly

overrepresented in Detroit.

Drug circumstances were more common in homicide incidents than were felony
circumstances. Incidents were defined as involving drug circumstances if the primary
motivation for the incident was the sale, distribution, or use of illegal narcotics. In some
cases drugs might have been present at the scene, but were not a contributing factor to the
incident. Similarly, there were incidents where drugs were not present at the scene, yet
circumstances surrounding the fatal altercation revolved around the drug trade. Of those
homicides with a known motive, slightly less than 20 percent involved felony
circumstances. It is hypothesized that this figure would be higher if information
concerning uncleared homicides was more readily available. Research suggests incidents
with missing victim-offender relationship information involve strangers (Williams and
F lewelling, 1987), and homicides involving strangers tend to involve a higher
concentration of felony circumstances (Cardarelli and Cavanagh, 1992; Decker, 1995;

Decker, 1996; Riedel and Rinehart, 1996; Rinehart, 1994; Wolfgang, 1958).

A problem with missing data was also evident with victim-offender relationship.
Victim-offender relationship was unknown in 38 percent of all incidents, even though only

31 percent of all cases were not cleared by arrest.23 The percentage of cases missing

 

2’ Reason would suggest that the victim-offender relationship should be known for all cleared cases. It is
not altogether apparent why the percentage of missing victim-offender relationship is higher than the

62

information on the victim-offender relationship is higher than was found in previous
research (Williams and Flewelling, 1987), especially when considering the one to two year
window in gathering the information. It is unlikely that missing victim-offender
relationship information is evenly distributed across each category because these incidents
are more likely to involve strangers (Williams and Flewelling, 1987). However, when the
relationship was known, the most frequent relationship involved acquaintances, followed

by strangers, and family.

There was limited consistency between patterns in temporal characteristics and
previous findings. Nearly half of all homicide incidents occurred between the hours of 4:00
pm. and midnight. While there was a correlation between time of day and homicides, no
significant pattern existed for season or weekday of incident. The lack of a clear
relationship for season is surprising, especially when one considers the extreme weather
patterns that exist in Michigan. Although it was hypothesized homicides would be
significantly higher in the spring and summer because personal contact is more frequent,

this does not appear to be the case in the Detroit sample.

Both physical evidence and witnesses were present at the vast majority of homicide
scenes. It is difficult to determine how useful physical evidence and witnesses were

because their sheer presence reveals nothing about their ability to identify a suspect, or to

 

percentage of uncleared cases. Possible reasons might include error on the part of the data collection
personnel, or the information was simply missing from the case folder.

63

add other pertinent information to the investigation. Spatial characteristics can affect the
presence of witnesses, but this relationship did not appear to be the case. Findings suggest
an almost equal number of witnesses in both public (85 percent) and private (82 percent)

settings.

Finally, nearly 65 percent of all victims scored either a medium (38 percent) or high
(26 percent) on the victim culpability scale. Researchers have suggested that victims
sometime play an active role in their own demise (Curtis, 1974; Levi, 1980; Lerner, 1971;
McDonald, 1976; Shultz, 1967; Silverrnan, 1974; Von Hentig, 1948; Wolfgang, 1958;
Wolfgang, 1967), and the label "victim" or "offender" are mere roles assigned after the
discovery of a crime. Since it is sometimes difficult to distinguish the victim’s actions
from the offender’s, the victim/offender labels do not always adequately capture the true
sense of who was the aggressor. It has been suggested that the victim’s lifestyle (e.g., drug
abuser, drug dealer) can have an affect on the perceived level of necessary investigative
effort, especially when no suspect has been initially identified (Bynum et. al, 1982; Waegel,
1982). The variable attempted to measure the level of victim culpability that could be

perceived by investigators.
Findings
Age was the only victim demographic variable that was significantly related to

clearance (Table 3). The relationship between victim age and clearance was nonlinear.

64

Clearance rates for victims in the 26-35 age group was lowest, while clearance rates were
highest for victims 17 years and younger. Of the three variables measuring the victim’s
previous criminal record, the victim’s previous violent criminal history was the only
significant variable. Incidents involving victims with previous violent criminal histories

were less likely to be cleared.

Four situational variables were significantly related to the dependent variable.
Crimes involving drug circumstance were significantly less likely to be cleared by arrest,
and those involving drug circumstances were also less likely to be cleared. Shift was also
significant, and those that were reported during the midnight shift were less likely to be
cleared by arrest. Crimes that occurred in private settings, such as a house or other building
were more likely to be cleared, and those where eyewitnesses were present at the scene,

were more likely to be cleared by arrest.

No other victim level variables were significantly related to clearance. The number
of homicides was almost evenly distributed among males and females, and whites and
nonwhites. Although information about victims’ previous drug or property crime records
did not affect clearance in the bivariate findings, results suggest that the general trend was
for clearance rates to be lower for victims with prior records. Table 3 suggests that
although victim culpability was not significantly related to case clearance, there was a trend

in the clearance rates. Clearance rates consistently decreased as the victim culpability score

 

65

Table 3. Percentage of cases cleared by arrest, by categories of independent variables‘.

 

Age of Victim’ Incidents Involving Drug Circumstances"
17 or younger 87.0 Yes 54.0
18-25 78.0 No 75.0
26-35 52.0
Over 35 70.0 Victim-Offender Relationship
Family 96.0
Gender of Victim Acquaintance 90.0
Male 69.0 Stranger 86.0
Female 70.0
Time Crime Reported“l
Race (Victim) Midnight Shift 59.0
White 70.0 Other Shift 78.0
Nonwhite 69.0
Season
Victims w/Drug Record Sept-Feb 65.0
Yes 64.0 Mar-Aug 72.0
No 72.0
Day of Week
Victims wNiolent Record“ Fri-Sun 71.0
Yes 60.0 Mon-Thurs 67.0
No 7 5.0
Spatial Characteristic"
Victim w/Property Record Private 80.0
Yes 66.0 Public 59.0
No 71.0
Physical Evidence Present
Weapon Yes 70.0
Gun 69.0 No 63.0
Other 70.0
Witnesses Present At Scene“
Incidents Involving Felony Circumstances Yes 72.0
Yes 69.0 No 53.0
No 69.0
Victim Culpability Scores
Low 77.0
Medium 67.0
High 61.0

 

 

'Analysis of Variance was used to determine significant differences in this table.

* p< .05
"p<.01

66

increased, indicating a negative relationship between the two variables. It was originally

hypothesized that crimes involving victims with higher culpability scores would be less

likely to be cleared. Based on the just world theory aVIcDonald, 1976), it was hypothesized

that detectives would perceive more culpable victims as less worthy of intensive

investigative effort because of the perceived relationship between their lifestyle and death.

The findings suggest that the victim’s lifestyle did not diminish the investigative effort of E

Detroit’s detectives.

There was no significant difference between the clearance rates for crimes that

 

involved the use of a gun and those that involved another type of weapon. Unlike the
significant relationship between drug circumstances and cases clearance, no such
relationship developed between felony circumstances and clearance. The time the incident
was reported, and season crime occurred were not significantly related to case clearance.
However, the data reveal that crimes that occurred during the midnight shift, and during the

fall or winter months were less likely to be cleared.

Several of the above relationships are inconsistent with the general findings of
previous research. Previous findings have generally not found victim demographic
variables to be significantly related to case clearance. The significance relationship
between victim age and clearance is contrary to previous findings (Bynum et al., 1982,

Riedel and Rinehart, 1996; Rinehart, 1994). In addition, although victim sex and race have

67

previously been shown to be marginally significant for isolated years (Rinehart, 1994), no

such relationship existed in the bivariate analysis.

Felony circumstances has generally been regarded as a strong predictor of case
clearance (Riedel and Rinehart, 1996).26 However, the clearance rate for felony versus
nonfelony crimes were the same (Table 3). It is possible the operational definition for
felony circumstances used here differed from that used by previous researchers. In the
present analysis, cases were identified as involving felony circumstances only if there was a
clear indication that a concurrent felony was being committed at the time of the homicide
(e.g., rape, robbery, burglary). Therefore, there might be a relationship between missing
circumstance information and the felony circumstance variable, suggesting that the
proportion of felony crimes might be higher or lower if all circumstances were known. It is

also possible that different criteria was used by other studies.

There is contradictory evidence surrounding the significance of the type of weapon
used in the crime. Prior research on the relationship between type of weapon and clearance
is mixed, with some finding presence and type of weapon significantly related to case
clearance (Cardarelli, 1993; Cardarelli and Cavanagh, 1992), and others finding otherwise
(Riedel and Rinehart, 1996). This analysis did not reveal a significant relationship. It

could be argued the relationship was influenced by the fact that the vast majority of

 

24Riedel and Rinehart (1996) found that 42.4% of felony homicides were cleared while 89.6% of nonfelony
homicides were cleared.

68

incidents in Detroit involved guns (Table 2). The limited variability in the type of weapon

could make it difficult to discern this relationship.

Another finding that was inconsistent with previous findings was the relationship
between victim-offender relationship and case clearance. Previously, victim-offender
relationship was found to be strongly correlated with case clearance (Bottomley and Pease,
1986; Rinehart, 1994). Missing information is a commonly cited problem when analyzing
the impact of this variable, and this also proved to be a problem in the current analysis. It

was hypothesized that the amount of missing victim-offender relationship data was related

 

to felony and/or drug circumstances, yet zero order correlations indicated otherwise.
Victim-offender relationship was not significantly correlated with drug circumstances, or to

felony circumstances.

There was limited consistency between the findings from the bivariate analysis and
the logistic regression results shown in Table 4. Logistic regression was chosen for the
multivariate analysis for two reasons. First, logistic regression is preferred over ordinary
least squares regression when the dependent variable is dichotomous. Second, logistic
regression is used when a nonlinear relationship is hypothesized, meaning that the
independent variables are not normally distributed in the population. This implies the
linear regression requirement of homoscedasticity is not met, and the error terms do not

have constant variance (Bachman and Paternoster, 1997).

69

Of the variables that measured victim’s previous criminal record (property, violent,
drug), only victim with violent criminal record was included because of multicollinearity.27
A correlation matrix indicated that the three variables were highly correlated, so violent
record was included because it has the most significant impact on the dependent variable.
Finally, victim age and victim culpability were used in their original ratio level form versus
the recoded groups that were used for the bivariate analysis. This was done in order to !

measure the optimal amount of possible significance.

The first regression model (Table 4) suggests that, net of all other variables, there

 

were no significant relationships between any of the independent variables and case
clearance. Incidents involving drug circumstances and incidents that occurred in private
settings were close to reaching the .05 significance level, suggesting that they might be
related to case clearance. The direction of the relationship between victim culpability score
and case clearance changes direction in the regression model. This indicates that one or
more independent variables were intervening in the original bivariate relationship. Zero
order correlations indicate that victim culpability is significantly correlated with drug
circumstances (.523), violent criminal history (.412), and felony circumstances (.252). This

suggests that incidents involving victims with higher culpability scores were more likely to

 

2’ An assumption of multiple regression is that the independent variables should be independent of each
other, and therefore uncorrelated (Bachman and Paternoster, 1997, p.491). Multicollinearity is a problem
in multivariate analysis when two independent variables are highly correlated with each other. While
relatively small correlation is acceptable, a correlation above approximately .7 is problematic (Ott, Larson,
Rexroat, and Mendenhall, 1992, p.530).

70

Table 4. Logistic Regression coefficients, significance levels, and odds ratio. All variables
included in model.

Vgiablg b Si, Odds

flag,
Victim Age 0.14 1.00 1.15
Victim Male -11.84 56.58 .00
Victim White 8.39 57.77 4424.39
Victim had Violent Record -1.79 1.40 .17
Gun Crime -7.03 52.98 .00
Felony Circumstances -0.64 1.64 .53

 

Private Location 3.13 1.64 22.93
Evidence Present 1.09 1.0 2.96
Witness Present 0.44 1.45 1.55
Midnight Shift -2.00 1.25 .13
September-February Incident -1.77 1.14 .17
Weekend Incident -0.56 .89 .57
Drug Circumstance -2.39 1.30 .09
Victim Culpability 0.32 .25 1.37
Victim-Offender Strangers -0.13 1.10 .88
N 120

Modified R2 .38

*Significance level <.05

 

involve drug and felony circumstances. Also, the victim was more likely to have a previous
criminal history for violent crimes. The direction also changed between victim age and
case clearance, once again suggesting that one or more variables were intervening in the

original relationship. Victim age was significantly correlated with victim race (.231), and

71

weapon (-.233). This suggests that black victims tended to be younger, and the crimes

tended to involve guns.

The results in Table 4 are weak due to missing data. The large standard errors for
victim sex, victim race and weapon are disproportionately high. The inflated odds ratio for
victim sex makes the problem more evident, indicating a serious problem with the model. If
Although the percentages of excluded cases for nonwhites (38 percent) and whites (41

percent) are similar, only 16 cases involving white victims are left in the analysis when i
i

 

victim-offender relationship is included. The small number of cases included in the i

analysis is primarily responsible for the questionable results.

A second logistic regression model was run using the same information included in
Table 4, except the victim-offender relationship was excluded from the analysis (Table 5).
It is hypothesized that one reason for the lack of any significant relationships in Table 4 was
because a large proportion cases were excluded from the analysis due to missing data.
When victim-offender relationship was included, 80 cases were excluded due to missing
data. However, when the victim-offender information was left out, only 39 cases are
excluded. It can be argued that including more cases in the analysis will make the analysis

more accurate.

When victim-offender relationship was excluded fi'om the analysis, five variables
were significantly related to clearance (Table 5). The data indicated that victims with

previous violent criminal records were significantly less likely to have their cases cleared.

72

Table 5. Logistic Regression coefficients, significance levels, and odds ratio. Victim-ofl‘ender
relationshijLinformation excluded from model.

 

 

 

 

V 'a le b Si, Odds
Ram
Victim Age -.01 .02 .99
Victim Male -.48 .95 .62
Victim White -.08 .73 .92
Victim had Violent Record -1.45** .48 .23
Gun Crime -.22 .70 1.24
Felony Circumstances -.10 .62 .90
Private Location 1.21 * .47 3.37
Evidence Present .72 .46 2.06
Witness Present -.65 .64 .52
Midnight Shift -.94* .46 .39
September-February Incident -.80 .41 .45
Weekend Incident -.04 .43 0.04
Drug Circumstance -1.86" .54 .16
Victim Culpability 0.25“ .10 1.28
N 161
Modified R2 .20
*Significance level <.05

A surprising finding was that victims with higher culpability scores were more likely to
have their cases solved. This finding contradicts the speculation of previous researchers
who suggested that victim’s lifestyle and social characteristics would negatively impact

investigative effort (Waegel, 1982).

73

Cases that involved victims with violent criminal records were significantly less
likely to be cleared by arrest. Crimes that occurred during the midnight shift, in public
settings, and involved drug circumstances were less also significantly less likely to be
cleared. Zero-order analysis indicates a significant correlation between drug circumstances
and three other independent variables (victim with previous violent record, felony
circumstances, and victim culpability). All three correlations are positive, and indicate that
drug circumstances are more likely to involve victims with violent criminal histories,
involve felony circumstances, and involve more culpable victims. The strongest
correlations are for victim culpability (.523) and felony circumstances (.310), yet are still
significant for victim’s violent criminal history (.159). Although incidents that were
reported during the fall or winter did not have a significant relationship to case clearance,

the significance level of .05 suggest that it was closely related to clearance.

Of all of the variables included in the analysis, location, evidence present, weekend
incident, and victim culpability were the only variables that were positively associated with
case closure. The direction of the relationship between victim race and felony circumstance
were opposite in the logistic regression versus the bivariate analysis. This could suggest

that one or more of the other variables somehow intervened in the original relationship.

The odds ratio in Table 5 indicates that certain variables have a greater impact on

the dependent variable. For example, the location of the homicide incident was one of the

74

better predictors of case clearance. Homicide incidents that occurred in private were over
three times has likely to be cleared compared to those that occurred in private.28 There are
several possible reasons for this relationship. The negative, and significant relationship
between spatial characteristics and weapon suggest crimes that public homicides were
significantly more likely to involve guns. In addition, they were significantly more likely to
involve felony circumstances, and occur during the midnight shift. Thus, it appears as if
weapon, felony circumstances, and time the incident was reported were intervening

variables.

Similar to linear regression, logistic regression provides a means to measure the
overall goodness of fit for the regression model. Like the R-square used in linear
regression, the modified R-square essentially indicates the reduction in error of prediction.
The measure is calculated by dividing the model chi-square by the log likelihood, and
multiplying the 100 (Menard, 1995, pp. 19-22). This statistic helps to determine the

amount of variance explained by the model (see Menard, 1995).

The modified R-square for model one (victim-offender strangers included)
suggests 38 percent of the variance is explained by the model. For model two, 20 percent
of the variance is explained by the model. However, a larger number of cases were

excluded from the first regression analysis because of missing data. The effect of these

 

26The above model was tested for multicollinearity using a standard test such as the VIF collinearity
statistic. The test indicated multicollinearity was not a problem with any of the models in this analysis.

75

missing data is evident in the inflated standard errors for victim sex, victim race, and

weapon. In addition, the weakness of the model was also evident in the inflated odds ratio

for victim sex. Thus, although the modified R-square for the second model was smaller

than the first, yet the lower standard errors suggest the model is better. In addition, none of

the variables in the first regression model were statistically significant, whereas several of E

the variables in Table 5 were statistically significant.

 

Chapter 6
CONCLUSIONS

This research was undertaken in response to steadily decreasing homicide clearance
rates since the 19603. Federal, state, and local governments have funneled tens of billions

of dollars into the criminal justice system, yet is still has to be established if those resources

 

were effective in assisting agencies in achieving their missions. Although there is some '
disagreement about the usefulness of using clearance rates as a measure of police

effectiveness, it is still one of the most commonly used methods. If one were to use

clearance rate as a measure of effectiveness, then it would be reasonable to assume that the l

resources allocated to police agencies have been less than effective. Official police
statistics indicate that homicide clearance rates have decreased 33 percent since the 19605
(Cardarelli and Cavanagh, 1992, p. 2), and homicide clearance rates tend to be substantially

higher than any other crime (see Figure 1 in Appendix A).

It is also reasonable to hypothesize that the nature of the crime itself may have
changed, and that this change is somehow related to decreasing clearance rates. However,
there is only limited support for this conclusion. These above findings suggest that the
patterns and comprise the typical homicide incident are relatively consistent with previous
findings. For example, the descriptive statistics presented in Chapter 5 suggest that the
victim, offender, and situational characteristics found in the previous literature are still

consistent with the nature of crime today. The typical homicide victim in Detroit was a

76

77

young black male who was killed with a gun. Just over half of all victims had at a previous

criminal record, and most were killed by someone that they knew. These findings should

be tempered with the fact that Detroit’s population is not representative of either the United

States or Michigan. Data presented in Figures 2 to 5 in Appendix A suggest that Detroit’s

population consists of a larger proportion of minorities and women. Also, economic

indicators (as measured by mean education and income level) suggest that there is a higher

concentration of poverty in the City of Detroit.

 

These demographic composition of Detroit is significant for two reasons. As will i
be discussed below, since Detroit’s population is not representative, it is difficult to
generalize findings from this research to other times and places. Second, although
nonwhites are still proportionately overrepresented among homicide victims, they are
substantially closer to their actual representation in the general population than in previous

findings.29

Although many characteristics of homicide have remained stable over time, it also
seems apparent that homicide has changed. One major influencing factor that has affected
the nature of criminal homicides seems to be the large scale manufacturing, trafficking,

selling, and use of illegal narcotics in our nation’s inner cities. Data revealed that nearly 30

 

2"Victimization rates for blacks found in Wolfgang’s (1958) research were similar to those in the Detroit
sample. However, at the time of Wolfgang’s study, blacks comprised only 12 percent of the total
population, thus they were significantly more overrepresented among homicide victims.

78

percent of all homicide incidents were motivated by drug circumstances (see Table 2 in
Chapter 5). It has been suggested that drugs, especially the crack epidemic, have
diminished many of the protective elements in our inner cities (Decker, 1996). This figure
seems even higher if one considers that a strict definition was used in this study for drug-
related homicides. A case was determined to be drug-related only if drugs were a

precipitating factor in the homicide incident. A possible explanation for the drug-violence E
nexus is that drug-related violence often involves young, heavily-armed males. It is argued

many of these heavily-armed youth tend to lack the ability to mediate conflict without

 

violence. The illegal drug industry has actively recruited young males, and this trend does
not seem to be decreasing (Blumstein and Heinz III, 1995). It seems that this is an

important area to further explore.

The bivariate analysis revealed that victim age was significantly related to case
clearance, and that younger victims were more likely to have their cases cleared. Incidents
involving victims with a previous criminal record were less likely to have their cases
solved. Previous research generally found that victim level variables did not significantly
affect clearance. These two variables are important because previous research found victim
level variables not to be significant predictors of case clearance (Bynum et al., 1982; Riedel

and Rinehart, 1996).

79

Two variables that were previously shown to be important predictors of case
clearance, felony circumstances and weapon, demonstrated no relationship to the dependent
variable. With felony circumstances, it is reasonable to assume that missing data or
different coding criteria affected the relationship. An incident was recorded as involving
felony circumstances only if there was a clear indication that another felony was the
motivation for the incident. Those instances, for example, where an offender might have
taken property after or during the commission of the crime, but the theft was not the
motivation for the crime, but only an afterthought, were not considered felony
circumstances because the theft was not the motivation for the crime. Thus, the different
criteria might have excluded cases that other researchers would have included. Also, more
cases might possibly have been coded as felony circumstances if more information was
available for open cases. Although felony circumstances was not significant, drug
circumstances were significantly related to case clearance. Zero order findings also
indicate a moderate (.310) correlation between felony circumstances and drug
circumstances. This indicates that incidents involving drug circumstances were also more

likely to involve felony circumstances.

It was hypothesized previously that spatial characteristics would be correlated with
clearance rates, suggesting that crimes in public would be more readily solved because

more witnesses would be present (Skogan and Antunes, 1979). The variable was

 

80

significant, though crimes occurring in private settings were significantly more likely to be
cleared. One possible explanation might be that these crimes were more likely to involve
family and fiiends. The data support this hypothesis. Nearly 40 percent of homicides that
occurred in private settings involved family members, compared to only 4 percent that

occurred in public settings. Similarly, nearly 40 percent of crimes that occurred in public

 

Ir
I
involved strangers, compared to slightly more than 20 percent of those that occurred in i
private settings. The Cramer’s V value of .39 indicates the two variables are moderately
correlated. i

Victim-offender relationship did not prove to be a significant predictor of clearance.
It is possible that the variable was insignificant because of missing data. It was difficult to
discern the victim-offender relationship for those incidents where an offender was not
identified. In a limited number of cases the information was available for open cases from
other sources such as information provided by witnesses. Finally, the presence of physical
evidence was not significantly related to case clearance. It is hypothesized that the
presence of physical evidence is potentially more valuable than the data indicate. The
usefulness of evidence is mediated by how the evidence is analyzed and integrated into the
investigation process. Technological advancements, such as Automated Fingerprint
Identification Systems (AF IS), have the potential for providing more assistance in the

clearance of cases. Unless these advancements are integrated into the investigation process,

81

physical evidence will continue to be useful only in preparing a case against an offender

that has already been identified.

When victim-offender relationship was included in the logistic regression model,
none of the independent variables were significantly related to clearance. Part of the
explanation for this might be that including the variable resulted in substantially more cases .-
being excluded because of missing data. When the variable was excluded, five variables

were significantly related with clearance. Consistent with the bivariate analysis, incidents

 

involving victims with previous violent records, and those that occurred during the

midnight shift were less likely to be cleared by arrest. However, a new development was

that crimes occurring during the midnight shift were significantly less likely to be cleared.

In addition to the above findings, two of the most significant findings are those that
are original to this piece of research. When controlling for all other variables, drug
circumstances remained a very significant predictor of case cleared. The consistency in this
relationship gives support to Decker’s (1996) conclusion that the drug trade is at least
partly responsible for decreasing clearance rates. The argument is that primary
relationships (e.g., family, or in some cases good friends) traditionally protected victims
from being murdered by family members for instrumental reasons (involving some sense of
gain such as financial reward). However, the drug trade has broken the protective nature of

these relationships (Decker, 1996). Furthermore, the correlation between drug

82

circumstances and felony circumstances adds an element to the discussion that has been

previously neglected.
Limitations of Findings

The major limitation in interpreting the above information concerns

 

generalizability. It is highly likely that the findings are not generalizable to all times and r
places. The primary reason for this concern steams from the information provided in ’2
Figures 2-5 in Appendix A that suggests Detroit is not representative of the population of

Michigan or the United States. However, one advantage to interpreting these results is that ‘I

the data, although technically secondary, were collected by the author and several of his
colleagues. It is argued that there is a high level of reliability and validity to the responses.
This author was present almost every day of the data collection process, and meetings were

held regularly to ensure that there was consistency in the data collection.
Implications

The findings presented above have implications to both the field of criminology,
and to the overall crime fighting/prevention strategies developed by components of the
criminal justice system. First, the descriptive data presented indicate that there are both
consistencies and inconsistencies between data presented here and findings fi'om previous
research. Racial minorities and males are still substantially overrepresented as both

homicide victims and offenders, yet the disparity is not as pronounced in Detroit. Also,

83

most offenders tend to know their victims prior to the criminal incident, yet there was a
substantially large percentage of missing victim-offender information. There have been
two major findings in this research, both of which have been largely untested in previous
research. First, nearly 30 percent of all homicide incidents that occurred in Detroit in 1994
and 1995 involved drug circumstances. Second, almost 64 percent of all victims scored
medium to high on the victim culpability, suggesting that they might have somehow played
a role in their own demise. These findings suggest that there have been some sociological
changes that have resulted in this change in the nature of homicide. As hypothesized by
Decker (1996), this change seems to at least in part be due to the influx of drugs into our
inner cities. The growing drugs-youth-violence nexus suggests although overall crime rates
might continue to drop, certain subpopulations might continue to experience high rates of

victimization (Blumstein and Heinz III, 1995)

These findings also suggest that the criminal justice system has become less
effective in solving crime, assuming that clearance rates are used as a measure of police
effectiveness. Homicide investigations tend to get the highest allocations of resources per
case for any crime, but it does not appear to be working effectively. There are two
apparently contrary operational implications for police departments. First, the
implementation of a case screening model that provides more resources to cases that are

more likely to be solved suggests less resources should be allocated to drug related

 

84

homicides because they result in significantly lower clearance rates. Second, it could also
be argued that extra attention should be given to drug or gang related homicides because it
is likely that the violence could possibly result in "homicide epidemics" (Decker, 1996, p.
446) that eventually consume even more departmental resources. If these rash of drug

related homicides are stopped at the onset by arresting the offenders quickly, it could F
positively affect homicide rates and clearance rates by taking these criminals off of the 1

street. It seems possible that a differential response might be necessary for different types

 

of homicides, such as drug-motivated honricides. The moderate correlation between drug
circumstances and felony circumstances indicates that departments may need to implement
a more coordinated approach to the investigation of drug and/or felony homicides. These
two types of homicide circumstances might in fact present situations where the sharing of

intelligence and information is even more essential.

These findings also suggest that more research attention needs to be given to the
drugs-guns-youth nexus. Researchers have found that younger offenders who are involved
in the drug trade are more violent than others, and they demonstrated a greater willingness
to solve disputes with violence (Inciardi and Pottieger, 1991). Although overall national
violent crime rates are lower, there appears to be a greater propensity to use violence as a

reasonable alternative to interpersonal conflict among some juveniles.

85

The data that were used for this analysis were limited in their suitability for
extensive testing of criminological theory. However, some inquiry was possible. Routine
activities theory has been found to be a useful fiarnework for interpreting the social ecology
of urban homicide (Messner and Tardiff, 1985). A routine activities approach to urban
homicide suggests sociodemographic characteristics of the victim, along with temporal F
factors, are systematically related to the location of the homicide. For example, incidents

between family members are more likely to occur at night and on the weekends when !

 

contact with family members is greatest (Messner and Tardiff, 1985).

There was limited support for a relationship between a victim’s routine activities
and the temporal and spatial characteristics of crime. Based on the routine activities theory,
a significant relationship would be predicted between the victim-offender relationship and
spatial characteristics, suggesting that incidents involving family members would be more
likely to occur in private settings. The theory would also posit that family crimes would be
less likely to involve firearms, and more likely to involve other types of weapons such as
knives and personals weapons (e.g., hands/feet). Finally, another hypothesis is that crimes
involving family members would be more likely to occur on weekends since contact

between family members would be most frequent.

Of all of the hypotheses just presented, only two were significant. The relationship

between victim-offender relationship and spatial characteristics was significant. Incidents

86

involving family members were more likely to occur in private settings, and incidents
involving strangers were most likely to occur in public settings. Stranger homicides were
most likely to involve guns (86 percent), and incidents involving family members were less
likely to involve guns (58 percent). However, there was no relationship between the day of
the week the crime was reported, and the victim-offender relationship. Crimes involving
strangers were not significantly more likely to occur on the weekends (57 percent) when
compared to incidents involving family members (50 percent). These findings suggest

limited support for the theory.

APPENDICES

APPENDIX A

APPENDIX A

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure]
National Clearance Rates
90%
80%
IE 70%
60%
g 50%
a 40%
3 30% +Overall
I-
a 20% +Violent
10%
0% +Property
v-! v r N Homicide
a a a E 8 it § a f +
Year
Figurez
Population Race Characteristics"
76%
.White
.Black
DOthRace
United &ates Michigan Detroit
Percent

 

 

 

87

88

 

 

 

 

 

 

 

Figure 3
Population Sex Characteristics*
54%
54% 51%
51%
52% ,
49% 49%
50% ,
O

48% 46 A I Males

46%. I Females

44%

42% ,

United States Michigan Detroit
Pe rcent
Figure 4
Pbptflion Erindionrl Onaeter'stim

 

 

H.S. Grad
Assocrates
Grad or
Professronal

E
l
a

Percent [hitching Level

 

 

 

 

 

89

Figure 5

 

Wan Income

$35,000
$30,000
$25,000
$20,000
$15,000
$10,000
$5,000
$0

Michigan

 

 

.Mean been:

 

'Source: United States Department of Justice, 1995

 

APPENDIX B

APPENDIX B

INSTRUMENT QUESTIONS USED FOR ANALYSIS

VICTIM CHARACTERISTICS

Sex of decedent:
1 = Male
2 = Female
3 = Missing

Race of decedent:
1 = White (Includes Mexican Americans)
2 = Black
3 = American Indian or Alaskan Native
4 = Asian or Pacific Islander
5 = Missing

Date of decedent’s birth; (DOB) Month, Day, Year
99-99-99 = Missing
WEAPON USED

What was the primary cause of death?

Note: Primary cause of death refers to the injuries that were responsible for the
death. Incidental, non-lethal injuries prior to death, and post-mortem mutilations
are not a cause of death.

1 = Shot, Handgun

2 = Shot, Rifle

3 = Shot, Shotgun

4 = Shot, other gun, Specify: _
5 = Shot, Unknown Gun

6 = Stabbed, Knife

90

91

7 = Stabbed, Other Instrument or Unknown

8 = Blunt Object (Would Include Gun, If Used As Gun)
9 = Personal Weapon (E.g., Hands, Feet)

10 = Asphyxiation, Suffocation, Drowning, Strangulation

11 = Fire
12 = Other, Specify:
99 = Missing

PREVIOUS CRIMINAL RECORD

Did the decedent have one or more prior arrests for drug offenses?

l = Yes, Specify:
2=No
3 = No Mention

4 = Suspected But Uncertain

Did the decedent have one or more prior arrests for violent offenses?

l = Yes, Specify:
2=No
3 = No Mention

4 = Suspected But Uncertain

Did the decedent have one or more prior arrests for property offenses?
1 = Yes, Specify: _
2=No
3 = No Mention
4 = Suspected But Uncertain

92
F NY R

Did any offender take or attempt to take property?
1 = Yes - Specify: __
2=No
3 = No Mention
4 = Suspected by Uncertain Specify: __
If Yes,

Was the taking of property incidental to the homicide? That is, did the offender kill
the victim for reasons unrelated to taking of property and then take the property? ‘

1 = Yes

2 = No

3 = No Mention

4 = Suspected by Uncertain

Did any offender sexually assault the victim?
I = Yes
2 = No
3 = No Mention
4 = Suspected by Uncertain

Did the homicide involve the robbery of a drug dealer that occurred during the selling of
drugs?

1 = Yes

2 = No

3 = No Mention

4 = Suspected But Uncertain

Did the homicide incident involve the robbery of a drug buyer that occurred during the
purchase of drugs?

1 = Yes

2 = No

3 = No Mention

 

93

4 = Suspected But Uncertain

Did any offender unlawfully enter a structure to commit a crime?

A structure would include a dwelling, garage, barn, public building, commercial
establishment, or vehicle used as a dwelling.

l = Yes
2=No
3 = No Mention ....

4 = Suspected But Uncertain

DRUG CIRCUMSTANCE

Were drugs a circumstance in the homicide incident?
1 = Yes
2=No
3 = No Mention
4 = Suspected But Uncertain

 

‘npu—I...

VICTIM-OFFENDER RELATIONSHIP

Prior to the incident in which the homicide occurred, were the decedent and offender
strangers?

l = Yes (If yes, skip all other relationship questions)

2=No

3 = No Mention

4 = Suspected but Uncertain

5 = Missing

Where the decedent and the offender members of the same family?
1 = Yes
2=No
3 = No Mention
4 = Suspected but Uncertain

94

5 = Missing

Was the decedent related to the offender by blood (consanguineously)?
1 = Yes, decedent was offender’s parent
2 = Yes, decedent was offender’s child
3 = Yes, decedent was offender’s sibling
4 = Yes, but not nuclear family members (Specify relationship: )
5 = No '1‘
6 = No Mention '
7 = Missing

All other relationships will be considered as acquaintanceships.

 

TEMPORAL CHARACTERISTICS H

Date incident occurred: month-day-year
Note: This date may be different from the date the incident was reported to the
police.
99-99-99 = Missing

Time incident occurred.
9999 = Missing

Day of week incident occurred:

1 = Sunday

2 = Monday

3 = Tuesday

4 = Wednesday
5 = Thursday

6 = Friday

7 = Sunday

95

SPATIAL HARACTERI TICS

Location where injury occurred that led to homicide.

A private location is, by community standards, not open to the general public. All
others are considered to be public.

1 = Public, Specify:
2 = Private, Specify:
9 = Missing

PHYSICAL EVIDENCE

Were fingerprints or other physical evidence found?
1 = Yes, Specify:
2 = NO 3'
3 = No Mention
4 = Suspected but Uncertain

 

PRESENCE OF WITNESSES

Were witnesses found at the crime scene interviewed?
1 = Yes
2=No
3 = No Mention
4 = Suspected But Uncertain
5 = No witnesses at the crime scene
9 = Missing

 

96

VICTIM CULPABILITY

Was the decedent a known gang member or part of a drug organization?

1 = Yes

2 = Yes, But Not At Time of Death

3 = Never

4=No

5 = No Mention
r-
1

Was the decedent drinking alcohol at the time of the incident?

1 = Yes

2=No

3 = No Mention

4 = Suspected but Uncertain i"

 

Was the decedent consuming illegal drugs at the time of the incident?

1 = Yes
2 = No
3 = No Mention

4 = Suspected but Uncertain

Was the decedent in possession of illegal drugs at the time of the incident?

1 = Yes
2=No
3 = No Mention

4 = Suspected but Uncertain

Was the decedent in possession of alcohol at the time of the incident?
1 = Yes
2=No
3 = No Mention
4 = Suspected but Uncertain

 

97

Did the decedent have a history of association with drug dealers/users?

1 = Yes
2=No
3 = No Mention

4 = Suspected but Uncertain

Did the decedent have a history of drug abuse?
1 = Yes
2=No
3 = No Mention
4 = Suspected but Uncertain

Did the decedent have a history of alcohol abuse?

1 = Yes
2=No
3 = No Mention

4 = Suspected but Uncertain

Was the decedent identified by police as a drug dealer?
1 = Yes
2=No
3 = No Mention
4 = Suspected but Uncertain

Was the decedent identified by police as a drug buyer?
1 = Yes
2=No
3 = No Mention
4 = Suspected but Uncertain

 

 

98

Did the decedent have one or more prior arrests for drug offenses?

1 = Yes

2=No

3 = No Mention

4 = Suspected but Uncertain
9 = Missing

Did the decedent have one or more prior arrests for violent offenses?

1 = Yes

2=No

3 = No Mention

4 = Suspected but Uncertain
9 = Missing

Did the decedent have one or more prior arrests for property offenses?

1 = Yes

2=No

3 = No Mention

4 = Suspected but Uncertain
9 = Missing

Was the decedent buying drugs at the time of the incident?

1 = Yes
2=No
3 = No Mention

4 = Suspected but Uncertain

Was the decedent selling drugs at the time of the incident?

1 = Yes

2=No

3 = No Mention

4 = Suspected but Uncertain

 

 

LIST OF REFERENCES

 

REFERENCES

Bachmarr, Ronet, and Paternoster, Raymond (1997). Statistical Methods Fgr Criminology
and Criminal Justice. New York: The McGraw-Hills Companies, Inc..

Bayley, D. H. (1996). Measuring overall effectiveness. In L. T. Hoover (Ed.), Mg 7'
W3 (PP- 37-54) Washington, DC: Police Executive Research
Forum.

 

Bernard, Thomas J. (1993) The intent to harm: Angry aggression as a form of violent '
crime. In Homicide: The Victim/Offender Connection. Cincinnati, OH: Anderson.

Bloch, P., and Bell, J. (1976). Managing investigations: The Rochester system.
Washington, DC: The Police Foundation.

Block, C., and Christakos, A. (1995). Chicago homicide fi'om the sixties to the nineties:
Major trends in lethal violence. In C. Block, and R. Block (Eds), Trends Risks
and Interventions in Lethal Violence. Washington, DC: National Institute of
Justice.

 

Block, R. (1976). Homicide in Chicago: A nine-year study (1965-1973). The Journal 9f
Criminal Law and Criminology, 66(2), 496-510.

----- (1981). Victim offender dynamics in violent crime. Journal of Criminal Law and
Criminology, 72, 743-761.

Blumstein, A., and Heinz 111, H. J. (1995) Youth violence, guns and the illicit-drug
industry. In C. Block, and R. Block (Eds), Trends Risks and Interventions in
Lethal Violence. Washington, DC: United States Department of Justice.

Boris, S. B. (1979, August). Stereotypes and dispositions for criminal homicide.
Criminology, fl(2), 139- 1 58.

 

99

100

Bottomley, K., and K. Pease (1986). Crime and Punishment: Interpreting the data.
Philadelphia, PA: Open University Press.

Bynum, T. S., Cordner, G. W., and Greene, J. R. (1982, November). Victim and offense
characteristics. Criminology, 20, 301-318.

Cardarelli, Albert P., and Cavanagh, D. (1992, November). Uncleged homicides in the
United States: An exploratog study of trends and patteme. Paper presented at the

annual meeting of the American Society of Criminology, New Orleans, LA.

Cardarelli, Albert P. (1993, October). Impact of uncleared killings on homicide research.
Paper presented at the annual meeting of the American Society of Criminology,
New Orleans, LA.

Cawley, D. F., Miron, H. J ., Araugo, W. J., Wasserman, R., Mannello, T. A., and Huffman,
Y. (1977, June). Managing criminal investigations manfl. Washington, DC:

US. Government Printing Office.

Chaiken, J .M., P.W. Greenwood, and J. Petersilia (1977). The criminal investigation
process: A summary report. Policy Analysis, 3(2), 187-217.

Cook, P. (1979) The Clearance rate as a measure of criminal justice system effectiveness.
Journal of Public Economics 11(1), 135-142.

Cordner, G. W. (1989). Police agency size and investigative effectiveness. Journal of
Criminal Justice 11(3), 145-155.

 

 

Daly, M., and Wilson, M. (1982). Homicide and kinship. American Anthromlogist, 8‘1 ,
372-378.

Decker, S. H. (1995) Reconstructing homicide events: The role of witnesses in fatal
encounters. Journal of Criminal Justice, 5, 451-460.

 

 

 

101

Decker, S. H. (1996, November). Deviant Homicide: A new look at the role of motives

and victim-offender relationships. Journal pf Research 61' Crime and Delinquency,
36(4), 427-449.

Denno, D, and Cramer, J. (1976). The effects of victim characteristics on judicial decision-

making." In W.F. McDonald (Ed.), Crimme1' Jp§tice m the Victim. Beverly Hills,
CA: Sage.

Durkheim, E. (1951). Suicide: A sgdy in sociology. John A. Spaulding and George 3
Simpson (trans), New York: Free Press. I ‘

Eek, J. (1983). Solving crimee: The investigation pf b_u_rglm and robm. Washington,

DC: The Police Executive Research Forum. 'r:

 

----- (1992). Criminal investigations. In G.W. Cordner and DC. Hale (Eds), fleet werks
in mlicing? Cperations and administration examined. Cincinnati, OH: Anderson.

Gaines, L. K., Bruce L., and Swanagin, R. (1983). Case screening in criminal
investigations: A case study of robberies. Police Studies 6(2), 22-29.

 

Goldstein, P. J. (1985). The drugs/violence nexus: A tripartite conceptual framework.
Journal of Drug Issues, fi, 493-506.

Gottfredson, M., and Travis H. (1990). A general theor_y of crime. Stanford, CA: Stanford
University Press.

Greenberg, B., and Wassherman, R. (1979). Managing criminal investigations.
Washington, DC: National Institute of Law Enforcement and Criminal Justice.

Greenwood, P. W., Chaiken, J. M., and Petersilia, J. (1977) The criminal investigatien
process. Lexington, MA: Lexington Books.

102

Hoover, L. T. (1996). Translating total quality management from private sector to policing.
In L. T. Hoover (Ed.), Qu_en_tifling Qualig in Policing (pp. 1-22). Washington, DC:

Police Executive Research Forum.

Horvath, F., and Meesig, R (1996) The criminal investigation process and the role of
forensic evidence: A review of empirical findings. Jeumfl pf Forensic Science,
41 (6), 963-969.

Inciardi, J. A., and Pottieger, A. E. (1991). Kids, crack, and crime. The Journal pf Drug
Issues, 21(2), 257-270.

Lee, E. S. (1995, July). Age patterns in homicide. In C. Block and R. Block (Eds),
Trends, Risks, and Interventions in Lethal Violence. Washington, DC: National
Institute of Justice.

Levi, K. (1980). Homicide as conflict resolution. Deviant Behavier: An Interdisciplim
Journal, 1, 281-307.

 

Luckenbill, D. F. (1977). Criminal Homicide As a Situated Transaction. Social Problems
2;, 176-186.

Lundsgaarde, H. F. (1977). M r' s a i : A ul i of Houst nhorrricide

patterns. New York: Oxford University Press.

Markides, K. S, and Tracy, G. S. (1977). The Effect of the Age Structure of a Stationary

Population on Crime Rates. The Journal of Criminal Law and Criminolegy, 61(3),
351-355.

Maxfield, M. G. (1989). Circumstances in Supplemental Homicide Reports: Variety and
validity. Criminelogy, 11 (4), 671-695.

Maxfield, M. G., and Babbie, E. (1995). Researeh Methods for Criminal Justig er_rd
Criminology. Belmont, CA: Wadsworth Publishing Company.

_ til-.1 m
g.

L...

 

 

103

McDonald, W.F. (1976). Criminal justice and the victim. In W.F. McDonald (ed.)
Criminal Justice and the Victim. Beverly Hills, CA: Sage.

Menard, S. (1995). Applied legietic reggession analysis. Thousand Oaks, CA: Sage

Messner, S. F., and Tardiff, K. (1985). The social ecology of urban homicide: An
application of the ‘routine activities’ approach. Criminology, 11(2), 241-267.

Morris, A. (1955). Homicide: An appreach te me preblem of crime, Boston, MA: Boston

University Press.

Ott, R. L., Rexroat, C., Larson, R., and Mendenhall, W. (1992). Statistics: A tool for the
sgial §cienees (Fifth Ed.). Boston: PWS-Kent.

Parker, R. (1995). Alcohel and homicide. A combipetiop 61 Mo American traditions.
Albany, NY: State University of New York Press.

----- (1989). Poverty, subculture of violence, and type of homicide. Social Forces 61, 983-
1007.

 

Riedel, M. (1995). Getting away with murder: An examination of arrest clearances. In C.

Block and R. Block (Eds), Trende, Rieks, and Interventiens in Lethal Violence (pp.
91- 98). Washington, DC: National Institute of Justice.

Riedel, M. (1981). Stranger homicides in an American city. Paper presented at the
American Society of Criminology. Washington, DC.

Riedel, M., and Przybylski, R. K. (1993). Stranger murders and assaults: A study of a

neglected form of stranger homicide. In A. V. Wilson (Ed.), Homieide: The
Vietim/Offeneer Cennectien (pp. 359-3 82). Cincinnati, OH: Anderson.

Riedel, M., and Rinehart, T. A. (1996). Murder clearances and missing data. Journal of
Crime and Jyetiee, 12 (2), 83-102.

 

 

104

Rinehart, T. A. (1994). An a_ryr_lysis pf murder elearances in Chicago: 1281-1991.

Unpublished master’s thesis, Southern Illinois University at Carbondale,
Carbondale, Illinois.

Rojeck, D. G., and Williams, J. L. (1993). Interracial v. intraracial offenses in terms of the
victim/offender relationship. In A. V. Wilson (Ed.), Hegcide; The
vietim/effender eonnectiop (pp. 249-266). Cincinnati, OH: Anderson.

Roth, S. F. (1995). Gun-related violence. In C. Block and R. Block (Eds), Trend_s, Risks,

and Interventions in lethal violence, Prmeedings pf the Third Annual Spring

Symmsium of me Homicide Research Working @gp. (pp. 265-278).
Washington, DC: National Institute of Justice.

Sarnaha, J. (1991). Criminal Justice. (2“ Ed). West Publishing Company: St. Paul, MN.

Seamon, T. M. (1990). The Philadelphia scene team experiment. The Police Chief, 61,
48-52.

Segall, W. E., and Wilson, A. V. (1993). Who is at greatest risk in homicides?: A
comparison of victimization rates by geographical region. In A. V. Wilson (Ed.),
Homicide: The victim/offender connection (pp. 343-356). Cincinnati, OH:
Anderson.

Skogan, W., and Antunes, G. E. (1979, Fall). Information, apprehension, and deterrence:
Exploring the limits of police productivity. Journal of Criminal lggtice, 3, 99-116.

Smith, M. D., and Parker, R. (1980). Type of homicide and variation in regional rates.
Social Forces fl, 136-147.

 

Stephens, D. W. (1996). Community Problem-Oriented Policing: Measuring impacts. In
L. T. Hoover (Ed.), uanti in uali in Policin (pp. 95-129). Washington, DC:
Police Executive Research Forum.

Sutherland, E., and Cressey, D. (1978). Principles of criminology (10"I ed.). Philadelphia,
PA: Lippincott.

105

Tennenbaum, A. N., and Fink, E. L. (1994). Temporal regularities in homicide: cycles,
seasons, and autoregression. Journal of Qu_a_r1titative Criminolog, 16, 317-342.

Topping, CW. (1952, November). The death penalty in Canada. In The Annals of the
American Academy of Political and Social Science, Murder and me penalty pf
death (pp. 150-185). Philadelphia, PA.

 

United States Census Bureau. (1990). LookUp 1990 Census Information [On-line].
Available: http://www.census.gov/cdrom/lookup/.

United States Department of Justice. (1995). Sourcemok of Cg’mjpal Jpsjce Statistics.

‘I
a.

Washington, DC.

United States Department of Justice. (1994). Violen ' on land Law Enforc ment
Act of 1994 (Briefing Bmk). Office of Community Oriented Policing Services:
Washington, DC.

Walker, 8., Spohn, C., and DeLone, M. (1996). The eolor of justice. Belmont, CA:
Wadsworth Publishing Company.

Waegel, W. B. (1982). Patterns of police investigation of urban crime. Journal of Police
Science and Administration QM), 452-465.

 

 

Williams, K., and Flewelling, R. (1988). The social production of criminal homicide: A
comparative study of disaggregated rates in American cities. Amerifl Sociology
Review 4_8_, 421-431.

 

Wilson, N. K. (1993). Gendered interaction in criminal homicide. In A. V. Wilson (Ed.),
Homieide: The victim/offender connection. Cincinnati, OH: Anderson.

Wolfgang, M. E. (1958). Patterns in criminal homicide. Philadelphia, PA: University of
Pennsylvania Press.

----- (1967). Studies in homicide. Philadelphia, PA: University of Pennsylvania Press.

"rrrrrirrrrr“