Kraemer Et Al., 2004

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 19

Behavioral Sciences and the Law

Behav. Sci. Law 22: 325–343 (2004)


Published online 1 March 2004 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/bsl.581

Comparing Single and Serial


Homicide Offenses
Gretchen W. Kraemer, J.D., Ph.D.,y
Wayne D. Lord, Ph.D.,yy
and Kirk Heilbrun, Ph.D.*

Serial homicide has attracted much attention, but little


empirical scientific investigation. This exploratory study
reports demographic information on a large sample of
serial homicide offenders (157 offenders, 608 victims),
and compares a subsample of serial homicide offenses
with a control group of single homicide offenses. Results
show that serial homicide offenders target more women
than men, and kill more strangers than family or friends.
Single homicide offenders kill men and women in equal
frequency, but kill family and friends more often than
strangers. Serial homicide offenders kill for apparent sex-
ual motivation more often than for any other reason, while
single homicide offenders kill most often out of anger.
Copyright # 2004 John Wiley & Sons, Ltd.

Serial homicide is a tragic crime, receiving serious law enforcement concern and
attention. Serial homicide has also captured the interest of popular culture, spawn-
ing a thriving business in works of fiction and true crime. Despite the intense interest

*Correspondence to: Kirk Heilbrun, Ph.D., Drexel University, Department of Psychology, Philadelphia,
PA 19102, U.S.A. E-mail: kirk.heilbrun@drexel.edu
y
Drake University; Whitfield & Eddy, P.L.C.
yy
Federal Bureau of Investigation, Child Abduction and Serial Murder Investigative Resources Center.
Preliminary findings from this project were presented at the 2000 American Academy of Forensic
Sciences meeting in Reno, Nevada, and the 2000 Biennial Convention of the American Psychology–Law
Society in New Orleans, Louisiana. This project was supported by an American Psychology–Law Society
Grant-in-Aid, received November, 1998 and an American Board of Forensic Psychology Dissertation
Award, received April, 1999. Data collection was made possible by the Visiting Scholar Program of the
Federal Bureau of Investigation’s National Center for the Analysis of Violent Crime.
This project was completed as part of the first author’s doctoral dissertation. Kirk Heilbrun served as
committee chair. Wayne Lord served as FBI supervisor for the project. Many thanks to the other members
of the dissertation committee: Donald Bersoff, Catherine Panzarella-Tse, and Michael Williams. Our
thanks also to Cynthia Lent, Kelly Newsome, Tim Templin, Dreama Long, and Carissa Goldbeck for
their valuable data collection and administrative assistance. Thanks to the FBI for its support of this
research project. In particular, we would like to recognize William Hagmaier III, Larry G. Ankrom,
Stephen Etter, Robert Morton, and Mark Young of the FBI for their research support.

Contract/grant sponsors: American Psychology–Law Society; American Board of Forensic Psychology;


Visiting Scholar Program of the Federal Bureau of Investigation’s Center for the Analysis of Violent
Crime.

Copyright # 2004 John Wiley & Sons, Ltd.


326 G. W. Kraemer et al.

in serial homicide, few empirical studies exist. It can be hard to distinguish reliable
facts from unproven assertions.
Most definitions of serial homicide include three elements: the number of
victims, time, and motivation (Keeney & Heide, 1995). The number of victims
required for an offender to be classified as a serial homicide offender varies from two
to ten (Dietz, 1986; FBI, 1994; Hickey, 1997; Ressler, Burgess, & Douglas, 1988).
The FBI definition originally required two or more separate homicide offenses
committed by a common offender (FBI, 1994). Since then, the official FBI
definition underwent review and revision in the course of enacting new legislation,
and now requires three homicides (Protection of Children from Sexual Predators
Act of 1998, 2000). The temporal element distinguishes mass homicide, which
occurs at one time, in one place, from serial and spree homicide, which occurs over
time, at separate locations. Serial homicide and spree homicide are usually distin-
guished by whether the offender has experienced an ‘‘emotional cooling-off’’
between offenses—offenders lack a cooling-off period in spree homicide, but have
a cooling-off period in serial homicide (Greswell & Hollin, 1994). The typical
motivation for serial homicide has been described as either sexual gratification or
internal psychological gratification (DeHart & Mahoney, 1994; Egger, 1998). Other
definitions exclude homicides committed in connection with military service,
political terrorism, or organized crime (Keeney & Heide, 1994). Because there is
little empirical evidence about serial homicide offenders’ motivations, and motiva-
tion must be inferred from observable behavior, it seems premature to include
motivation as part of the definition rather than a variable for investigation.
Many authors have provided detailed, intriguing case studies of serial homicide
offenders (see, e.g., Egger, 1998; Pollock, 1995). To understand how individual
case studies relate to one another, however, we need a sense of the population of
serial homicide offenders. Given sufficiently detailed information about a large
number of serial homicides, it would be possible to compare individual offenders or
offenses and distinguish common elements from unique aberrations. One of the best
ways to gain useful information about any population is to study a randomly selected
and representative group sample. Unfortunately, few group studies of serial homi-
cide offenders exist. The conspicuous absence of serial homicide group studies is
perhaps understandable—serial homicide is a rare crime and the most detailed
records of the crime, those of law enforcement, usually are not available to
researchers.
Despite these difficulties, several investigators have conducted studies involving
more than one serial homicide offender. One of the first, and the only published
study to include direct interviews of offenders, included 36 sexual homicide
offenders, 25 of whom were serial offenders (Ressler, et al., 1988). Hickey (1997)
compiled a database including 399 offenders accounting for between 2,536 and
3,860 victims. Godwin (2000) studied 107 offenders and 728 victims. The study by
Holmes and DeBurgers’ (1988) included 110 offenders, but provided no statistical
data on the sample. Other studies include samples of 20 offenders or fewer, and rely
primarily on the media as a source of data (Jenkins, 1988; Keeney & Heide, 1994;
McKenzie, 1995; Warren, Hazelwood, & Dietz, 1996). Although media accounts
provide an accessible source of information, reports may include inaccurate facts not
yet corrected through additional investigation. In addition, relying solely on media
accounts may over-sample highly publicized cases (Egger, 1990).

Copyright # 2004 John Wiley & Sons, Ltd. Behav. Sci. Law 22: 325–343 (2004)
Comparing single and serial homicide offenses 327

The limited available research provides basic information about serial homicide
offenders. Serial homicide offenders are described as killing when they are in their
mid-twenties (Brittain, 1970; Fox & Levin, 1999; Hickey, 1997; Langevin, Ben-
Aron, Wright, Marchese, & Handy, 1988). Godwin (2000) reported the mean
offender age as 30. Most serial homicide offenders, like most homicide offenders,
are male (Godwin, 2000; USDOJ, 2001). Some studies exclude female serial
offenders because they are a rare subset of a group of rare criminals, but critics of
this approach argue that exclusion of female offenders eliminates the opportunity to
learn about gender differences among offenders (Hickey, 1990; Keeney & Heide,
1995).
Although there are roughly as many African American homicide offenders as
Caucasian homicide offenders, because there are fewer African Americans in the
overall population, African Americans are more heavily represented in the popula-
tion of homicide offenders (Fox & Levin, 1994; USDOJ, 2001). In serial homicide,
offenders are more likely to be Caucasian than African American, in numbers
consistent with the relative proportion of Caucasians and African Americans in the
United States’ population (Godwin, 2000; Hickey, 1990, 1997; Keeney & Heide,
1994; Ressler, Burgess, Douglas, Hartman, & D’Agostino, 1986).
The available research also provides useful information on victims of serial
homicide offenders. According to the Uniform Crime Reports (UCR), most homi-
cide victims are between 20 and 29 years old, with 30–39 year olds forming the next
most frequent victim age group (USDOJ, 2001). Between 8 and 12% of homicide
victims each year are children. Like homicide generally, young and middle-aged
adults are the most common targets of serial homicide (Godwin, 2000; Hickey,
1997). Children (under 18 years old) form a higher percentage of victims of serial
homicide than victims of homicide. In Hickey’s study, 24% of serial offenders killed at
least one child; Godwin reported that 27% of victims were under the age of 17. The
gender of victims varies distinctly between homicide generally and serial homicide.
According to the UCR for the past ten years, most homicide victims are male. By
contrast, most victims of serial homicide offenders are female (Brittain, 1970; Egger,
1998; Godwin, 2000; Langevin, et al., 1988). Godwin reported that 67% of the
victims in his study were female. Racial differences in homicide victims are similar to
those of homicide offenders. According to the UCR, the number of Caucasian victims
is roughly equal to the number of African American victims, but because African
Americans are less prevalent in the population they are overrepresented as homicide
victims. Like homicide generally, serial homicide is usually intraracial, with offenders
killing victims of the same race (Godwin, 2000; Hazelwood & Douglas, 1980; Hickey,
1997). The racial frequencies of serial homicide victims are roughly equivalent to the
prevalence of different races in the U.S. population. For example, in Godwin’s study,
80% of the victims were Caucasian, 16% were African American, and 4% were
members of other racial groups.
Victims of homicide are usually friends, family, or acquaintances of the offender
(USDOJ, 2001). Serial homicide, however, is typically described as a stranger-to-
stranger crime (Egger, 1998; Godwin, 2000). Godwin reported that 90% of the
victims in his study were strangers to their assailants; Hickey (1997) found 62% of
victims were strangers.
The differences in gender and relationship between homicide victims and serial
homicide victims suggest different reasons for killing. Borrowing from ecological

Copyright # 2004 John Wiley & Sons, Ltd. Behav. Sci. Law 22: 325–343 (2004)
328 G. W. Kraemer et al.

biology models, we observe that animals kill for two reasons: competition and
predation (Lord, Boudreaux, Jarvis, Waldvogel, & Weeks, 2002). Most homicide
offenses are between two males who know each other, and can be understood as
eliminating competition of peers (Daly & Wilson, 1988). Serial homicide offenders,
however, although also predominantly male, more frequently target women and
children who are strangers. Killing such victims may be a form of predation, to
acquire a desired resource such as sex.
Serial homicide is often mentioned in conjunction with profiling. Offender
profiling involves working backward from a crime scene to develop a description,
or profile, of the person who committed the crime (Homant & Kennedy, 1998).
Profiling is often described as an art (Ault & Reese, 1980; Douglas, Ressler, Burgess,
& Hartman, 1986). In practice, expert law enforcement officers trained to recognize
commonalities among crime scenes collaborate to develop the most likely profile of
an individual based on evidence of that crime and the group’s collective knowledge
about other crimes.
Another method of case analysis currently used by law enforcement is criminal
investigative analysis (CIA). CIA is a comprehensive, multidisciplinary method of
assessing the facts of a criminal act or series of criminal acts from forensic
investigative, scientific, and behavioral perspectives. Using both inductive and
deductive reasoning, CIA incorporates a detailed understanding of case facts,
empirical research, clinical psychiatric and psychological principles, and the experi-
ence and insight of trained practitioners. CIA, in many respects, is comparable to
the process of team-centered clinical diagnosis. With CIA, the goal is not to obtain a
‘‘profile’’ of the suspected offender, but to use behavioral science principles to
narrow the investigative focus, resulting in a more efficient use of investigative
resources. One of the major criticisms of profiling, and to a lesser extent CIA, is its
current lack of scientific foundation (Alison, Bennell, Mokros, & Ormerod, 2002;
Homant & Kennedy, 1998). In particular, offender profiling relies on trait theory,
which posits that persons behave in consistent and identifiable ways because of
stable personality traits, while current thinking focuses on theories of personality
that accommodate situational variability and receive greater empirical support
(Alison, et al., 2002). Further, the reliability and validity of profiling has not been
established (Homant & Kennedy, 1998; Pinizzotto & Finkel, 1990). The only
published account of reliability assessment in profiling presents percent agreement
between an FBI profiler reading a ten minute case summary, and trained profilers
listening to the presentation and categorizing the offender (FBI, 1985). Percent
agreement, although not the ideal measure of interrater reliability, ranged from 51.7
to 92.6%. Despite these limitations, profiling is not likely to cease pending empirical
support by scientific data. Indeed, law enforcement officers request profiling
assistance more often each year (Wilson, Lincoln, & Kocsis, 1997). As criminal
investigative analysis increases in popularity, it too may fall under criticism for lack
of proven empirical support. However, good science can inform and shape the
practice of profiling and criminal investigative analysis, much as other research has
supported the predictive superiority of actuarial over unstructured clinical predic-
tions (Borum, 1996; Grove & Meehl, 1996).
For this study, we set out to answer some basic questions about serial homicide.
Specifically, we sought to determine whether a set of empirically valid demographic
parameters characterized serial murderers as a group, and whether such parameters

Copyright # 2004 John Wiley & Sons, Ltd. Behav. Sci. Law 22: 325–343 (2004)
Comparing single and serial homicide offenses 329

reliably distinguished serial from single homicide offenses. We defined serial


homicide as two separate events, occurring at least 24 hours apart, by a common
offender. This definition was used by the FBI at the time this study was conducted,
and provided a broad base from which to gather information. (The current
definition requires three homicides for classification as a serial offender and was
changed after detailed review in the course of drafting the Protection of Children
from Sexual Predators Act of 1998.) Having a broad definition was important,
considering the exploratory nature of the study. In this study, we consider serial
homicide in two ways. First, we gathered basic demographic information about
serial homicide offenders, victims, their relationship, and the offenders’ reasons for
committing the crimes. Second, we compared a smaller group of the first offenses of
the serial homicide group to a group of offenses by single homicide offenders on
homicide event characteristics.

METHOD

Participants

We began by constructing a working database of serial offender names, drawing


from FBI case files and secondary sources such as media accounts. In the working
database, 93% of offenders were male. Of the cases that contained racial informa-
tion, 68% were Caucasian, 27% were African American, and 5% were from other
minority groups. Offender names were selected in no particular order, but without
the benefit of a formal random number generator. FBI case files provided most of
the data for this study, providing a national sample. Local law enforcement officers,
prosecutors, and medical examiners were contacted to obtain additional informa-
tion. Cases were eliminated from the study if the case file contained insufficient
information to mount an archival search, or if local offices no longer possessed
original documents or declined to assist with the study. Cases were also eliminated if
the offenses occurred in less than 24 hours because those events did not fall within
our definition of serial homicide. Conviction for at least one homicide was
confirmed for each case entered into the study. The resulting sample included
157 offenders and 608 victims. Demographic information is presented in the
‘‘Results’’ section.
Next, we obtained a sampled group of single homicide cases from the FBI’s
computerized violent crime database, in no particular order, but again without use
of a formal random number generator. The selected cases met the following search
criteria: Crime of homicide, offender was more than 18 years of age at the time of the
offense, and offender was in custody. Conviction for the homicide was confirmed
and offender criminal histories were reviewed to ensure the offender had no other
arrests or convictions for homicide. The resulting sample included 133 victims and
195 offenders (the number of offenders is greater than the number of victims
because several offenders worked together). A sub-sample of serial homicide
offenses was selected for comparison with the single homicide group. One hundred
and thirty-three first offenses were selected randomly from the sample of serial
offenders. The resulting serial sample included 147 offenders, five of whom were
single offenders who teamed up with a serial offender. Demographic information

Copyright # 2004 John Wiley & Sons, Ltd. Behav. Sci. Law 22: 325–343 (2004)
330 G. W. Kraemer et al.

(excluding the single homicide members of a serial homicide team) is presented in


the Results section. After the initial research was completed, identifying information
was deleted from the case file.

Coding

Each victim was considered a separate homicide event. We used the Violent Crime
Apprehension Program (VICAP) form for data collection.1 In addition, we gathered
data on the relationship between the victim and the offender, the apparent motive
for the crime, and the level of confidence that the victim was, in fact, killed by the
offender (conviction, confession plus law enforcement linkage of crimes, or law
enforcement linkage of crimes without confession).

Variables

We obtained demographic information on the offender and the victim, information


on the offender’s style of approach, use of a vehicle, location of events (initial contact
site, murder site, and body recovery site), symbolic artifacts or writings at the crime
scene, positioning of the body, use of restraints, the body’s state of dress, use of
weaponry, and medical forensic information including cause of death, location and
type of trauma, and evidence of sexual assault. Particularly vulnerable victims were
identified in the database to explore the hypothesis that serial homicide offenders
target vulnerable victims. Vulnerable victims included children, the elderly, home-
less, prostitutes, or mentally retarded persons. Because most of the variables in the
study are categorical, items were dummy coded into dichotomous variables for entry
into analyses.

Age

Age was coded into five year intervals, consistent with UCR practice. For offenders,
the groups were under 20, 20–29, 30–39, 40–49, and 50 and older. For victims, the
groups were birth-12, 13–18, 19–29, 30–39, 40–49, 50–59, 60–69, and 70 and
older. The childrens’ ages were subdivided based on Godwin’s rationale of
differences in sexual maturity and mobility.

Relationship

Victims were rated as being family, including blood relatives and spouses, para-
mours (required a continuing physical relationship, not a one-time interaction),
acquaintance or friend, stranger, and unknown.

1
The coding form and instructions are available upon request.

Copyright # 2004 John Wiley & Sons, Ltd. Behav. Sci. Law 22: 325–343 (2004)
Comparing single and serial homicide offenses 331

Reason for the Crime

We use ‘‘intent’’ (rather than ‘‘motive’’) to describe the reasons for the offense.
Motive suggests an internal psychological drive, or at least a known reason, for the
crime. Here, where the offender confessed, we used the known reason, and we
approximated the reason based on the physical evidence in other cases. We
considered five reasons for committing homicides: sex, profit, emotion based,
psychosis, and unclear (see Cloud, 1996; Dietz, 1985). To receive a positive coding
for sex, an offender must have confessed to killing for sex, or the crime scene must
have contained physical evidence of a sexual act such as the presence of semen or
genital/anal trauma. In the literature, serial homicide frequently is described as a
sexual crime, and ambiguous evidence is sometimes considered evidence of sexual
intent. We purposefully adopted a conservative coding scheme to ensure that any
findings about sexual intent were supported by evidence in the case. Emotion-based
crimes were triggered by anger, often by an argument. The psychosis category was
rarely used, and was positively coded only if the record contained medical evidence
that the offender suffered from command hallucinations at the time of the crime.
Although command hallucinations is not the only mode of psychosis possible, it was
reliably coded in the existing records, and thus formed a discrete and measurable
category. If two reasons were present at the same time, both were coded.

Interrater Reliability

One hundred and seventy serial homicide cases (28%) were coded by two raters and
analyzed using inter-item correlation and Cronbach’s alpha (see Table 1). Mean
alpha was 0.76. Coefficients ranged from 0.15 for approach to 1.0 for gunshot wound
as cause of death. The coefficient for approach was particularly low because the
amount of information available to research raters at the conclusion of the case was
different than that available to the investigative agent who first completed the form.

RESULTS

Characteristics of Serial Homicide Offenders

Ninety-five percent of serial homicide offenders were male, 68% were Caucasian,
and the mean age was 31 years (see Table 2). Men were overrepresented as serial
homicide offenders in comparison with the population, but were roughly equal in
proportion to homicide offenders reported in the UCR (USDOJ, 2001). The
prevalence of Caucasian serial offenders was lower than the prevalence of Caucasian
persons in the population, but was higher than the prevalence of Caucasian
offenders reported in the UCR. The prevalence of African American serial offenders
was higher than the prevalence of African Americans in the population, but lower
than the prevalence of African American offenders reported in the UCR. There were
few differences in the incidence of other minorities in comparison with either the
population or UCR offenders.

Copyright # 2004 John Wiley & Sons, Ltd. Behav. Sci. Law 22: 325–343 (2004)
332 G. W. Kraemer et al.

Table 1. Reliability analysis

Variable Number of cases Correlation Alpha

Approach 170 0.12 0.15


Location
Differing sites 160 0.86 0.92
Body recovery 157 0.36 0.45
Murder 120 0.29 0.35
Initial contact 141 0.44 0.53
Body disposal
Moved 145 0.26 0.33
Method of disposal 161 0.37 0.49
Manner of disposal 166 0.63 0.77
Cause of death
Gunshot wounds 164 1.00 1.00
Stab wounds 164 0.87 0.93
Cutting wounds 164 0.42 0.56
Blunt force injury 164 0.77 0.87
Strangulation 164 0.80 0.88
Crushing injury 165 0.87 0.93
Undetermined 164 0.62 0.76
Location of injury
Head 166 0.63 0.73
Torso 164 0.76 0.86
Arms 164 0.58 0.73
Legs 164 0.47 0.64
Breasts 164 0.84 0.91
Genitalia 164 0.82 0.90
Buttocks 164 0.81 0.89
Anus 164 0.88 0.94
Defensive injuries 125 0.92 0.96
Extent of trauma
Blunt force 157 0.72 0.84
Stab wounds 44 0.97 0.98
Cut wounds 35 0.77 0.85
Gunshot wounds 29 0.96 0.98
Weapon used
Gun 165 0.98 0.99
Blade 165 0.95 0.97
Club 165 0.79 0.88
Ligature 165 0.79 0.88
Personal 168 0.70 0.82
Other 166 0.79 0.88
Number 167 0.58 0.73
Rare events
Carving 164 0.60 0.74
Symbolic 165 0.64 0.78
Victim bound 167 0.17 0.15
Clothing present 156 0.38 0.52
Bitemarks 168 0.99 0.99
Unusual torture 169 0.67 0.80
Sexual assault 126 0.49 0.65

Ninety-four percent of victims of serial homicide offenders were killed by one


offender acting alone. Of those offenders for whom data were available, 61% were
employed at the time of the offense (N ¼ 64, 104 known). The mean level of
education was 11.51 years. Thirty-one percent were married at the time of the
offense (N ¼ 34 of 108 known) and 16% had been married in the past (N ¼ 17 of

Copyright # 2004 John Wiley & Sons, Ltd. Behav. Sci. Law 22: 325–343 (2004)
Comparing single and serial homicide offenses 333

Table 2. Offender demographic comparison

Sample (fo) Census (fe) 2 UCR (fe) 2

Gender (f )
Serial 132.12** 4.0*
Male 149 77 141
Female 8 80 15
Serial subsample 126.48** 6.42*
Male 137 70 128
Female 5 72 14
Single 139.12** 0.23
Male 178 96 176
Female 17 99 19
Race (f )
Serial 37.67** 34.26**
Caucasian 107 121 72
African American 44 20 81
Other 6 16 4
Serial subsample 94.01** 8.15*
Caucasian 81 109 65
African American 56 18 73
Other 5 15 4
Single 186.76** 14.42**
Caucasian 94 150 90
African American 88 25 100
Other 13 20 5

Expected frequencies were calculated based on reported totals and solving for . Expected frequencies for
2000 UCR figures were calculated using known values. Although the 2000 Census permitted respondents
to endorse multiple categories for race, because those numbers exceeded 100%, numbers from
respondents selecting only one race were used.
*p < 0.05; **p < 0.01.

108). Offenders were convicted of homicide a mean of 4.4 times (SD ¼ 3.70), with a
mode of one conviction, and a range of 1–13 convictions.
Serial homicide victims were most likely female (67%) and Caucasian (71%), and
were an average of 33 years old (see Table 3). Victims of serial offenders were
significantly more likely to be female and less likely to be male than expected based
on the population, or reported homicide victims in the UCR. The prevalence of
Caucasian victims was lower than that in the general population, and greater than
the prevalence of Caucasian victims reported in the UCR. Conversely, the pre-
valence of African American victims was greater than the prevalence of African
American persons in the population and lower than the prevalence of African
American homicide victims reported in the UCR. Most victims were strangers to
their offenders (N ¼ 409; 67%), followed by acquaintance (N ¼ 130; 21%), family
(N ¼ 33; 5%), paramour, and unclear (N ¼ 18 each; 3%). Over half of the victims
were killed for sexual purposes (N ¼ 334). Eighteen percent (N ¼ 109) were killed
for profit, 17% (N ¼ 106) for emotional reasons, 17% (N ¼ 102) were unclear, and
4% (N ¼ 25) of the killings were motivated by the offender’s psychosis.

Comparing Single and Serial Offenses

Demographic data on the single homicide offender group were consistent with the
prevalence of the genders of offenders reported in the UCR (see Table 2). The single

Copyright # 2004 John Wiley & Sons, Ltd. Behav. Sci. Law 22: 325–343 (2004)
334 G. W. Kraemer et al.

Table 3. Victim demographic comparison

Sample ( fo) Census ( fe) 2 UCR ( fe) 2

Gender ( f )
Serial 63.65** 623.04**
Male 201 298 463
Female 407 310 145
Serial subsample 25.31** 173.86**
Male 36 65 101
Female 97 68 32
Single 0.48 42.14**
Male 69 65 101
Female 64 68 32
Race ( f )
Serial 84.06** 146.12**
Caucasian 432 468 298
African American 145 77 295
Other 27 63 15
Serial subsample 40.56** 19.61**
Caucasian 84 102 65
African American 41 17 65
Other 7 14 3
Single 42.89** 50.38**
Caucasian 77 102 65
African American 42 17 65
Other 14 14 3

There were 15 Latino, nine Asian, and three ‘‘other ethnicity’’ victims in the serial sample. Expected
frequencies were calculated based on reported totals and solving for . Expected frequencies for 2000
UCR figures were calculated using known values. Although the 2000 Census permitted respondents to
endorse multiple categories for race, because those numbers exceeded 100%, numbers from respondents
selecting only one race were used.
*p < 0.05; **p < .01.

offender group, however, had fewer African American offenders and more
Caucasian offenders than expected based on UCR data. Likewise, the single
homicide group compared to the census population in the same way the serial
group compared to the population: more males and fewer females, and more African
Americans. There was one exception: The single group had fewer Caucasians than
would be expected based on population data. Too many cases had missing data
on education level and marital status to permit analysis.
Demographic data on the single homicide victim group were notably different
from the prevalence of gender and race in the population of victims reported in the
UCR (see Table 3). Victims were male (N ¼ 69) and female (N ¼ 64). More victims
were Caucasian (N ¼ 77; 58%) than African American (N ¼ 42, 32%). In contrast
to the UCR data, victims in the single group were less likely to be male and more
likely to be female. Likewise, victims were less likely to be African American and
more likely to be Caucasian or another minority race. In comparison to the 2000
U.S. Census data, there was a proportionate number of males and females, signi-
ficantly more African Americans and fewer Caucasians than expected.
We tested group differences with chi square and t-tests where appropriate.
Table A1 in the Appendix provides descriptive data and resulting statistics. Next,
we ran discriminant function analyses to discover which variables most distin-
guished serial offenses from single offenses.

Copyright # 2004 John Wiley & Sons, Ltd. Behav. Sci. Law 22: 325–343 (2004)
Comparing single and serial homicide offenses 335

Chi Square Analyses

Offender Characteristics

The split between male and female offenders was more extreme for serial offenders
than single offenders (single offenders, 10% female, 90% male; serial offenders, 4%
female, 96% male). The single group followed the 90% male, 10% female pattern
reported in the UCR for the prior ten years. Ages of offenders were fairly equivalent,
and differences were not statistically significant. There were slightly more Caucasian
offenders in the serial group and slightly more African American offenders in the
single homicide group; however, these differences were not statistically significant.

Victim Characteristics

Single and serial offenders targeted victims by gender in significantly different ways.
The proportion of males to females killed by single homicide offenders was
approximately 1:1; serial homicide offenders, however, killed 2.7 females to every
male. There were no significant differences in the ages, races or vulnerability
statuses of the victims killed.

Intent and Relationship

Single and serial offenders significantly differed in their reasons for committing the
crime. Single offenders were significantly higher in emotion-based crimes
(2(1) ¼ 17.14, p < 0.001). Serial homicide offenders were more likely to commit
sexually based crimes (2(1) ¼ 23.30, p < 0.001). Likewise, there were differences in
relationship between single and serial homicide offenders. Single offenders killed
more paramours (2(1) ¼ 9.58, p < 0.01) and acquaintances (2(1) ¼ 23.76,
p < 0.001). Serial offenders killed more strangers (2(1) ¼ 43.85, p < 0.001).

Approach, Event Locations, and Body Disposal

The only significant difference in approaches was a greater relative likelihood of


single offenders to utilize a blitz attack (surprise with overpowering force)
(2(1) ¼ 5.88, p < 0.05). Single offenders were more likely to have the initial
contact, murder, and body disposal in the same location (2(1) ¼ 15.44,
p < 0.001); serial offenders, however, were more likely to use all different locations
for their crimes (2(1) ¼ 8.02, p < 0.01). Generally, there were few differences in
event sites between single and serial homicide offenders. Single offenders were
statistically more likely to murder (2(1) ¼ 9.99, p < 0.01) and dispose of the
victim’s body (2(1) ¼ 6.94, p < 0.01) in the victim’s own home and to initially
contact the victim in a vehicle (2(1) ¼ 6.31, p < 0.05).
There were few significant differences regarding the manner and method of body
disposal between serial and single homicide offenders. There were no significant
differences in whether or not the body was moved: 31% of single homicide offenders
moved the body, whereas 36% of serial homicide offenders did so. Single homicide

Copyright # 2004 John Wiley & Sons, Ltd. Behav. Sci. Law 22: 325–343 (2004)
336 G. W. Kraemer et al.

offenders were more likely to leave the body with a lack of concern for discovery
(2(1) ¼ 6.01, p < 0.05). In contrast, serial offenders were more likely to leave the
body appearing to initial investigators as if death had been natural (2(1) ¼ 4.09,
p < 0.05); however, the group sizes on this difference are extremely small (zero
single, four serial). The only significant difference in method of body disposal was
the increased likelihood that serial homicide offenders dump the body in a remote
location (2(1) ¼ 13.76, p < 0.001).

Medical and Associated Crime Scene Evidence

Although there were few differences in forensic findings, the differences are
important. Single homicide offenders were statistically significantly more likely to
kill victims with guns (2(1) ¼ 7.88, p < 0.01); serial offenders were more likely to
use strangulation (15% single: 47% serial; 2(1) ¼ 32.67, p < 0.001), ligature
weapons (2(1) ¼ 5.44, p < 0.05), and their hands or feet (2(1) ¼ 6.68, p < 0.01).
There were no differences between the groups in defensive injuries (2(1) ¼ 1.81,
p > 0.05). Single homicide offenders were more likely to cause harm to the victim’s
torso (2(1) ¼ 10.40, p < 0.001); serial offenders were slightly more likely to cause
injury to the victim’s head (2(1) ¼ 5.83, p < 0.05) and anus (2(1) ¼ 3.89,
p < 0.05). The only difference in extent of trauma was the greater number of
gunshot wounds to victims of single homicide (t(1) ¼ 2.16, p < 0.05). Serial offen-
ders were significantly more likely to bind their victims (2(1) ¼ 4.27, p < 0.05) and
were much more likely to sexually assault their victims (27% single: 59% serial;
2(1) ¼ 27.23, p < 0.001).

Step-Wise Discriminant Function Analyses

Two discriminant function analyses tested the usefulness of crime scene variables in
discriminating between single and serial homicide events. Using victim and offender
characteristics, intent, and relationship variables, 72.2% of the records were
correctly classified into single or serial homicide (canonical correlation ¼ 0.51;
Wilks  ¼ 0.74, 2(4) ¼ 76.95, p < 0.001). The following variables entered the
equation: stranger relationship, sexual intent, psychotic intent, and unclear intent.
It would appear, therefore, that intent is more useful than demographic variables in
determining whether a homicide is serial in nature.
The second equation tested all variables. Percent correct classification increased
from 72.2% to 76.0% (canonical correlation ¼ 0.57; Wilks  ¼ 0.67, 2(5) ¼ 97.37,
p < 0.001). The following variables entered the equation: stranger relationship,
strangulation as cause of death, natural manner of disposal, all event location sites
the same, and victim of another race than African American or Caucasian.

Female Victims Only

To test whether differences between single and serial offenses would remain if the
victim gender were the same, we ran a stepwise discriminant function analysis using

Copyright # 2004 John Wiley & Sons, Ltd. Behav. Sci. Law 22: 325–343 (2004)
Comparing single and serial homicide offenses 337

female victims only. Offender characteristics, relationship, intent, approach, differ-


ences in event sites, manner of disposal, method of disposal, cause of death, and
location of injury variables were entered into the analysis.
The percent correct classification between single and serial offenses was 78.6%,
an improvement of 2.6% over the previous analysis (canonical correlation ¼ 0.63,
Wilks  ¼ 0.60, 2(8) ¼ 77.83, p < 0.001). The variables entering the analysis were,
in order, stranger relationship, strangulation, remote dumping method of disposal,
paramour relationship, all event sites different, ‘‘none of the above’’ method of
disposal, disposal in a building, and psychotic intent. According to the frequencies,
serial homicide offenses were more likely to be stranger relationship (serial, N ¼ 72
of 97 (74%); single, N ¼ 20 of 64 (31%)), involve strangulation (serial, N ¼ 54
(56%); single, N ¼ 19 (30%)), use three different event sites (serial, N ¼ 28 (29%);
single, N ¼ 7 (11%)), use remote disposal sites (serial, N ¼ 14 (14%); single, N ¼ 0),
‘‘none of the above’’ disposal sites (serial, N ¼ 29 (30%); single, N ¼ 11 (17%)), and
involve psychotic intent (serial, N ¼ 5 (5%); single, N ¼ 0). Single homicides, in
contrast, were more likely to occur between paramours than serial homicides (serial,
N ¼ 0; single, N ¼ 11 (17%)), and were more likely to involve building disposal sites
(serial, N ¼ 28 (29%); single, N ¼ 22 (34%)).
Next, we calculated the relative frequency of intent (sex, profit, or emotion) by
type of offense (single or serial) for female victims. Sexual intent was more frequent
in serial crimes, although not significantly so (serial, N ¼ 64 (66%); single, N ¼ 34
(53%); 2(1) ¼ 2.68, p < 0.10). Profit was rare in each category (serial, N ¼ 4 (4%);
single, N ¼ 5 (8%); 2(1) ¼ 0.99, p ¼ ns). Emotion-based crimes were more fre-
quent in single homicides (serial, N ¼ 11 (11%); single, N ¼ 24 (38%); 2(1) ¼
15.51, p < 0.001).

DISCUSSION

Demographic data from our sample of serial offenders were consistent with the
observation that serial homicide offenders are young, white males (Egger, 1998;
Godwin, 2000; Hickey, 1997). Comparing sample data with census and UCR data
adds more depth to the picture. This sample of serial offenders was somewhere
between the proportion of races found in the population and the UCR: Most were
Caucasian, some were African American (although African Americans were over-
represented in comparison to Census data) and fewer were other minorities.
Although homicide is more frequently committed by men than by women (90%
male to 10% female), serial homicide is even more male dominated (95% male to
5% female).
According to available research, victims of serial homicide are young, vulnerable,
Caucasian women (Egger, 1998; Godwin, 2000; Hickey, 1997). This sample did
contain more females than males, contrasting with homicide generally, where men
are more often victims. Fewer victims were Caucasian than predicted based on
census data, but more were Caucasian than predicted based on UCR data. Thus, in
this sample, serial offenders targeted females more than males, and Caucasians were
not overselected.
Comparing serial and single homicide offenses allowed three meaningful dis-
criminators between single and serial homicide: stranger relationship, death by

Copyright # 2004 John Wiley & Sons, Ltd. Behav. Sci. Law 22: 325–343 (2004)
338 G. W. Kraemer et al.

strangulation, and similarity in event locations. Although leaving a body as if death


were natural did enter the predictive equation, it is not a practically significant
predictor. Serial cases in which the body was left as if death were natural were
usually intrafamilial homicides. In such cases, criminal investigation usually begins
after a suspicious pattern of natural-appearing deaths is discovered. Thus, this factor
is not as useful to the typical homicide investigation, which typically begins when a
body is discovered. Prediction improved when victims were separated by gender.
Using a remote location for disposing the body is significantly predictive of serial
offenders. Also, when considering only female victims, distinctions in intent and
relationship became significantly predictive—single homicide offenders tended to
kill people they knew, usually out of anger, but serial offenders tended to kill
strangers for sexual reasons. Serial homicide offenders exhibited more planning by
moving the victim or body from one location to another, by using restraints, and by
disposing of the body in a remote location. Although not predictive in the
discriminant function equation, use of weaponry differed between serial and single
offenders. Serial offenders were more ‘‘hands on,’’ killing through strangulation or
beating with hands or feet, while single offenders more frequently used guns.
Although the results of this study tend to support some of the classical char-
acteristics attributed to serial versus single homicide offenders and offenses,
investigators and other criminal justice practitioners should remain aware of the
complex and multifaceted motivations and behaviors exhibited by those who kill.
Our findings generally support the differentiation of serial homicides from other
types of murder by their deliberate, premeditated, sexually predatory nature, and by
the lack of interpersonal conflict and provocation common in nonserial offenses
(Brantley & Kosky, in press). No set of empirical criteria however, is absolutely
predictive.
Investigators and other criminal justice practitioners should utilize these research
findings as an additional tool in the arsenal of available information targeted at
identifying potential serial homicide cases and separating these events from non-
serial murders. Homicide cases should be considered particularly suspect when they
involve female victims, murdered by strangulation, with apparent sexual intent,
whose remains are transported and disposed of in remote locations, who display
evidence of the use of restraints, and who have no overt indications of a pre-existing
relationship with their killers. Such cases should be investigated rigorously and the
investigative findings shared with other law enforcement agencies. Additionally,
these and other potential serial homicide cases should be expeditiously entered into
regional, state, and national homicide linkage databases for subsequent comparative
analysis.
Like all research, this study has its share of limitations. Because the study was
archival and the cases were not collected for the purposes of research, some cases
have missing data and reliability is less certain. The sample size was smaller than
ideal, although larger than any previous study of its particular focus. The compar-
ison group provided a useful counterpoint to serial homicide data, but the group was
not representative of homicide generally, as demographic data from the single
sample differed from that reported in the UCR. These differences are likely an
artifact from the source of data: reporting cases to the FBI is voluntary. Although
some states automatically transfer data on all homicides from state systems to the
FBI, the body of cases submitted may reflect the judgment of individual law

Copyright # 2004 John Wiley & Sons, Ltd. Behav. Sci. Law 22: 325–343 (2004)
Comparing single and serial homicide offenses 339

enforcement officers that the case warranted submission to the FBI, perhaps
because the case was particularly unusual or difficult. The study remains unique,
however, in its unprecedented utilization of detailed case specific investigative
information as a basis for the comparative empirical analysis of a relatively large
sample of serial and single homicide offenses.
Again, although this study offers comparative empirical support for some of the
traditional assertions present in the serial homicide literature, it also shows that
comparative empirically based analyses are both possible and needed. It further
demonstrates that such studies can produce scientifically valid results relevant to
both theoretical debate and practical criminal investigation. It also shows that
comparative analysis is possible and can yield relevant results. There is sufficient
information on serial homicide to permit researchers to advance beyond case studies
and conclusions unsupported by empirical evidence, bringing scientific rigor to this
important field of study.

REFERENCES
Alison, L., Bennell, C., Mokros, A., & Ormerod, D. (2002). The personality paradox in offender
profiling: A theoretical review of the processes involved in deriving background characteristics from
crime scene actions. Psychology, Public Policy, and Law, 8(1), 115–135.
Ault, R. L., & Reese, J. T. (1980). A psychological assessment of crime profiling. FBI Law Enforcement
Bulletin, 49, 22–25.
Borum, R. (1996). Improving the clinical practice of violence risk assessment. American Psychologist, 51,
945–956.
Brantley, A. C., & Kosky, R. H. (in press). Serial murder in the Netherlands. FBI Law Enforcement
Bulletin.
Brittain, R. P. (1970). The sadistic murderer. Medicine, Science and the Law, 10, 198–207.
Cloud, M. Y. (1996). Abducted and murdered children: Behavioral based analyses of victims, offenders, and
remains disposal methods [dissertation]. Los Angeles, CA: University of California, Los Angeles.
Daly, M., & Wilson, M. (1988). Homicide. New York: Aldine de Gruyter.
DeHart, D. D., & Mahoney, J. M. (1994). The serial murderer’s motivations: An interdisciplinary review.
Omega—Journal of Death and Dying, 29, 29–45.
Dietz, P. E. (1985). Sex offender profiling by the FBI: A preliminary conceptual model. In M. H.
Ben-Aron, S. J. Hucker, & C. D. Webster (Eds.), Clinical criminology: The assessment and treatment of
criminal behavior (pp. 207–219). Toronto: Clarke Institute of Psychiatry.
Dietz, P. E. (1986). Mass, serial and sensational homicide. Bulletin of the New York Academy of Medicine,
62, 477–490.
Douglas, J. E., Ressler, R. K., Burgess, A. W., & Hartman, C. R. (1986). Criminal profiling and crime
scene analysis. Behavioral Sciences and the Law, 4, 401–421.
Egger, S. (1990). Serial murder—An elusive phenomenon. New York: Praeger.
Egger, S. (1998). The killers among us: An examination of serial murder and its investigation. Englewood
Cliffs, NJ: Prentice-Hall.
Federal Bureau of Investigation. (1985, August). Classifying sexual homicide crime scenes: Interrater
reliability. FBI Law Enforcement Bulletin, 13–17.
Federal Bureau of Investigation. (1994). Child Abduction and Serial Killer Unit, Morgan P. Hardiman Task
Force on Missing and Exploited Children [brochure].
Fox, J. A., & Levin, J. (1994). Overkill: Mass murder and serial killing exposed. New York: Plenum.
Fox, J. A., & Levin, J. (1999). Serial murder: Popular myths and empirical realities. In M. D. Smith, &
M. A. Zahn (Eds.), Homicide: A sourcebook of social research (pp. 165–175). Thousand Oaks, CA: Sage.
Godwin, G. M. (2000). Hunting serial predators: A multivariate classification approach to profiling violent
behavior. Boca Raton, FL: CRC Press.
Greswell, D. M., & Hollin, C. R. (1994). Multiple murder: A review. The British Journal of Criminology,
34, 1–14.
Grove, W. M., & Meehl, P. E. (1996). Comparative efficiency of informal (subjective, impressionistic)
and formal (mechanical, algorithmic) prediction procedures: The clinical–statistical controversy.
Psychology, Public Policy and the Law, 2, 293–323.

Copyright # 2004 John Wiley & Sons, Ltd. Behav. Sci. Law 22: 325–343 (2004)
340 G. W. Kraemer et al.

Hazelwood, R. R., & Douglas, J. E. (1980). The lust murderer. FBI Law Enforcement Bulletin, 18–22.
Hickey, E. (1990). The etiology of victimization in serial murder: An historical and demographic analysis.
In S. Egger (Ed.), Serial murder: An elusive phenomenon (pp. 53–71). New York: Praeger.
Hickey, E. (1997). Serial murderers and their victims (2nd ed.). Belmont, CA: Wadsworth.
Holmes, R. M., & DeBurger, J. (1988). Serial murder. Newbury Park, CA: Sage.
Homant, R. J., & Kennedy, D. B. (1998). Psychological aspects of crime scene profiling: Validity
research. Criminal Justice and Behavior, 25, 319–343.
Jenkins, P. (1988). Serial murder in England 1940–1985. Journal of Criminal Justice, 16, 1–15.
Keeney, B., & Heide, K. M. (1994). Gender differences in serial murderers: A preliminary analysis. The
Journal of Interpersonal Violence, 9, 383–398.
Keeney, B. T., & Heide, K. M. (1995). Serial murder: A more accurate and inclusive definition.
International Journal of Offender Therapy and Comparative Criminology, 39, 298–306.
Langevin, R., Ben-Aron, M. H., Wright, P., Marchese, V., & Handy, L. (1988). The sex killer. Annals of
Sex Research, 1, 263–301.
Lord, W. D., Boudreaux, M. C., Jarvis, J. P., Waldvogel, J., & Weeks, H. (2002). Comparative patterns in
life course victimization: Competition, social rivalry, and predatory tactics in child homicide in the
United States. Homicide Studies, 6(4), 325–347.
McKenzie, C. (1995). A study of serial murder. International Journal of Offender Therapy and Comparative
Criminology, 39, 3–10.
Pinizzotto, A. J., & Finkel, N. J. (1990). Criminal personality profiling: An outcome and process study.
Law and Human Behavior, 14, 215–233.
Pollock, P. H. (1995). A case of spree serial murder with suggested diagnostic opinions. International
Journal of Offender Therapy and Comparative Criminology, 39, 258–268.
Protection of Children from Sexual Predators Act of 1998, 28 U.S.C. § 540B (2) (2000).
Ressler, R. K., Burgess, A. W., & Douglas, J. E. (1988). Sexual homicide: Patterns and Motives. New York:
Lexington.
Ressler, R. K., Burgess, A. W., Douglas, J. E., Hartman, C. R., & D’Agostino, R. (1986). Sexual killers
and their victims: Identifying patterns through crime scene analysis. Journal of Interpersonal Violence, 1,
288–308.
U.S. Census Bureau. (2000). DP-1. Profile of general demographic characteristics: 2000: Census 2000
Summary File 1 (SF 1). Retrieved 18, November 2002, from http://factfinder.census.gov.
United States Department of Justice (USDOJ), Federal Bureau of Investigation. (2001). Crime in the
United States: Uniform Crime Report for the United States 2000. Washington, DC: U.S. Government
Printing Office.
Warren, J. I., Hazelwood, R. R., & Dietz, P. E. (1996). The sexually sadistic serial killer. Journal of
Forensic Sciences, 41, 970–974.
Wilson, P., Lincoln, R., & Kocsis, R. (1997). Validity, utility and ethics of profiling for serial violent and
sexual offenders. Psychiatry, Psychology and Law, 4, 1–12.

Copyright # 2004 John Wiley & Sons, Ltd. Behav. Sci. Law 22: 325–343 (2004)
Comparing single and serial homicide offenses 341

APPENDIX
Table A1. Comparison of single and serial homicide: frequencies and chi square analyses

Source Single Serial 2

Offender characteristics
Gender
Female 15 (10%) 6 (4%) 4.19*
Male 129 (90%) 130 (96%) 0.15
Age
<20 6 (5%) 5 (4%) 0.10
20–29 74 (56%) 77 (58%) 0.14
30–39 39 (29%) 42 (32%) 0.16
40–49 10 (8%) 7 (5%) 0.57
50þ 4 (3%) 2 (2%) 0.68
Race
African American 63 (47%) 55 (41%) 0.98
Caucasian 64 (48%) 73 (55%) 1.22
Other 6 (4%) 5 (4%) 0.10
Victim characteristics
Gender 17.14***
Male 69 (52%) 36 (27%)
Female 64 (48%) 97 (73%)
Age
1–12 9 (7%) 8 (6%) 0.06
13–18 12 (9%) 18 (14%) 1.35
19–29 41 (31%) 49 (37%) 1.08
30–39 31 (23%) 21 (16%) 2.39
40–49 21 (16%) 10 (8%) 4.42*
50–59 5 (4%) 11 (8%) 2.39
60–69 5 (4%) 4 (3%) 0.12
70þ 9 (7%) 9 (7%) 0.00
Race
African American 42 (32%) 41 (31%) 0.02
Caucasian 77 (58%) 84 (63%) 0.77
Other 14 (11%) 7 (5%) 2.53
Vulnerability 55 (41%) 64 (48%) 1.23
Relationship
Family 10 (8%) 6 (5%) 1.06
Paramour 14 (11%) 2 (2%) 9.58**
Acquaintance 66 (50%) 28 (21%) 23.76***
Stranger 40 (30%) 94 (71%) 43.85***
Unclear 3 (2%) 3 (2%) 0.00
Intent
Sex 38 (29%) 77 (58%) 23.30***
Profit 9 (7%) 5 (4%) 1.21
Emotion 51 (38%) 21 (16%) 17.14***
Psychotic 0 6 (4%) 6.14*
Unclear 14 (11%) 23 (17%) 2.54
Approach
Unknown 56 (42%) 62 (47%) 0.55
Deception 26 (20%) 36 (26%) 1.72
Surprise 6 (5%) 10 (8%) 1.06
Blitz 48 (36%) 30 (23%) 5.88*
Location
All the same 79 (59%) 47 (35%) 15.44***
Recovery 7 (5%) 7 (5%) 0.00
Initial contact 33 (25%) 43 (32%) 1.84
All different 13 (10%) 30 (23%) 8.02**

Continues

Copyright # 2004 John Wiley & Sons, Ltd. Behav. Sci. Law 22: 325–343 (2004)
342 G. W. Kraemer et al.

Table A1. Continued

Source Single Serial 2

Body recovery
Victim residence 45 (34%) 26 (20%) 6.94**
Any other residence 14 (11%) 14 (11%) 0.00
School 2 (2%) 6 (5%) 2.06
Commercial 7 (5%) 11 (8%) 0.95
Street 17 (13%) 17 (13%) 0.00
Vice area 6 (5%) 2 (2%) 2.06
Woods 21 (16%) 27 (20%) 0.92
Field 8 (6%) 10 (8%) 0.24
Vehicle 5 (4%) 2 (2%) 1.32
Other 1 (1%) 4 (3%) 4.06*
Murder
Victim residence 48 (36%) 25 (19%) 9.99**
Any residence 20 (15%) 15 (11%) 0.82
School 3 (2%) 5 (4%) 0.52
Commercial 9 (7%) 6 (5%) 0.64
Street 15 (11%) 13 (10%) 0.16
Vice area 6 (5%) 1 (1%) 3.67
Woods 9 (7%) 21 (16%) 5.41*
Field 6 (5%) 7 (5%) 0.08
Vehicle 7 (5%) 1 (1%) 4.64*
Other 1 (1%) 7 (5%) 4.64*
Initial contact
Victim residence 45 (34%) 37 (28%) 1.13
Any other residence 18 (4%) 13 (10%) 0.91
School 4 (3%) 2 (2%) 0.68
Commercial 23 (17%) 18 (14%) 0.72
Street 17 (13%) 33 (25%) 6.31*
Vice area 7 (5%) 11 (8%) 0.95
Woods 2 (2%) 1 (1%) 0.34
Field 4 (3%) 1 (1%) 1.83
Vehicle 10 (8%) 1 (1%) 7.68**
Other 1 (1%) 4 (3%) 1.83
Body disposal
Moved 41 (31%) 10 (8%) 1.53
Manner of disposal
Displayed 9 (7%) 10 (8%) 0.07
Concealed 32 (25%) 44 (33%) 2.79
Lack of concern 92 (69%) 72 (54%) 6.01*
Left as natural 0 4 (3%) 4.09*
Method of disposal
Buried 10 (8%) 5 (4%) 1.77
Covered 15 (11%) 16 (12%) 0.04
In water 11 (8%) 8 (6%) 0.51
Building 51 (38%) 40 (30%) 2.02
Container 3 (2%) 7 (5%) 1.66
Vehicle 7 (5%) 2 (2%) 2.88
Scattered 1 (1%) 1 (1%) 0.00
None 37 (28%) 39 (29%) 0.07
Remote 3 (2%) 20 (15%) 13.76**
Body missing 0 1 (1%) 1.00
Cause of death
Gunshot wound 40 (30%) 20 (15%) 7.88**
Stab wound 35 (26%) 20 (15%) 4.62*
Cutting wound 11 (8%) 8 (6%) 0.42
Blunt force injury 37 (28%) 33 (25%) 0.17
Strangulation 20 (15%) 62 (47%) 32.67***
Smothering 5 (4%) 6 (4%) 0.12

Continues

Copyright # 2004 John Wiley & Sons, Ltd. Behav. Sci. Law 22: 325–343 (2004)
Comparing single and serial homicide offenses 343

Table A1. Continued

Source Single Serial 2

Poison 0 2 (2%) 2.06


Undetermined 4 (3%) 6 (5%) 0.48
Defensive injuries 22 (38% of known) 17 (27% of known) 1.81
Location of injury
Head 95 (71%) 109 (82%) 5.83*
Torso 68 (51%) 41 (31%) 10.40***
Arms 23 (17%) 19 (14%) 0.35
Legs 14 (11%) 7 (5%) 2.37
Breasts 6 (5%) 11 (8%) 1.70
Genitalia 4 (3%) 10 (8%) 2.86
Buttocks 2 (2%) 2 (2%) 0.00
Anus 2 (2%) 8 (6%) 3.89*
Extent of Injuries (t test)
M blunt force trauma 1.76 (1.90) 1.65 (1.98) 0.45
M No. stab wounds 3.74 (10.10) 2.99 (10.13) 0.60
M No. cutting wounds 0.98 (3.91) 1.20 (4.60) 0.41
M No. gunshot wounds 0.69 (1.56) 0.35 (0.95) 2.16*
Weapon used
Gun 41 (31%) 22 (17%) 6.98**
Blade 41 (31%) 31 (23%) 1.61
Club 25 (19%) 24 (18%) 0.01
Ligature 13 (10%) 26 (19%) 5.44*
Personal 50 (38%) 70 (53%) 6.68**
Other 15 (11%) 8 (6%) 2.16
Other characteristics
Carving 4 (3%) 5 (4%) 0.13
Symbolic 7 (5%) 7 (5%) 0.00
Victim bound 21 (16%) 34 (26%) 4.27*
Clothing present
Fully dressed 76 (57%) 52 (39%) 8.27**
Partially dressed 38 (29%) 47 (35%) 1.65
Nude 15 (11%) 26 (20%) 3.72*
Bitemarks 4 (3%) 3 (2%) 0.12
Unusual torture 21 (16%) 17 (13%) 0.42
Sexual assault 36 (27%) 78 (59%) 27.23***

All tests have one degree of freedom.


*p < 0.05; **p < 0.01; ***p < 0.001.

Copyright # 2004 John Wiley & Sons, Ltd. Behav. Sci. Law 22: 325–343 (2004)

You might also like