Файл: [D._Kim_Rossmo]_Geographic_Profiling(BookFi.org).pdf

© 2000 by CRC Press LLC

Figure 9.3

 Serial murder by day of week

.

Figure 9.4

    Distance to crime site.

© 2000 by CRC Press LLC

otherwise it drops to 1.6. This appears to suggest a desire on the part of
offenders to distance the remains of victims encountered relatively close to
home.

While only 61% of all sites were known to police, this figure increased

to 67% for body dump sites and 70% for encounter sites. These two location
types are the most important in geographic profiling, and it is usually suffi-
cient to know one or the other for the purposes of a criminal geographic
targeting analysis (see Table 10.7). Approximately 12% of the crime sites were
within the killer’s residence. Such locations are likely unknown to police.

Multiple responses were allowed for site description, and 238 encounter site

types, 144 body dump site types, and 476 total site types were recorded. Streets
and residences are the most common crime locations. The “other” classification
typically refers to deserted lots or waste ground. Residential and commercial
land use predominated. The majority of incidents occurred in outside public
places, followed by inside private places. Serial murderers prefer to travel by
vehicle.

9.4.4 Crime Parsing

A crime is often treated as having a single location, but depending upon crime
type there may be various sites connected to a single offence. These have
different meanings to the offender and, consequently, distinctive choice prop-
erties. For serial murder these location types include victim encounter, attack,
murder, and body dump sites. For serial rape they include victim encounter,
attack, rape, and victim release sites. Serial arson normally involves only one
location, the fire setting site. While these particular actions can all occur in a
single place, the majority of cases involve two or more locations.

Eight possible combinations, called crime location sets, result from the

four different murder site types. Breaking an offence down into its crime
location set is referred to as crime parsing. While the specific location set for
a given crime is a function of victim selection and encounter site character-
istics, it also implies something about the offender’s mobility, search strategy,
and level of organization. Generally, the greater the organization and mobility
of an offender, the greater the potential complexity (i.e., the more separate
locations) of the crime location set.

Table 9.7

 presents the percentage breakdown for the eight crime location

sets. Movement between victim encounter (

E

), attack (

A

), murder (

M

), and

body dump (

D

) sites are represented by an arrow (

→

). For example, 

EAM

→

D

indicates that the encounter, attack, and murder sites were in the same place,
but the body dump site was in a different location.

There was a high level of geographic consistency in the M.O. of this

sample as most offenders repeatedly employed the same crime location set.
Approximately 85% of the total number of serial murder victims (n = 178)

© 2000 by CRC Press LLC

fell into the most common crime location set used by their killer, and 96%
into either the first or second most common crime location set. This implies
that crime location set might be used as an assessment characteristic for the
linking of serial offences.

59

An understanding of consistency, change, and progression in offender

behaviour is an important principle of linkage analysis. Warren et al. (1995)
found about half of the 119 quantified behaviours they examined from serial
rape crime scenes remained consistent over time. Other offender behaviours
either showed progression (e.g., degree of planning, protection of identity,
and use of bindings), or exhibited inconsistent change. They suggest the
pathological aspects are more constant. “The idea of consistency and change
represents an important area of serial crime: in theory, it helps to define
relevant dimensions of classificatory paradigms, which can inform investi-
gative efforts to link crimes perpetrated by the same offender” (p. 255).
Measures of consistency and change in offender geographic behaviour assist
in this effort (see, for example, Dettlinger & Prugh, 1983).

9.4.5 Clusters

Short-term spatial selectivity has been observed in the hunting behaviour of
animals and certain predators repeatedly visit the same forage sites (Smith,
1974a, 1974b; Smith & Sweatman, 1974). Geotropism is also found in serial
killers, many of whom return to favoured sites to hunt victims, or dispose
of bodies in cluster dumps or forest “graveyards” (Newton, 1992). Beyond
convenience, these private “totem places” may also be significant to the
offender’s fantasies.

Table 9.7    Crime Location Sets

Crime Location Set

Percentage

Number of Sites

Percentage

E 

→

 A 

→

 M 

→

 D

1.7%

4

1.7%

E 

→

 A 

→

 MD

1.7%

3

26.4%

E 

→

 AM 

→

 D

21.3%

EA 

→

 M 

→

 D

3.4%

EA 

→

 MD

1.1%

2

33.1%

E 

→

 AMD

29.2%

EAM 

→

 D

2.8%

EAMD

38.8%

1

38.8%

Total

100%

100%

59 

The design of ViCLAS incorporates certain geographic profiling principles, and queries

can be based on similarities in crime location sets in conjunction with other search criteria.

© 2000 by CRC Press LLC

The Yorkshire Ripper and the Son of Sam were both known to revisit

the areas of their previous crimes in pursuit of new victims (Ressler &
Shachtman, 1992).

When we went to New York to talk to the ‘Son of Sam,’ David Berkowitz,”
says [FBI Special Agent] Robert K. Ressler … “he told us that on the nights
when he couldn’t find a victim to kill he would go back to the scene of an
old crime to relive the crime and to fantasize about it. Now that’s a heck of
a piece of information to store somewhere to see whether other offenders
do the same thing. (Porter, 1983, pp. 49–50)

LeBeau (1985) notes “the proclivity of chronic serial offenders to use

repeatedly the same geographic and ecological space ... The geographical and
ecological patterning of the serial offender may be tangible information
which is discerned by police investigators and utilized in apprehension” (p.
397). He determined through nearest neighbour analysis that the mean dis-
tance between crime scenes for chronic serial rapists in San Diego varied
from 0.12 to 0.85 miles, averaging 0.35 miles (1986). Büchler and Leineweber
(1991) observe bank robbers in Germany follow similar patterns of escape,
information which may help police connect crimes. Davies and Dale (1995b)
found several cases of geographic “backtracking” in their study of British
rapists. They propose national access to intelligence information on former
crime sites, and people and places of significance to an offender is an impor-
tant investigative resource. Serial killer Monte Rissell was apprehended only
after he returned to his former rape sites in Alexandria, Virginia (Ressler &
Shachtman, 1992).

An analysis of the pattern of crime sites in the SFU serial murder study

showed a tendency towards aggregation (59% of 

R

 scale values were smaller

than 1). One of the main influences on divergence from randomness is target
backcloth; not surprisingly, a lack of uniformity leads to clustering. Another
factor appears to be opportunity. If an offender was successful in a particular
neighbourhood once, then why not again. Other potential targets may have
been noticed and remembered during the commission of the first crime. In
many ways, these influences are similar to those documented in repeat vic-
timization studies of property crime (Farrell & Pease, 1993; Pease & Laycock,
1996). Fantasy also plays a role in drawing an offender back to a particular
location. A crime site situated close to a previous offence is referred to as a
contagion location, and is generally regarded for the purposes of geographic
profiling as a nonindependent event.

© 2000 by CRC Press LLC

9.4.6 Trip Distance Increase

While it has been suggested individual offender crime trips increase in dis-
tance over time, there has been little empirical testing of this hypothesis. In
an attempt to see what influence, if any, time had on the journey to crime,
distances travelled by serial murderers, both individual and collective analyses
were conducted. This approach is limited, however, as any particular journey
to crime may not have originated from the offender’s residence. Also, a killer
could have been responsible for unknown murders, in addition to preceding
crimes such as burglary, robbery, or sexual assault. And some offenders
moved residence in the midst of their murder series, complicating journey-
to-crime comparisons.

The Manhattan distance from offender residence to location of victim

encounter was measured for every known crime in each killer’s series. Where
two or more victims were attacked during the same day, only the first crime
was counted. The offences were chronologically ordered, outliers excluded,
and significant gaps ignored. Distance from offender residence to encounter
site was plotted against crime number (see Rossmo, 1995a).

Table 9.8

 summarizes these results. Half of the serial murder cases showed

a significant increase (defined as a slope with an 

R

2

 > 0.300), while the other

half exhibited no significant change; none showed a significant decrease in
crime trip distance. Average journeys varied considerably by offender (rang-
ing from 1 to 40 km), therefore slopes were compared to each killer’s mean
crime trip distance. This figure represents the incremental increase in crime
trip distance expressed as a proportion of the mean crime trip distance. The
overall mean percentage increase is 16% (standard deviation = 17).

Figure 9.5

 shows crime trip distance changes in the aggregate. Wide

variations in hunting range distort the pooled data, therefore the
unweighted mean natural logarithm of crime trip distance was plotted
against offence number. Only those cases with a minimum of five crime
trips were included in the analysis. When a linear trend line is fitted to the
data, a significant increase over time is observed (

R

2

 = 0.519). A better fit

is found, however, with a third-order polynomial expression (

R

2

 = 0.880),

as shown in 

Figure 9.5

.

Table 9.8    Crime Trip Distance Increase

Serial Murderer

Slope

Crime Trip Distance (km) 

Minimum Mean Maximum

Proportion

Mean

2.66

4.47

24.44

57.95

0.16

Standard Deviation

3.97

7.39

13.20

50.31

0.17