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

© 2000 by CRC Press LLC

idence, and victim residence spatial conjunction (Amir, 1971; Normandeau,
1968). Crime journey research findings are often presented in the form of
the percentage of offences that fit into the neighbourhood triangle (Plough-
man & Ould, 1990; Rand, 1986).

Research using neighbourhood or mobility triangles raises the question

of how to define the concept of “neighbourhood.” Census tract delineations
are often used, but this is a problematic assumption. Census tracts are only
rough approximations of neighbourhoods and more sophisticated concepts
of mobility, routine activities, and target selection are available (e.g., Brant-
ingham & Brantingham, 1981; Felson, 1986). Other studies do not define
the concept of neighbourhood, but rather use the subjective interpretations
of respondents (see, for example, DeFrances & Smith, 1994).

The most useful presentation of crime trip data is the distance-decay

approach, a graphical curve that shows the number of trips for several dif-
ferent radii (e.g., half-mile increments) from the offender’s residence (see,
for example, Capone & Nichols, 1976; Rhodes & Conly, 1981; also Baldwin
& Bottoms, 1976). Such formats allow for an inspection of the distance-decay
function, providing more information for further analysis and a fuller under-
standing of the nature of crime journeys.

Van Koppen and de Keijser (1997) have questioned the accuracy of

distance decay findings in the journey-to-crime literature. Using randomly
generated data that lacked distance decay, they were able to show the aggre-
gation of individual crime trips led to a distance-decay result. Their main
point is correct. Individual behaviour cannot be safely inferred from aggre-
gate data, an error referred to as the ecological fallacy.

Research at a certain spatial level of analysis may produce conclusions

that, while valid at that scale, are invalid at different levels (Goodall, 1987).
When research arguments are made by relating results derived from one
geographic level of analysis to another, an ecological fallacy occurs. This
usually involves the application of correlates derived from areal data to indi-
viduals, though an ecological fallacy can occur by moving in either direction
through the spatial analytic framework (Brantingham & Brantingham, 1984).
The scale problem, as geographers refer to it, can be a serious difficulty for
attempts to generalize from geographical areal research as changes in focus
may mask the true nature of relationships (Langbein & Lichtman, 1978;
Taylor, 1977).

For example, combining individual crime trip distances without first

standardizing the data

35

 often obscures the existence of the buffer zone. The

method of van Koppen and de Keijser, however, actually created non-random
distances with a bias towards shorter crime trips, therefore introducing

35 

One method of standardization is to divide every individual crime trip distance by the

offender’s mean crime trip distance.

© 2000 by CRC Press LLC

distance decay (Rengert, Piquero, & Jones, 1999). True random data would
show a uniform distribution. They also did not distinguish between studies
of destination and origin distance decay. Finally, they ignored research that
properly established distance decay in serial criminals (e.g., Davies & Dale,
1995b; Rossmo, 1995b; Sapp et al., 1994; Warren, Reboussin, Hazelwood,
Cummings, Gibbs, & Trumbetta, 1998).

It has been suggested that as an offender’s criminal career matures, jour-

ney-to-crime distances lengthen and size of hunting area increases (Brant-
ingham & Brantingham, 1981; Canter & Larkin, 1993). After the arrest of
David Berkowitz, police searching his apartment “found maps of Connecti-
cut, New York, and New Jersey, marked and annotated in such a way that
investigators took them as evidence that Berkowitz was planning to extend
his killing grounds” (Time-Life Books, 1992b, p. 179). Hungarian Sylvestre
Matuschka killed 22 people and injured 75 more in engineered train wrecks
(Nash, 1992).  Upon his apprehension in 1932, authorities seized railway
schedules and maps for France, Italy, and The Netherlands – all part of his
plan to cause future wrecks monthly (Seltzer, 1998).

The FBI believes that the first attack in a serial murder series is the one

most likely to be closest to the offender’s home (Warren et al., 1995). Both
Barrett (1990) and Canter (1994) note that the first crime of a serial offender
may be spontaneous and impulsive, and victim selectivity tends to decrease
over time (see, for example, the discussion on changes in Jeffrey Dahmer’s
murder pattern, in Ressler & Shachtman, 1992). These observations indicate
the importance of a temporal dimension in the analysis of spatial crime
patterns (see Kind, 1987a; LeBeau, 1992; Newton & Newton, 1985; Newton
& Swoope, 1987).

In an FBI study of travel distances and offence patterns for 108 U.S. serial

rapists responsible for a total of 565 offences, Warren et al. (1995) observed
that the rape closest to the offender’s residence was the first in 18%, but the
fifth in 24% of the cases. A British study of 79 rapists and 299 sexual offences
(Davies & Dale, 1995a) found no significant difference in distance between
first and last crime trip for prolific offenders (those with five or more rapes).
Noting that habitual burglars and robbers travelled longer distances, Davies
and Dale (1995a) suggest that for those rapists with a history of break and
entry, the first rape in a series might well be their fiftieth housebreaking. Any
maturation in crime pattern would therefore have occurred long ago. This
problem is compounded by the fact that sexual assaults are notoriously
underreported, and the first known rape might actually be the offender’s
second or third (Davies & Dale; Leyton, O’Grady, & Overton, 1992). Such a
misinterpretation happened in the Vampire Killer case where Sacramento
County police believed the second murder was actually the first (Ressler &
Shachtman, 1992).

© 2000 by CRC Press LLC

The journey-to-crime concept implies a home base from which the trip

commenced, yet some offenders may not have a residence. Convicted crim-
inals also tend to be less residentially stable than noncriminals, and psycho-
paths in particular are nomadic. Rossmo found that a significant number of
criminal fugitives in Canada are willing to travel thousands of miles to avoid
prison (1987; Rossmo & Routledge, 1990). Marvell and Moody (1998)
observed that a small but highly active group of major criminals were
extremely mobile, for reasons that included fear of the police, conflicts with
other criminals, or general wanderlust. These offenders engaged in a succes-
sion of road trips involving temporary stays in different regions, or perma-
nent moves of residence every few months or years from one state to another.
It is therefore important not to confuse transiency with crime trips.

Travelling offenders are in the minority, and fewer than 10% of the

criminals Canter (1994) studied were of “no fixed abode” at the time of their
arrest. Davies and Dale (1995b) determined that 22% of those rapists for
whom they had such information were itinerant. Victims were confronted
at their homes in 41% of the cases,

36

 and within public areas (including

apartment building common areas) in 58% of the cases. They observed that
“some rapists were obviously drawn to areas where potential victims were
accessible, such as red-light districts ... The distance the offenders travelled
was clearly related to the proximity of their own residence to these locations”
(Davies & Dale, 1995a, p. 13).

These places of victim accessibility include both nodes (e.g., entrances

to train stations or apartment blocks), and routes used by females commuting
to work, school, shopping, and entertainment areas. Because the value of
these “victim hunting grounds” depended upon female activity level, their
desirability is influenced by time of day. Distinct clusters of contact sites were
noted, some associated with victim availability, others with residences of
people significant to the offender (Davies & Dale, 1995a). Ted Bundy’s FBI
wanted poster alerted people to his preferred target areas — beaches, ski
resorts, discotheques, and college campuses (Ressler & Shachtman, 1992).
Such finding are consistent with Brantingham and Brantingham’s (1981,
1993b) pattern theory and model of crime site selection, discussed in the
next section.

36 

House (1993) examined offence venue for 61 Canadian cases of stranger sexual assault

committed by 30 offenders (40% of whom were serial rapists, each responsible for an
average of 2.8 crimes). He found 61% of the crimes occurred outdoors and 39% inside (38%
of which were in the victim’s residence).

Table 7.1   Journey–to–Crime Research

3

Source

Crime

Location

Year

Crime Trip Distance

Comments

Aitken et al. (1994)

sex motivated child 

murders

Great Britain

1960–1991

91.6% < 5 mi

> 5 mi if offender travel or 

victim abduction 
indicated

Alston (1994)

stranger serial sexual 

assault

British Columbia

1977–1993

31.1% < 0.5 km; 44.4% < 1 

km; 55.6% < 1.5 km; 60.0% 
< 2 km; 75.6% < 3 km

distance to nearest 

offender activity node

Amir (1971)

rape

Philadelphia

1958–1960

72% within home area (5 

blocks)

mobility triangles

Baldwin & Bottoms 

(1976)

property crime

Sheffield

1966

47% < 1 mi; 69% < 2 mi

Baldwin & Bottoms 

(1976)

breaking offence

Sheffield

1966

54.4% < 1 mi; 74.8% < 2 mi

Baldwin & Bottoms 

(1976)

larceny offence

Sheffield

1966

51.9% < 1 mi; 74.3% < 2 mi

Baldwin & Bottoms 

(1976)

taking & driving 

offence

Sheffield

1966

45% < 1 mi; 63.3% < 2 mi

Boggs (1965)

homicide & assault

St. Louis

most likely within residential 

area

Boggs (1965)

rape & robbery

St. Louis

most likely within 

nonresidential area

Bullock (1955)

homicide

Houston

1945–1949

40% < 1 block; 57% < 0.4 mi; 

74% < 2 mi

Canter & Hodge 

(1997)

serial murder

U.S.

40 km; body dump site: 9 

km/90 km mean min./max. 
(25%< 5 km; 50% < 15 km)

89% marauders; 11% 

commuters

Canter & Hodge 

(1997)

serial murder

Britain

24 km; body dump site: 6 

km/36 km mean min./max.

86% marauders; 14% 

commuters

© 2000 by CRC Press LLC

Table 7.1   Journey–to–Crime Research (Continued)

Source

Crime

Location

Year

Crime Trip Distance

Comments

Canter & Larkin 

(1993)

serial  rape

Greater London & 

SE England

1980s

1.53 mi mean min. crime trip 

distance

87% marauders; 13% 

commuters

Capone & Nichols 

(1976)

robbery

Miami

1971

1/3 < 1 mi; 1/2 < 2 mi; 2/3 < 

3 mi

Capone & Nichols 

(1976)

armed robbery

Miami

1971

26% < 1 mi; 45% < 2 mi; 59% 

< 3mi

Capone & Nichols 

(1976)

unarmed robbery

Miami

1971

36% < 1 mi; 60% < 2 mi; 75% 

< 3mi

Chappell (1965)

burglary

England

1965

50%/85% < 1 mi (< 21/14 

years)

Davies & Dale 

(1995b)

stranger rape

England

1965–1993

17% < 0.5 mi; 29% < 1 mi; 

52% < 2 mi; 60% < 3 mi; 
69% < 4 mi; 76% < 5 mi

approach site; 72%/24% < 

1.8 mi (</> 26 years)

DeFrances & Smith 

(1994)

all offences

U.S.

1991

43% in own neighbourhood 

(violent crime 44.7%; 
murder 44.5%; rape 59.6%)

sample survey of state 

prison inmates

Erlanson (1946)

rape

Chicago

1938–1946

87% within home 

neighbourhood

home neighbourhood = 

police precinct

Farrington & Lambert 

(1993)

burglary & violent 

offences

Nottinghamshire

1991

69.2%/55.3% < 1 mi; 

80.7%/67.8% < 2 mi 
(burglars/violent offenders)

younger & smaller 

offenders lived closer

Gabor & Gottheil 

(1984)

total of 10 crimes

Ottawa

1981

1.22 mi (70.5% in–towners)

out–of–towners, NFAs, & 

n/k excluded

Gabor & Gottheil 

(1984)

homicide

Ottawa

1981

0.54 mi (71% in–towners)

out–of–towners, NFAs, & 

n/k excluded

Gabor & Gottheil 

(1984)

rape & indecent 

assault

Ottawa

1981

1.43 mi (90% in–towners)

out–of–towners, NFAs, & 

n/k excluded

© 2000 by CRC Press LLC