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

© 2000 by CRC Press LLC

This raises an important issue. The above analysis assumes that increases

over time in crime trip distance are linear, an assumption that may not be
warranted. If journey-to-crime distance grows proportionately (e.g., if a given
crime trip is, on average, 10% longer than the previous trip), then the rela-
tionship is best expressed through a power curve. Distances might also
increase in steps or after significant thresholds (as suggested by 

Figure 9.5

).

Such a growth process could result from an offender first exploring direc-
tional alternatives before increasing crime trip distance. The exact nature of
the relationship requires further research, preferably with larger data sets of
offenders responsible for lengthier series of crimes.

The FBI maxim that the first crime in a series is the one closest to the

offender’s residence was tested. Of the serial murder cases in the SFU research,
50% involved at least one crime trip distance of a mile or less (Chase, Olson,
Dahmer, Brady, Brudos, and Collins). First offence was closest in 41% of the
cases, and distance to first crime averaged 40% of mean distance (standard
deviation = 42). The FBI belief in offender residence proximity to first offence
appears to be a reasonable though not universal rule, consistent with the
increase of crime trip distance over time. It was found to apply in just under
half of the cases.

The argument that journey-to-crime distances increase over time is based

on the supposition that offenders learn from experience and increase their
spatial knowledge accordingly. Warren et al. (1995) observed that serial  rap-
ists who travel greater distances appear to have longer criminal careers; they
also found distance indicators to be significantly related to time between
successive offences.

Figure 9.5 

   Mean logarithm of crime trip distance over time.

© 2000 by CRC Press LLC

Expansion in hunting region, however, does not necessarily result in

victims being selected exclusively from the perimeter of the search area. It
does indicate that increasingly large target areas are available for the offender
to hunt within. While this implies crime trip distances lengthen over time
(albeit at a rate slower than the growth of the hunting area), it also suggests
an increase in the variation of crime journeys. To examine this possibility,
the standard deviation for crime trip distance was calculated in each serial
murder case for every crime trip after the first. When the mean standard
deviations for all cases are plotted against offence number (

Figure 9.6

), a

significant increase is observed over time (

R

2

 = 0.790). This suggests that as

a serial murderer’s career progresses, his or her crime trips may increase in
both distance and variation. 

Fully half of the cases in the study, however, showed no significant

increase in journey-to-crime distance over time. Indeed, the nearness and
least-effort principles, and the tendency for crime locations to cluster, miti-
gate against expansion. Assuming learning underlies crime trip growth, why
does an offender need to expand his or her spatial knowledge? The most
obvious reasons include greater victim search opportunities and lowered risk
of police detection. Previous crime sites become tarnished and unattractive
because of the risk of increased community vigilance, as shown by the con-
cern of child killer Westley Allan Dodd over losing his hunting ground.

60

Displacement, occasioned by police intervention or community response,
was responsible on several occasions for the shift or expansion of a serial
murderer’s hunting area (e.g., Bianchi, Sutcliffe, Dahmer, Ramirez, and
Berkowitz).

Figure 9.6

    Crime trip distance mean standard deviation over time.

© 2000 by CRC Press LLC

Increase in size of hunting area is most likely to occur in cases where an

offender has the capability and need to learn, and will benefit from doing
so. In this context, those with the capability to learn include organized mobile
offenders who possess well-developed mental maps. Offenders with a need
to learn include those whose hunting style, body disposal methods, or offence
timing result in their crimes being linked, generating community fear and a
significant police response. Offenders who benefit from learning include
those who hunt close to home or prefer victim types that are available in
several different areas (i.e., there is a uniform target backcloth).

The case of Joel Rifkin is a good example of an offender with little need

to alter his crime trip distances, which were a consistent 37 kilometres from
residence to victim encounter site. Rifkin was sane, drove an automobile, and
was familiar with large areas of New York and Long Island. His victims were
street prostitutes and their strangled bodies were dumped, often in steel
drums, at remote sites that spanned thousands of square miles. Consequently,
the murders were not linked and the police failed to realize that a serial killer
was operating in their jurisdiction. Rifkin also lived a significant distance
from the red-light district in Lower Manhattan where he picked up most of
his victims.

60 

In times of short food supply, the nomadic Naskapi Indians employed scapulimancy – a

technique in which a heat-cracked caribou shoulder is used as a map – to help them find
game on the Labrador plateau. Anthropologists have theorized that this divination process
produced effective results because of its randomizing effect, preventing grassland and tundra
areas from being overhunted (Moore, 1957).

© 2000 by CRC Press LLC

Geographic Profiling

10.1 Mapping and Crime Analysis

Crime mapping has become a common analytic practice in many police
agencies. The capability to spatially manipulate and display offence-related
data is the result of the power and availability of geographic information
system (GIS) software. “GISs are automated systems for the capture, storage,
retrieval, analysis, and display of spatial data” (Clarke, 1990, p. 11; see also
Anderson, 1992; Garson & Biggs, 1992; Goodchild, Kemp, & Poiker, 1990a,
1990b; Miller, 1993; Tomlin, 1990; Waters, 1995a; Wendelken, 1995a). The
ability to store and integrate geographic attributes and other data produces
a powerful crime analysis tool.

Approximately 30% of police agencies with more than 100 officers now

use computer-mapping software, and the International Association of Crime
Analysts (IACA) estimates the need for GIS experts in law enforcement has
grown ten-fold over the last 15 years (Waters, 1998). In a survey of 2004 U.S.
police departments, 85% of respondents stated that mapping was a valuable
tool and reported both increasing interest and implementation (Mamalian
& La Vigne, 1999). This growth has been prompted largely by greater access
to digital arrest, incident, and calls for service data. The survey found crime
clustering and hot spot analyses were the most common mapping applica-
tions, and reported that mapped crimes allow comparisons with such exter-
nal information as census, city planning, parks, property assessment, utility,
and community data.

In 1996 the National Institute of Justice established the Crime Mapping

Research Center (CMRC) to strategically support and guide this trend.
According to Dr. Nancy La Vigne, Director of the CMRC, “[maps] make
sense on an intuitive level. It’s human nature to respond to this kind of

10

© 2000 by CRC Press LLC

graphical representation. What you get is a far more sophisticated under-
standing of what’s happening on the streets” (Waters, 1998, p. 47). This allows
patrol officers, detectives, and police managers to quickly comprehend local
crime patterns and trends. Not only is it possible to integrate a variety of
different data sources in a GIS crime map, but dimensions of both space and
time can be explored through this technique.

The Spatial and Temporal Analysis of Crime (STAC) software, developed

by the Illinois Criminal Justice Information Authority (ICJIA), was one of
the first systems designed for studying the existence, location, and size of
crime hot spots (Block, 1993a; Block & Block, 1995). It has been used to
analyze homicides, drug incidents, liquor-related crimes, rapid transit
impacts, gang turfs, and community problems. The New York Police Depart-
ment (NYPD) has successfully pioneered the use of crime mapping and
organizational accountability in their CompStat (computer statistics) pro-
cess. Geo-MIND (Geographically-linked Multi-Agency Information Net-
work and Deconfliction) is an interdepartmental criminal information
network developed in Westchester and Rockland Counties, New York State,
to assist in tactical and strategic police decision making. It uses vehicle track-
ing, GIS mapping, and incident monitoring to manipulate and link infor-
mation.

The potential application of geographic information systems to the inves-

tigation of serial murder was recognized several years ago. Because a GIS can
store geographic attributes and integrate spatial and other data for analytic
purposes, it is a useful tool in the reduction of linkage blindness and the
identification of crime series. “‘Geographically coded information from
police records can be used to detect crime trends and patterns, confirm the
presence of persons within geographic areas, and identify areas for patrol
unit concentration’” (Rogers, Craig, & Anderson, 1991, p. 17, quoting from
a 1975 International Association of Chiefs of Police (IACP) report). Rogers
et al. suggest it might be possible to identify serial murder solvability factors
with a GIS through retrospective analyses of known cases. Such knowledge
may then be used to assist police investigators in their efforts to clear unsolved
murders. Much work has been done along these lines over the past decade.

Crime mapping and analysis may also help detect serial criminals. Tech-

niques employed by epidemiologists to assess the likelihood of epidemic
disease outbreaks can be useful for determining if a predator is active in a
given area. The underlying concept is the same — the significance of a pattern
of incidents (crime or disease reports) is tested through spatial-temporal
clustering statistics to ascertain if the problem is real or merely a random
fluctuation. This method was applied in 1999 to help measure the probability
of a serial murderer targeting prostitutes in Vancouver’s Downtown Eastside