Criminal Geographical Profiling

download Criminal Geographical Profiling

of 8

Transcript of Criminal Geographical Profiling

  • 8/3/2019 Criminal Geographical Profiling

    1/8

    Analysing the offence locations and residential base

    of serial arsonists in New Zealand

    MICHAEL J. EDWARDS & RANDOLPH C. GRACE

    Department of Psychology, University of Canterbury, Christchurch, New Zealand

    Abstract

    The purpose of the present paper was to test the applicability of the Canter and Larkin (1993) circle theory of environmentalrange for offending by serial arsonists in New Zealand. Police files for 45 serial arsonists convicted between 1988 and 2003

    were obtained, and maps were produced for each offender indicating arson sites as well as the home address at the time ofoffending. Criminal range circles produced according to the Canter and Larkin method encompassed all offences in 84% ofcases, consistent with prior research. Offenders were classified as marauders or commuters depending on whether the homebase was contained within the criminal range circle. Contrary to previous research, a predominant marauder pattern was notfound; the present sample was equally divided between marauders and commuters. Regressions of distance between homebase and furthest offence on the distance between the two furthest offences yielded a slope of 0.93, indicating that formarauder offenders, the home base tended to be situated eccentrically near the circumference of the criminal range circle.No demographic or offence-related variables were found that reliably differentiated between marauders and commuters.Overall, these results suggest that the criminal range circle may provide only limited information for predicting the homebase of serial arsonists in New Zealand.

    Psychological profiling is an investigative strategy

    used by law enforcement authorities that attempts togenerate a detailed description of an unknown

    offender (Bartol & Bartol, 1994; Egger, 1999). From

    a practical standpoint, one of the most important

    items to help identify an offender is their place of

    residence. For serial offenders, the locations of

    crimes can provide useful clues regarding the

    offenders residence. As a result, geographical

    profiling (i.e., analysis of the spatial distribution of

    crime sites) has become a focus of research in

    forensic psychology (Rossmo, 1997). However,

    Australasian research on geographical profiling has

    been limited, and there have been no studies to date

    conducted within New Zealand.One of the most important assumptions in

    geographical profiling is that the selection of offence

    sites is related to the residential address or home base

    of the offender. Brantingham and Brantingham

    (1981) suggested that the everyday behaviour of

    offenders (and law-abiding citizens) defined an

    activity space. Because individuals return to their

    home base or domicile after moving through their

    activity space, it is reasonable to expect that mentalrepresentations (cognitive maps) of significant

    environmental features would be determined to a

    large extent by the location of their home. Thus,

    serial offending may reveal a domocentric pattern

    with respect to the offenders residential base.

    An early pioneering study was conducted by White

    (1932, cited in Pyle, 1974), who examined 481 cases

    of crime against people and property in Indianapolis.

    White analysed distances from the offenders resi-

    dence to their offence locations. The study had two

    important findings. First, that crimes against people

    were committed extremely close to the offenders

    home (mean distance of 0.84 miles [1.35 km]); andsecond, that crimes against property were committed

    at more considerable distances from the offenders

    residence (mean distance of 1.72 miles [2.77 km]).

    White suggested that property offenders were more

    likely to move out of their neighbourhoods to

    commit their crimes, in contrast to those offending

    against persons. Whites results were consistent with

    Correspondence: R. C. Grace, Department of Psychology, University of Canterbury, Private Bag 4800, Christchurch, New Zealand.

    E-mail: [email protected]

    Australian Psychologist, November 2006; 41(3): 219 226

    ISSN 0005-0067 print/ISSN 1742-9544 online The Australian Psychological Society Ltd

    Published by Taylor & Francis

    DOI: 10.1080/00050060600637626

  • 8/3/2019 Criminal Geographical Profiling

    2/8

    a later study by Amir (1971), who found that crime

    locations of serial rapists in Philadelphia tended to

    radiate outward from a fixed point.

    Canter and Larkin (1993) proposed an influential

    model that is relevant for geographical profiling: the

    circle theory of environmental range. They distin-

    guished between the criminal range, which is a

    region that includes all offence sites of the offender,and the home range, which, similar to the

    Brantingham and Brantingham (1981) activity space,

    is an area familiar to the offender in which they

    operate in all noncriminal activities. Canter and

    Larkin assumed that both the criminal range and

    home range could be adequately described as circles.

    Although there is a variety of ways in which the

    criminal range could be specified, Canter and Larkin

    suggested a simple procedure: The criminal range

    was the circle whose diameter was a line drawn

    between the two offence sites furthest apart. They

    further proposed two possible models of the spatial

    distribution of offending: the commuter and mar-

    auder models, which differ in terms of whether or not

    the home base is located within the criminal range.

    Note that in terms of distinguishing between the

    commuter and marauder models the home range

    circle does not need to be specified, and Canter and

    Larkin did not propose a method for doing so.

    As shown in Figure 1, the commuter model

    describes the geographical distribution of crime sites

    by an offender who travels away from their home

    base into another area to commit their offences.

    Thus, there is no overlap between the location of the

    criminal range and the offenders home base. Notethat this hypothesis does not imply that the criminal

    range is unfamiliar to the offender, but rather that it

    is an appreciable distance from the area in which he

    [sic] habitually operates as a non-offender (Canter

    & Larkin, 1993, p. 65).

    The second model of the circle theory proposed by

    Canter and Larkin (1993) is the marauder model.

    According to this model, the offender moves out

    from his or her home base to commit crimes and

    then returns once the offence is completed, as shown

    in Figure 2. The offence sites tend to be distributed

    systematically around the home base, and there is

    significant overlap between the home range and

    criminal range. Thus, the marauder model implies a

    relationship between the offence locations and theoffenders home, and suggests that the geographical

    distribution of crimes may provide information about

    the home base.

    Canter and Larkin (1993) provided support for

    their theory based on a study of 45 serial rapists in

    London, who had committed an average of 5.6

    offences each. They found that for 91% of the

    offenders, all the crime sites were contained within

    the criminal range circle constructed as described

    above. This suggests that serial rapists may operate in

    a limited geographic area, in accord with the circle

    theory of environmental range. They also found that

    39 (87%) of these offenders lived within the

    proposed circle, consistent with the marauder model.

    Canter and Larkin (1993) formulated an addi-

    tional extension of the circle theory: the home range

    hypothesis. They proposed that if criminals are

    offending within a circular criminal range, and that

    their home base is within this proposed area, it might

    be possible to make some further generalisations

    about the relative location of their home base within

    the criminal range. Thus, the home-range hypothesis

    applies only to those offenders whose criminal

    behaviour is consistent with the marauder model.

    The procedure developed for the home rangehypothesis concerns the relationships between two

    variables: the distance between the two furthest

    Figure 1. The commuter model of spatial behaviour (Adapted

    from Canter & Larkin, 1993).

    Figure 2. The marauder model of spatial behaviour (Adapted from

    Canter & Larkin, 1993).

    220 M. J. Edwards & R. C. Grace

  • 8/3/2019 Criminal Geographical Profiling

    3/8

    offence sites (X) and the distance between the

    furthest offence site and the offenders home base

    (Y). If the home base is situated at the centre of the

    criminal range, then the distance from the home base

    to the furthest offence location would be half of the

    maximum distance of those two furthest offences.

    Thus, for a sample of marauder offenders, a

    regression of maximum distance between offencesites (X) on a maximum distance from the offenders

    home (Y) will have a slope between 0.50 and 1.00. A

    slope near 0.50 would imply that the home base is

    typically situated in the center of the circle, whereas a

    slope close to 1.0 would imply that the home base is

    eccentrically placed near the circumference of the

    proposed circle. Canter and Larkin (1993) found

    that the distance between the two furthest locations

    and maximum distance from the home base to an

    offence site produced a strong positive correlation,

    r0.93, p50.001. The regression equation was

    y0.84x 0.61, suggesting that the home base was

    eccentrically placed near the circumference and

    unlikely to be close to the centre. Moreover, Canter

    and Larkin noted that the average minimum distance

    travelled per offender was 1.53 miles (2.46 km),

    which is greater than the constant term in the

    regression (0.61 miles; 0.98 km). This suggests that

    offenders may have a buffer zone or a minimum

    distance that they are willing to travel to commit an

    offence. Overall, Canter and Larkins work is

    important because it suggests that serial offending

    for a presumably impulsive crime such as rape may

    be described in terms of systematic geographic

    relationships, and that the majority of offenders areadequately characterised by the marauder model.

    A limited number of follow-up studies have found

    additional support for the circle theory of environ-

    mental range with other types of offenders.

    Lundrigan and Canter (2001) analysed the body

    disposal sites selected by 126 U.S. and 29 British

    serial killers convicted since 1960. They found that

    the offenders lived within the criminal range circle in

    89% and 86% of the American and British cases,

    respectively. Kocsis and Irwin (1997) studied serial

    rapists, arsonists, and burglars in Australia (N24,

    22, and 27, respectively). Most offenders in each

    group lived in city or metropolitan areas. They foundthat the criminal range circle encompassed all

    offending for 79%, 82%, and 80% of each sample,

    respectively. The home base was located inside the

    criminal range for 71%, 82%, and 48% of cases,

    indicating that the marauder model described the

    majority of serial rapists and arsonists. Burglars,

    however, were equally likely to be marauders or

    commuters. This suggests that geographical pattern-

    ing might not be the same for all types of offenders,

    and so the utility of the circle theory may be limited.

    Because this result was based on a relatively small

    sample size, Kocsis, Cooksey, Irwin, and Allen

    (2002) examined the offences committed by a larger

    sample of burglars (N58) in four rural Australian

    towns. Similar to the Kocsis and Irwin (1997)

    results, only 50% of offence distributions were

    consistent with the marauder model. They also

    examined a number of potential demographic

    correlates (e.g., gender, ethnicity), but found thatnone was reliably associated with marauder versus

    commuter offence patterns.

    Overall, prior research suggests that the circle

    theory of environmental range and the marauder

    model in particular may apply to different types of

    serial crime, including rape, arson, and murder.

    Burglary appears to be different, perhaps because it is

    a crime for which the motivation is generally

    instrumental (Douglas, Burgess, Burgess, & Ressler,

    1992). Nevertheless, the number of studies examin-

    ing the predictions of the Canter and Larkin (1993)

    model is still relatively small and so its generality may

    be limited. Thus, the present study examined the

    geographical patterning of fires set by serial arsonists

    in New Zealand. This country likely represents a

    different profile in terms of environmental and

    geographic factors that impinge on offending beha-

    viour compared to prior research, which has studied

    offenders in primarily urban (Canter & Larkin, 1993;

    Kocsis & Irwin, 1997) or rural (Kocsis et al., 2002)

    areas. Parks and nature reserves are relatively

    common within even urban environments in New

    Zealand, offering many potential targets for arso-

    nists. Moreover, the present sample was nationwide

    and thus more heterogeneous in terms of a mixtureof urban and rural environments. Finally, in addition

    to testing the predictions of the circle theory of

    environmental range and the marauder model, we

    were also interested to determine whether there were

    any reliable correlates with marauder versus com-

    muter offending patterns.

    Method

    Data sample

    Official records of all arsonists convicted between

    January 1988 and June 2003 were obtained from theNew Zealand Police and Fire Service. We selected

    the 45 most recent serial arson offenders for

    inclusion in the study. Participants were required to

    be 14 years or older at the time of their first arson

    offence. In addition, each offender must have

    committed a minimum of three arson offences, and

    had no regular prior contact with the targets before

    committing the offence. The offenders had to reside

    at a fixed home base for at least a minimum of two

    arson offences. Eight offenders who were identified

    as having no fixed home base were excluded from the

    Serial arsonists 221

  • 8/3/2019 Criminal Geographical Profiling

    4/8

    study. After the maps had been constructed, we

    discovered that one offender had been acting as a

    hired torch for another person who had selected

    the targets. This offender was excluded from the

    study, leaving a final sample size of 44.

    All participants were required to have committed

    their multiple arson offences within the specified

    international criteria that define serial offences(Douglas et al., 1992). That is, the offender must

    have committed at least three arsons, successive

    offences had to have been committed on separate

    days (24-hr time frame), and no more than 52 weeks

    apart. The first wave of serial offences at their first

    home base was used for the study. If an offender

    committed multiple offences before and after 2 years

    in prison, then the first phase of serial offences was

    used for that offender. Serial offending was deter-

    mined by the offence dates and times as identified on

    offenders records. If the offender had committed

    spree (i.e., multiple fires, different locations, within a

    24-hr time frame) or mass (i.e., multiple fires,

    same location, within a 24-hr time frame) offences

    on a particular day, the first offence was identi-

    fied and used for the study, with the remaining

    offences excluded from the study. The majority

    of the offenders in this sample (72%) were serial-

    only offenders. The offenders address at the time

    of the arsons was recorded, as well as each offence

    site.

    Once the sample had been identified, contact was

    made with various police stations in order to access

    file information for the offenders. The police files

    supplied information such as the gender, ethnicity,age, police districts where the serial arsons had been

    committed, arson charges, sentences imposed, occu-

    pation, previous criminal convictions, marital status,

    accomplices involved, total number of arsons, other

    offences, and the type of arsons committed.

    When the home base and all offence locations had

    been identified by street address for the sample, the

    New Zealand Fire Services computerised geo-

    graphic information system was used to construct

    the maps. For each offender, home base and offence

    sites were located in terms of gridpoint coordinates.

    Once the point coordinates were encoded, straight-

    line distances from the home base to each offencewere computed and later converted into travelling

    kilometres.

    Procedure

    For each offender, a map was printed on an A3-sized

    sheet, with home base and offence site locations

    marked. In some cases (such as walkways in wildlife

    reserves) the site locations were plotted at the

    midpoint of the target pathway or street. The

    criminal range was then determined according to

    the Canter and Larkin (1993) method by measuring

    the distance in millimetres between the two furthest

    offence sites using a standard ruler. A circle was

    drawn using this distance as the diameter. Next, we

    assessed how many offence sites for each offender

    were contained within the proposed criminal range

    circle, and whether the home base was located within

    the circle. Cases where the home base or offence sitesfell on the circumference of the criminal range were

    counted as being within the circle. If the home base

    fell within the circle the offender was classified as

    a marauder; otherwise they were designated a

    commuter.

    The Canter and Larkin (1993) home range

    hypothesis was tested using the group of offenders

    classified as marauders. This required obtaining the

    distance between the two furthest offence sites (X)

    and the distance between the home base and the

    offence site furthest from the home base (Y) for each

    marauder. Distances were measured as straight lines

    in millimetres and converted to kilometres. The

    home base to furthest offence site distances were

    then regressed on the distances between the two

    furthest offence sites.

    All statistical tests used the .05 significance level.

    Results

    Of the 44 arsonists in the sample, 39 (89%) were

    male and five (11%) were female. Thirty-eight (86%)

    were Caucasian ethnicity and six (14%) were Maori.

    The average age at the time of apprehension was 25

    years (range: 14 52 years; SD9.10). The 44arsonists committed a cumulative total of 214 arson

    offences, with the mean arson series analysed

    comprising 4.86 offences (range: 3 14; SD2.91).

    Offences were committed in police districts across

    New Zealand. Twelve arsonists committed their

    offences in Wellington, 10 in Christchurch, five in

    Central (Taumarunui, New Plymouth, Palmerston

    North), three in North Shore (Waitakere), Auckland

    City, Waikato, Eastern (Hastings, Gisborne), two

    in Counties Manukau, and one each in Northern

    (Whangarei), Bay of Plenty, Tasman (Nelson,

    Blenheim), and Southern (Dunedin, Invercargill).

    One offender committed arsons in the police districtsof both Auckland City and Waikato. The district

    with the most offenders was Wellington (n12).

    First, we tested the circle theory of environmental

    range by determining the number of offenders for

    which all offence sites were contained in the criminal

    range. The criminal range encompassed all offence

    sites in 37 of 44 cases. This percentage (84%) is

    similar to that obtained in previous studies and

    provides additional evidence that the Canter and

    Larkin (1993) assumption that criminal range can be

    described in terms of a circle is valid.

    222 M. J. Edwards & R. C. Grace

  • 8/3/2019 Criminal Geographical Profiling

    5/8

    Next we determined the number of cases for which

    the home base fell within the criminal range. Visual

    inspection of the maps indicated that the offenders

    home base fell within the criminal range in 22 out of

    44 cases (50%). This percentage is smaller than

    those previously reported by Canter and Larkin

    (1993) and Kocsis and Irwin (1997), and indicates

    that offenders choice of arson locations were equallylikely to be described in terms of the marauder and

    the commuter models.

    The Canter and Larkin (1993) home range

    hypothesis was tested using results from those

    offenders who were classified as marauders. The

    first prediction of the home range hypothesis is that

    the distance between the offenders home base and

    the furthest offence (Y) has a strong positive

    correlation with the distance between the two

    furthest offences (X). Figure 3 shows a scatterplot

    of the distances between the home base and

    the furthest offence, and the distance between

    the two furthest offences. The correlation was

    r0.99, p5 .001, consistent with the home range

    hypothesis.

    The second prediction was that the regression

    equation for X and Y will have a slope between 0.50

    and 1.00. A slope near 0.50 indicates that the home

    base tends to lie near the centre of the criminal

    range, whereas slopes near 1.0 mean that the home

    base tends to be close to the circumference. The

    regression equation was Y0.35 0.93X. This

    indicates that the home range generally lies within

    the criminal range circle, but suggests that it is not

    close to the centre but tends to be eccentricallyplaced near the circumference of the circle, similar to

    the Canter and Larkin (1993) results.

    Visual inspection of Figure 3 indicates that there

    are three outliers that might have an undue influence

    on the regression parameters. Thus, we repeated the

    analysis with these points removed. The resulting

    regression equation was Y0.040.79 X. Although

    there was some decrease in the slope, it still indicates

    that the home base tended to be nearer the

    circumference than the centre of the circle.We also examined whether there was evidence for

    a safety area or buffer zone around the home base.

    Existence of a safety area would be suggested by a

    constant term in the regression that is positive but

    less than the average minimum distance of offence

    sites from the home base. The average minimum

    distance of offences from the home base for these

    offenders was 0.51 km. The constant terms of

    0.35 km and 0.04 km in the regressions reported

    above are less than the average minimum offence

    distance. This is evidence for a minimum distance

    that an arsonist is willing to travel to commit their

    offences from their home base. These results are

    consistent with the Brantingham and Brantingham

    (1981) proposed buffer zone around the marauding

    offenders residential base.

    Examination of New Zealand Police records

    established that the offenders in the study did have

    a home base at the time of their offences, and in

    every serial episode except one (an offender who

    committed arsons in two police districts) their home

    address was in the same police district as their

    offence locations. Using straight-line distances it was

    found that 82% of the offenders set fires within 5 km

    of their home address, and 68.5% of offences(pooled across offenders) occurred within 5 km of

    the home. The mean travelling distances from the

    Figure 3. Scatterplot of the relationship

    between distance of the two furthest

    offence sites and distance between the

    offenders home and furthest offence site.

    The line is the best-fitting regression

    equation, Y0.35 0.93X.

    Serial arsonists 223

  • 8/3/2019 Criminal Geographical Profiling

    6/8

    home base to the offence sites was 6.63 km, with a

    standard deviation of 10.71 km. Across the sample,

    the minimum distance travelled was 0.00 km and the

    maximum was 66.01 km.

    We investigated whether there were any demo-

    graphic or offence-related variables that might be

    correlated with the commuter versus marauder

    offending pattern. However, the two offender groupsdid not differ significantly on any of the variables. For

    marauders and commuters, respectively, the average

    age (in years) was 26.32 and 23.68, t(42)0.96, n.s.;

    the average sentence imposed (months of imprison-

    ment) was 20.23 and 20.00, t(42)0.03, n.s.; the

    average number of arsons committed in the serial

    episode was 5.36 and 4.36, t(42) 1.14, n.s.; and the

    average number of total convictions in their criminal

    history (i.e., including crimes other than arson) was

    25.95 and 43.91, t(42) 1.15, n.s. Marauders and

    commuters were equally likely to be male (91% vs.

    86%, respectively, w20.23, n.s.), to be Caucasian

    rather than Maori ethnicity (82% vs. 91%, w20.77,

    n.s.), to be single rather than married or in a de facto

    relationship (91% vs. 77%, w20.89, n.s.), to have

    committed their offences alone rather than with an

    accomplice (41% vs. 59%, w21.45, n.s.), and to

    have set fires to vegetation and bush rather than to

    man-made structures and objects (e.g., schools,

    houses, vacant buildings, rubbish bins, etc; 36% vs.

    32%, w20.10, n.s.).

    Finally, we examined whether the population of

    the community in which the offender lived, differed

    for marauder- versus commuter-type offenders. The

    home bases represented a wider range of urban andrural locations than previous studies, and it is

    possible that the failure to find a predominant

    marauder pattern might depend on this variation.

    Population data were obtained from the 2001 census

    (Statistics New Zealand, 2001). However, the

    correlation between marauder versus commuter

    and population was not significant, Spearmans

    r0.21, n.s. Thus, we could find no demographic

    or offence-related variable that was correlated with

    the marauder versus commuter pattern.

    Discussion

    The present study tested whether the circle theory of

    environmental range proposed by Canter and Larkin

    (1993) was applicable to serial arsonists in New

    Zealand. In particular, we were interested to

    determine whether the marauder model, according

    to which offenders move away from a fixed home

    base to commit their crimes, adequately charac-

    terised the majority of crime sites of arsonists, as it

    has for every type of serial offender except burglars in

    prior studies (Canter & Larkin, 1993; Kocsis &

    Irwin, 1997; Kocsis et al., 2002; Lundrigan &

    Canter, 2001). We found that circles with a diameter

    determined by a straight line between the two

    furthest offence sites contained all offences in 84%

    of cases. This is comparable to the percentages

    reported previously (91% for rapists, Canter &

    Larkin, 1993; 79%, 82%, and 70% for rapists,

    arsonists, and burglars, Kocsis & Irwin, 1997; and

    71% for burglars, Kocsis et al., 2002). Thus, ourdata provide further support for the generality of the

    circle theory and its assumption that serial offenders

    tend to commit crimes within a well-defined

    geographic area.

    Offenders were classified as marauders or com-

    muters depending on whether the home base fell

    within the criminal circle. Contrary to previous

    research (Kocsis & Irwin, 1997), offenders were

    equally likely to fit the marauder and the commuter

    patterns. This result represents the first failure to find

    a dominant marauder pattern with serial offenders

    other than burglars. It suggests that the utility of the

    Canter and Larkin (1993) circle theory and mar-

    auder model for geographical profiling may be

    limited. In our study, knowing the criminal range

    provided no additional information in terms of

    predicting the offenders home base.

    Canter and Larkin (1993) further specified a home

    range hypothesis for those offenders whose crime

    sites were consistent with the marauder model. Their

    approach involves regressing the distance between

    the home base and the furthest offence on the

    distance between the two furthest offences. The

    slope is interpreted as where the home base is

    typically situated relative to the circle, with valuesnear 0.50 indicating a home base near the centre and

    values approaching 1.0 suggesting that the home

    base is eccentrically placed near the circumference.

    Our slope was 0.93, which suggests that the home

    base was usually near the circumference. This value

    is higher than that obtained in previous studies (0.84

    for Canter & Larkin, 1993; 0.77, 0.60, and 0.65 by

    Kocsis & Irwin for rapists, arsonists, and burglars,

    respectively; and 0.85 by Kocsis et al., 2002),

    although when three potential outliers were removed

    the slope decreased to 0.79. Whether or not these

    outliers are included, the results suggest that for

    those offenders consistent with the marauder pat-tern, the criminal range circle provides relatively little

    information in terms of predicting the location of the

    home base. Given that our sample was equally

    divided between marauders and commuters to begin

    with, the predictive power of the Canter and Larkin

    (1993) model for the present data is clearly limited.

    One difference between the current study and

    prior research is that our sample was obtained across

    a wider geographical range (from Whangarei in the

    far north of New Zealand to Invercargill in the deep

    south) and there was greater variability in terms of

    224 M. J. Edwards & R. C. Grace

  • 8/3/2019 Criminal Geographical Profiling

    7/8

    offenders residing in urban versus rural areas. Thus,

    one possible reason for our failure to find a

    consistent geographical pattern in offence site loca-

    tions was the hetereogeneity of the New Zealand

    environment. From the perspective of environmental

    psychology (Stokols, 1995), one would expect that

    increased variability across a sample in terms of

    population density and features of the natural andman-made landscape would increase the variability

    of behaviour.

    Presumably there are some factors that influence

    whether a serial offenders choice of crime locations

    is consistent with a marauder or commuter pattern.

    Yet we could find no demographic or offence-related

    variable that reliably differentiated between marau-

    ders and commuters in our study. However, it is

    possible that the failure to find any significant

    correlates of geographical patterning may have been

    due to a lack of statistical power.

    Brantingham and Brantingham (1981) proposed

    the existence of a safety zone around the offenders

    home base; that is, a minimum distance that he or

    she was willing to travel before committing an

    offence. Similar to previous studies, we found that

    the average minimum distance travelled by the

    marauder-type offenders was larger than the inter-

    cept in the regression analysis to test the home-

    range hypothesis. This suggests that arsonists prefer

    to move away from their home base before setting a

    fire.

    Although we found evidence for a safety zone,

    previous research on criminal mobility has shown

    that criminals do not travel very far (e.g., just a fewkilometres) from their home base to commit their

    offences. A study of arsonists in England by Fritzon

    (2001) found that the mean distance travelled from

    home to an offence site was 2.06 km. We found that

    the average distance travelled was 6.63 km, suggest-

    ing that arsonists in New Zealand travelled relatively

    longer distances when offending. However, 68.5% of

    all the offences occurred within 5 km of the

    offenders home base. Thus, the majority of fires

    were still started near the offenders home base.

    One limitation of the study should be acknowl-

    edged. The practice in New Zealand for arson files

    held by police is that they are destroyed after 5 yearsif the damage to property or target were less then

    $20,000. However, for arsons in which the targeted

    property was valued at more than $20,000, files are

    stored in archives. Thus, small-time arsonists could

    have been underrepresented in this study, because

    the available police files would have dated back to

    only 1999. Thus our results may be applicable only

    to arsonists who commit relatively serious offences.

    The Canter and Larkin (1993) circle theory is

    based on the two furthest offence sites, which are

    used to develop the criminal range of a serial

    offender. The effectiveness of the circle theory to

    establish the offence region might be improved if

    other pertinent features are taken into account.

    Features such as topographical and geographical

    characteristics could be beneficial when constructing

    the criminal range. For example, urban structures,

    arterial roads, highways, zoning, land use and rapid

    transit stations, bus stops, physical and psychologicalboundaries, and natural barriers such as coastlines

    may all be useful when determining the offence

    region of the serial arsonist.

    Overall, our results suggest that the applicability of

    the Canter and Larkin circle theory of environmental

    range to offending by serial arsonists in New Zealand

    may be limited. Although the criminal range circle

    encompassed all offences in the large majority of

    cases, the circle provided little information in terms

    of predicting the offenders home base. Our sample

    was equally divided between marauder and commu-

    ter patterns. Of course, no single profiling tool such

    as the circle theory will prove decisive in all

    investigations. For a successful profiling strategy we

    must also use other tools such as pattern analysis of

    temporal and target selection, crime scene analysis,

    and method of operation. Despite our failure to find

    evidence that serial arsonists in New Zealand could

    be described as predominantly marauders or com-

    muters, the circle theory remains a useful framework

    for spatial analysis of offending patterns and addi-

    tional research should improve its utility for

    investigators in Australasia. For example, future

    studies could explore potential correlates of spatial

    patterning in serial arsonists, including whetherfactors such as expressive and instrumental motiva-

    tions might play a role in determining choice of

    offence locations and travelling distances to offend

    (Fritzon, 2001).

    Acknowledgements

    We thank Dave Haslett and Mark Chubb for their

    help with this study.

    ReferencesAmir, M. (1971). Patterns in forcible rape. Chicago: University of

    Chicago Press.

    Bartol, C. R., & Bartol, A. M. (1994). Psychology and law: Research

    and application (2nd ed.). Belmont, CA: Brooks.

    Brantingham, P. J., & Brantingham, P. L. (1981). Notes on the

    geometry of crime. In P. L. Brantingham, & P. L. Brantingham

    (Eds.), Environmental criminology (pp. 27 54). Beverley Hill:

    CA Sage.

    Canter, D., & Larkin, P. (1993). The environmental range of serial

    rapists. Journal of Environmental Psychology, 13, 63 69.

    Douglas, J. E., Burgess, A. W., Burgess, A. G., & Ressler, R. K.

    (1992). Crime classification manual. New York: Lexington

    Books.

    Serial arsonists 225

  • 8/3/2019 Criminal Geographical Profiling

    8/8

    Egger, S. A. (1999). Psychological profiling: Past, present,

    and future. Journal of Contemporary Criminal Justice, 15, 242

    261.

    Fritzon, K. (2001). Examination of the relationship between

    distance travelled and motivational aspects of firesetting

    behavior. Journal of Environmental Psychology, 21, 45 60.

    Kocsis, R. N., Cooksey, R. W., Irwin, H. J., & Allen, G. (2002). A

    further assessment of the circle theory in psychological

    profiling of burglary crimes. Australian and New Zealand Journal of Criminology, 35, 17 42.

    Kocsis, R. N., & Irwin, H. J. (1997). An analysis of spatial patterns

    in serial rape, arson and burglary: The utility of the circle

    theory of environment range of psychological profiling.

    Psychiatry, Psychology and Law, 4, 195 206.

    Lundrigan, S., & Canter, D. (2001). Spatial patterns of serial

    murder: An analysis of disposal site location choice. Behavioral

    Sciences and the Law, 19, 595 610.

    Pyle, G. F. (1974). The spatial dynamics of crime. (Department of

    Geography Research Monograph no. 159). Chicago: Univer-

    sity of Chicago.

    Rossmo, D. K. (1997). Geographic profiling. In J. L. Jackson, &

    D. A. Bekerian (Eds.), Offender profiling: Theory, research and

    practice (pp. 159 176). Chichester, UK: John Wiley & Sons.Statistics New Zealand (2001). 2001 Census of population and

    dwellings - final counts. Retrieved on 30 June 2005 from: http://

    www.stats.govt.nz/census/final-counts.htm

    Stokols, D. (1995). The paradox of environmental psychology.

    American Psychologist, 50, 821 837.

    226 M. J. Edwards & R. C. Grace