|
Understanding patterns of criminal behaviour is important for both the
prevention and detection of crime. In this paper we will focus on the
crime of burglary and concentrate, in particular, on work concerned with
spatial and temporal patterns of this crime. Research in this area has
revealed a number of findings that have direct relevance to both policy
making and operational policing. For instance, studies demonstrate that
rather than being random, crime tends to cluster in space (e.g. Bailey
and Gatrell 1995), and that directing police (Sherman and Weisburd 1995)
or crime prevention (e.g. Kodz and Pease 2003) resources to such 'hotspots'
can have a crime reductive effect. However, the areas delineated as hotspots
of crime are typically large and even in areas where the risk of victimisation
is high, not all households are victimised (e.g. Budd 1999). Thus, to
optimise the efficiency of crime reduction effort, a more precise understanding
of the dimensions of risk is needed. Research demonstrates that prior
victimisation is a very good predictor of future risk (e.g. Anderson et
al 1995; Polvi et al. 1990; Farrell and Pease 1993) and that when it occurs,
repeat victimisation tends to occur swiftly (e.g. Polvi et al. 1991; Anderson
et al. 1995; Johnson et al. 1997). The implications of such findings for
crime prevention are difficult to overemphasize; the burglary event should
trigger preventive action focused on the burgled home.
Interventions that have focused on reducing repeat victimisation have
realized impressive reductions in crime. For instance, in the original
demonstration project, which included a target hardening initiative directed
towards victims of crime, relative to a reference area, burglary within
the target area was reduced by up to 70 percent(Forrester et al 1988).
Numerous replication studies have subsequently been published (e.g. Johnson
et al. 2001), thereby illustrating the external validity of this approach
in areas where repeat victimisation is a problem.
Two rival (or complementary) hypotheses have been articulated to account
for repeat victimisation. The first, referred to as the flag account (see
Pease 1998) suggests that certain properties effectively advertise their
vulnerability, which attracts any passing opportunistic offender. In this
way, burglaries that occur at the same location are considered to be independent
events, with the only thing in common being the property targeted. According
to the second hypothesis, on the other hand, a subsequent event committed
at a property is considered conditional upon the first. Thus, following
an initial crime, the risk of victimisation is 'boosted' (Pease 1998).
Here, the assumption is that the same offender, or group of offenders,
will be involved in the crime series and that experience gained during
the first event is put to use later.
Determining which of these explanations is correct, or the extent to which
each contributes towards an explanation of the phenomenon, has important
implications for crime reduction theory and practice. For instance, if
a series of crimes were usually committed by the same offender, this would
aid the investigative process and increase the probability of detections.
Moreover, if a particular offender favored a specific Modus Operandi (MO)
which he or she applied consistently, this too would inform strategies
aimed at preventing further events.
Patterns of crime provide a partial means of determining the relative
importance of boost and flag explanations. For instance, according to
the flag account, we would anticipate that repeat crimes would be equally
likely to occur over a variety of intervals of time. Thus, the finding
that incidents of repeat victimisation tend to occur swiftly (e.g. Polvi
et al. 1991), which reflects an event-related signature, is difficult
to reconcile with the flag account (unless the rate of victimisation of
the home is so high that events occur every few weeks, which is thankfully
exceptional).
Specific a priori predictions also can be made regarding the way
in which repeat incidents are committed. Within the forensic psychology
literature it is assumed that there will be a fairly high degree of similarity
in the way in which the same offender commits a series of crimes (e.g.
Bennell and Canter 2002; Ewart, Oatley and Burn 2004). For instance, a
particular offender may favor a specific tool or entry point to gain access.
Further evidence for this theory comes from recent work by Adderley and
Musgrove (2003). They used features of Modus Operandi (including spatial
and temporal preferences) to identify crimes that could potentially have
been carried out by a particular network of offenders who worked together.
Using the MO preferences of the network, they were able to increase the
accuracy of a list of the network's possible offences from 10-15 percent
to 55 percent. Taken together, these findings suggest that MO can be a
useful way of distinguishing between the strategies of different offenders/networks
of offenders.
If repeat crimes are committed by the same offender(s), we would expect
a series of repeat events to have similar MOs. Ratcliffe and McCullagh
(2001) examined the similarity of the MOs used in repeat burglaries, operationalized
as the point of entry (POE, e.g. back door versus ground floor window)
and means of entry (MOE, e.g. kicked in versus window smashed) used. The
results indicated a greater degree of consistency for repeat crimes committed
closest together in time.
Interviews with offenders provide still further support for the boost
account (Pease 1998). Consideration of the reasons typically given by
offenders for returning to the same properties suggests that these are
bounded by rational choices that are entirely commensurate with the boost
account. These include familiarity with the house layout, the risks involved,
and the known availability of saleable goods (Ericsson, 1995). Thus, the
overwhelming evidence from the research undertaken is that the same perpetrators
are responsible for the bulk of offences against the same target (Farrell
and Pease 2001).
More recent research suggests that repeat victimisation may represent
a special case of a more general space-time pattern of victimisation,
with implications for effective crime prevention and crime pattern theory.
Specifically, research conducted in the UK (Bowers and Johnson 2005; Johnson
and Bowers 2004) and Australia (Townsley et al. 2003) suggests that the
risk of victimisation is communicable, with the risk of victimisation
following an initial burglary not only affecting the burgled home but,
in a similar way to the spread of a communicable disease, also extending
to properties nearby. As with repeat victimisation conceived narrowly,
the (communicated) risk of burglary to nearby properties (within 400m
of each other) was shown to be elevated for a short period of time, typically
one-month, after which risks returned to pre-event levels. This pattern
of space-time clustering has been referred to as the 'near repeat' phenomenon
to reflect the association with repeat victimisation.
Across all of the published studies (Bowers and Johnson 2005; Johnson
and Bowers 2004; Townsley et al. 2003), the starting point for the research
was essentially a variant of the predictions generated by the boost account.
The rationale for which was that having burgled one property, offenders
would become more familiar with and, consequently, target nearby households.
Good reasons for this hypothesis exist. For instance, houses nearest to
each other are likely to share more features that may inform offender
targeting decisions than those located further away. Such features include
access and escape routes, internal and external architectural layouts,
levels of natural surveillance, and the availability of desirable goods.
As already noted, the results of the studies validated the hypothesis,
demonstrating that burglary clusters in space and time. However, one question
that currently remains unanswered concerns the similarity in the way in
which near repeat burglaries are committed. If similar, this would suggest
a common offender or group of offenders. If substantially different, an
alternative conclusion may be warranted.
The aim of the current study was to examine this issue using police recorded
crime data for the County of Merseyside, UK. As with the study conducted
by Ratcliffe and McCullagh (2001), two forms of MO were examined, these
being point and method of entry. The specific hypothesis was that if near
repeats are committed by the same offenders we would expect a greater
degree of similarity in the way in which pairs of such incidents were
committed than would be expected on the basis of chance. Furthermore,
we also would expect there to be a greater degree of similarity for events
that occurred close to each other in both space and time than for crimes
which occurred close to each other on one dimension alone (close in space
but not time or vice versa). The importance of the latter point warrants
further emphasis as one criticism that could be directed at a set of confirmatory
findings is that we might expect events that occurred close together to
have similar MOs even if they were committed by unrelated offenders. The
reason for this is that it is plausible that opportunities, in terms of
point of entry and to some extent method of entry, may cluster in space.
To illustrate this point, consider that as already discussed, houses on
the same street will most probably share similar architectural features
which in turn would provide similar potential opportunities for access
to the houses for opportunistic offenders. However, it is important to
note that we are concerned here with the similarity in the MOs for events
that occur close together in both space and time. That is, even
if specific opportunities that encourage offenders to use specific MOs
cluster in space, there is no reason to anticipate that they would systematically
cluster in time also.
To summarise, research demonstrates that burglary clusters in space and
time, placing the burgled home and those nearby at an elevated risk of
victimisation in the near future. In relation to repeat victimisation
proper, interviews with offenders, analyses of the temporal signature
of repeat victimisation, and the similarity of MOs across successive events
at the same property provide support for the boost explanation of this
phenomenon. Thus, the aim of the current study is to examine whether analyses
of police recorded crime data provide evidence to support the same conclusion
concerning near repeats.
DATA
Police recorded crime data concerned with 3,562 domestic burglaries that
occurred in a study area of Merseyside between April 1997 and March 1998
were analysed. The study area and time period selected were chosen to
allow direct comparisons with our earlier work. For the purposes of illustration
the study area is shown as Figure 1. For each burglary event, the data
included the following fields of information:
- a unique reference number
- the address of the offence, stored as a free text field
- the grid reference of the offence location (x and y coordinates)
- the date of the offence
- the point of entry used to gain access to the property
- the method of entry used
Figure 1. Map of Merseyside and the Grid Used.
Measuring the Similarity of the Modus Operandi Used
To address the research hypothesis, in the analyses that follow, all burglaries
are compared with all others in discrete pairs. With 3,562 crime events,
this means that 6,342,141 comparisons were made.
For each burglary pair, the distance and time between the
two events was calculated. In addition, the MOs for each event were also
compared. The more MOs that are considered and match, the more confident
we can be that there is a connection between the events. This is because
the discriminatory power of the test increases with the number of cases
considered.
On Merseyside, as in many other forces, there are several different MO
fields that help to describe the burglary event. For example, MOs cover
point of entry, means of entry, relationship with offender, time of offence,
description of victim, description of location, and description of weapon.
For each of these fields standard codes, which have an associated look
up table, are used to describe how the crime was committed. When inputting
data, the officer is requested to pick the most relevant fields to complete.
Consequently, the frequency with which each field is completed typically
varies according to the type of crime. For instance, the most frequently
completed fields for burglary are point of entry, means of entry, and
location of the event.
In addition to considering the reliability of the MO data, it is important
to consider the relevance of the different variables in relation to the
experimental hypothesis. The aim of the current research was to see how
consistently different incidents of burglary were carried out within an
area. Thus, as we were concerned with property crime, some MO data such
as the victim's relationship with the offender, the description of the
victim, and the description of any weapon used were considered less relevant.2
Hence, as a result of data availability, reliability, and relevance, the
analyses that follow are restricted to consideration of two elements of
the MOs used; point of entry (available for 84% of incidents) and means
of entry (available for 88 % of incidents).
In line with earlier research (Johnson and Bowers, 2004), and in response
to the findings concerned with the communicability of risk, as a starting
point 'near repeats' were defined as burglaries occurring within 400m
and one-month. Using these criteria, analyses were conducted to contrast
the similarity of the MOs used in near repeat burglaries with those used
in what, for the purposes of this research, would be considered random
burglary pairs. To maintain a focus on near repeats, the analyses that
follow exclude comparisons of repeat offences at the same location.
Method of Entry Used (MOE)
Figure 2 shows the proportion of burglaries for which the means of entry
(MOE) was the same for pairs of burglary that were within 400 metres of
each other and for those that were greater than 400 metres apart. The
x axis shows the number of months elapsing between events. Figure 2 shows
that for pairs of burglary events that occurred within 400m of each other,
the MOE used across the two offences were the same for 17 to 20 percent
of burglary pairings. Despite a subtle trend for burglaries that occurred
within one-month of each other (20%) to be more similar to each other,
no clear temporal signature is apparent. For burglaries that occurred
further apart from each other, the percentage of burglary pairs for which
the MOs were the same was lower across all chronometric intervals. The
difference achieved statistical significance (z=3.06, p<.005, two-tailed).
To interpret the above analyses it is important to compare the observed
results with what we would expect on the basis of chance. For the current
analyses, the probability that two unrelated burglaries would share the
same MOE can be calculated by considering the frequency with which the
different MOEs were actually used. Across all offences, 42 different methods
were used. The prevalence of different types of MOE varied (e.g. bodily
pressure was used in 26% of cases, whereas the offender sneaked into the
house in around 8% of cases), and hence it is more likely that the MOE
for two unrelated events would be the same for some means of entry than
for others. Using this information and applying basic statistical theory
it is possible to calculate the probability (P(matched MOE)) that the
MOE for two different events would match on the basis of chance. The formula
is described as follows:
P(matched MOE)= P(MOE1=1 and MOE2=1) + P(MOE1=2 and MOE2=2) + ….. +
P(MOE1=42 and MOE2=42)
Hence,
P(matched MOE)= P(MOE1=1)x P(MOE2=1) + P(MOE1=2) x P(MOE2=2) + … +
P(MOE1=42) x P(MOE2=42)
For the current data, P(matched MOE)=0.165
Figure 2. Percentage of Burglary Pairs for which the
Means of Entry was the Same.
Considering the results shown in Figure 2, for burglary
events that occurred within 400 metres of each other, the proportion (Mean
= 19.1%) of burglary pairs for which the means of entry was the same for
both events was slightly above the chance level. A one-sample t-test confirmed
that the difference was statistically significant (t(11)=12.32, p<.001,
two-tailed). For burglary pairs that were more than 400 metres apart,
the equivalent proportions (Mean = 16.6%) were almost identical to the
chance level (t(11)=0.21, p=0.84, two-tailed).
Point of Entry Used (POE)
The proportion of cases for which the same POE was used in pairs of events
is shown in Figure 3. The number of cases for which the POE were the same
again ranged between 15 to 20 percent, and this was consistently higher
for burglaries that were nearer to each other than those further away.
This difference achieved statistical significance (z=3.06, p<.005,
two-tailed). Once more, there was no evidence of a distinct temporal signature
in the pattern of results - the rate of matches appears to be fairly consistently
independent of the time that elapsed between events. However, in line
with the findings for MOE, there was a slightly larger percentage (21.3%)
of near repeats (burglaries that occurred within 400m and one-month of
each other) that had the same POE.
In the same way as described above, it is possible to derive an estimate
of the probability that two burglary events would share the same MO on
the basis of chance. For the POE used, 20 different methods were identified
in the current data set. The frequency with which different points of
entry varied (e.g. the front door was used in 27% of incidents, the rear
window on the upper floor in 5% of cases), and hence it is more likely
that the POE for two unrelated events would be the same for some entry
points, such as the front door, than others.
Consequently, for the current data:
P(matched POE)=0.174
Thus, for burglaries that occurred within 400m of each other, the probability
that the same MO in this case the
Figure 3. Percentage of Burglary Pairs for which the
Point of Entry was the Same.
POE used (Mean = 20.0%), was the same for two events was, on average,
above the chance level of 17.4 percent. The difference was statistically
significant (t(11)=15.85, p<.005). In contrast, the probability that
burglaries that occurred over 400 metres from each other (Mean =17.3%)
would have the same MO did not differ from what would be expected on the
basis of chance alone (t(11)=0.73, p=0.48).
Point and Method of Entry
The analyses reported so far have considered one aspect of MO in isolation.
In this section we consider the frequency with which pairs of burglary
events share both of the MO attributes analysed. Figure 4 shows
that the percentage of cases for which both the means of entry and the
point of entry matched was fairly low- varying between 5 and 8 percent.
However, to interpret this finding it is important to consider what we
would expect on the basis of chance. From our earlier calculations we
know that:
P(matched MOE) =0.165
P(matched POE)= 0.174
However, it is unlikely that the configuration of any two MOs will be
independent. For instance, it is likely that some MOEs and POEs will go
hand-in-hand (e.g. bodily pressure and a back door). For this reason,
using the burglary data, an analyses was undertaken to determine the frequency
with which each MOE and POE combination occurred. By doing this, it was
possible to calculate the probability that for any two burglaries both
the MOE and the POE would match by chance alone, accounting for variations
in the co-occurrence of certain MOEs and POEs. The results indicated that
the probability that any two burglaries would have the same two MOs on
the basis of chance was 0.0497.
Therefore, for near repeats the frequency with which (8% of comparisons)
the MOs of both events were the same was almost twice the figure we would
expect on the basis of chance. Considering pairs of burglaries that occurred
within 400m of each other (independent of when), a one-sample t-test confirmed
that the average (Mean = 6.6%) percentage for which both MOs were the
same was above the chance level (t(11)=13.12, p<.0001, two-tailed).
For events that occurred greater than 400 metres apart, the percentage
of pairs for which both MOs matched was not significantly different to
chance (t(11)=-0.117, p=0.91, two-tailed).
Consistent with the above findings, Figure 4 illustrates that pairs
of burglaries that occurred closer together were more likely to have been
committed in a similar way than those further apart. Again, the difference
achieved statistical significance (z=3.06, p<.005, two-tailed). In
contrast to the results above, however, for events that occurred within
400m of each other, the results also indicate a fairly strong space-time
interaction. Thus, burglary pairs that occurred within 400 metres of each
other were more likely to have the same MOs (POE and MOE) if they occurred
less than one month apart. No such effect is discernible for burglary
pairs that occurred further than 400 metres apart.
Figure 4. Percentage of Burglary Pairs for which both
MOs were the Same (Months Apart).
The temporal unit of analyses used here (months) may have masked the
full extent of the effect observed. For this reason, we conducted further
analyses that considered a smaller period of time: the number of days
between burglary events. The results are shown in Figure 5.
Figure 5 shows that for offences that occurred within 400 metres and one
day of each other, the percentage of events for which both MOs were the
same (16%) was over three times greater than we would expect on the basis
of chance. Considering events that occurred within 400m of each other
more generally, there was a much greater tendency for crimes committed
within one week of each other (M=9.12%, range 7.01% to 15.80%) to share
the same MOs (POE and MOE) than for those that occurred one month or more
apart (M=6.48%, range 6.03%-6.63%). Furthermore, the rate with which crimes
that occurred within 400m and one week of each other shared the same two
MOs was almost twice the equivalent figure for burglaries that occurred
over 400m apart (M=5.05, range 4.70-5.66).
Figure 5. Percentage of Burglary Pairs for which both
MOs were the Same (Days Apart).
To explore this pattern of results further, Figure 6 shows the same analyses
for events that occurred within, or further away than, 200m (rather than
400m) of each other. The apparent trend is similar to that shown in Figure
5, but the pattern is clearly amplified. Hence, for burglaries occurring
within one day and 200 metres of each other, both the MOE and POE matched
in approximately 23 percent of cases (almost five times greater than chance).
Again, the time-decay signature was only evident for burglaries that occurred
nearest to each other (within 200m).
Are near repeats more likely to have the same MOs?
The above analyses suggest that, relative to burglaries that occur some
distance apart (>400m), there is a significant difference in the consistency
with which the same MOs are used for events that occur closer together
in space. There also is evidence to suggest that there is a space-time
interaction, such that events that occur close together in space and
time, particularly within 200m and a few days, are more likely to share
the same MOs for both means and point of entry. In relation to the time-space
interaction, the above figures have shown visual trends in the data and
have given some indication of the probabilities involved in the patterns
found. However, it is important to test whether these trends are statistically
significant or not.
Figure 6. Percentage of Burglary Pairs for which both
MOs were the Same (Same Day).
To do this, we used a series of logistic regression models. Logistic
regression models are used to estimate the likelihood that a particular
set of results would be observed in a trial, given certain values of one
or a series of independent variables. Expressed in a slightly different
way, given a certain configuration of one set of variables, what is the
probability that we will observe a specific pattern in another? Here the
question of interest was whether or not burglaries that occur close together
in space and time (near repeats) were more likely to share the same MO
than those that occurred further apart (in space and time).
Table 1 shows the result of three different logistic regression models.
In each case, the dependent is the likelihood that a pair of crimes would
share the same MO. For the first model, 'matches' are defined on means
of entry only, in the second on point of entry only, and for the third
on both means of entry and point of entry. The independent variables in
each case are the distance between crimes, the time between crimes, and
the distance-time interaction variable expressed as odds ratios of the
independent with the dependent (e.g. 
in the equation below). The general form of the logistic regression equation
is therefore:
Where the odds ratio is equal to 1, this indicates that there is no relationship
between the likelihood that two burglaries will share the same MO(s) and
the independent variable of interest. Positive odds ratios of over 1 indicate
that there is a greater likelihood that the MO(s) will be the same for
two crimes for increasing values of the independent variable. In relation
to the coding of the independent variables for the distance between crimes,
a value of 1 indicates that two burglaries occurred within 400 metres
of each other (zero that they occurred more than 400m apart). For the
time between events, a value of 1 indicates that the events occurred less
than one month apart (zero more than one month). The associated p-values
shown in Table 1 indicate whether or not the odds ratios of the independent
variables are statistically significant.
Table 1. Logistic Regression Models for Matching MOs.
The first model, shown in the first row of the table, indicates that
burglaries which occur close together in space (but not necessarily time)
are more likely to share the same MOE. Equally, it is clear that the MOE
is more likely to be consistent for burglaries that occur within one month
of each other. The space-time interaction term also was significant, indicating
that burglary pairs defined as 'near repeats' (in other words those within
400 metres and occurring less than a month apart) are more likely to share
the same MOE than those that are not near repeats. Importantly, this effect
is over and above the main effects of space (where victimised houses physically
close to each other are more likely to have matching means of entry) and
time (where houses victimised within a certain time of each other are
more likely to have matching means of entry). Thus, whilst burglaries
that occur near to each other in space are more likely to share the same
MOE, they are even more likely to share the same MOE if they also
occur close together in time, although it should be noted that the low
value of the odds ratios indicates that the patterns are fairly subtle,
albeit statistically significant.
The second model shows the same trends for matching POE. It is interesting
to note that for this MO there is no main effect of time. Thus, burglaries
that occur within short intervals of each other are no more likely to
share the same POE than events for which the elapsed time is longer. This
contrasts with the findings for MOE and may suggest that there are temporary
'fashions' in the MOE employed by offenders, but that the generally preferred
POE endures. Importantly, however, there is a significant space-time interaction
indicating that burglary events that occur close in space and time
are more likely to share the same POE.
The third model shows the results for the likelihood that both MOs will
be the same for any two burglaries. Here, in all cases the odds ratios
are larger than in the other two models. This means that there was a stronger
association between when and where crimes were committed and how they
were committed when both MOs were considered simultaneously. In particular,
we see that near repeats are 1.17 times more likely to have both MOs matching
than other burglary pairs. This can be compared to likelihood ratios of
1.07 for matching means of entry and 1.08 for matching points of entry.
Thus, in line with the patterns revealed in graphs 4-6, the results indicate
that crimes that occur close in both space and time are more likely to
share the same MOs than those that do not.
DISCUSSION
Earlier work discussed in the introduction demonstrates clearly that the
risk of victimisation is communicable. One of the central assumptions
made in relation to this finding is that the same offenders, or their
associates, are responsible for crimes that form part of a space-time
cluster series (near repeats). If this conclusion is valid, then we would
expect to see certain patterns in the way that crimes are committed. Given
the evident behavioural consistency in the way in which offenders commit
crimes, if near repeats are committed by the same offenders we would expect
them to be committed in a more similar way to their antecedents than other
crimes. In the current paper, we tested this hypothesis by comparing the
consistency with which crimes committed close together in time and space
(near repeats) were conducted with other pairs of burglaries. A number
of findings emerged.
For the means of entry used to gain access to properties, the rate with
which the same MO was used for crimes that were committed over 400m away
from each other did not differ from what would be expected on the basis
of chance. In contrast, for crimes committed within 400m of each other,
the rate was significantly greater than would be expected. For crimes
committed nearby to each other, there was subtle evidence of more consistency
for burglaries that were also committed shortly after each other.
Similar results were observed for the point of entry used.
A much clearer pattern of results was apparent for the analyses that considered
the rate with which both MOs were consistent between events. Again,
for crimes that occurred over 400m from each other, the rate with which
the same MO (POE and MOE) was used did not differ from what would be expected
on the basis of chance. However, for crimes committed within 400m of each
other, the rate was much higher. There was also clear evidence of a space-time
interaction. Crimes that occurred within 400m and one month of each other
were almost twice as likely to share the same MO as those that occurred
within the same period of time but further away from each other. And,
importantly, crimes that occurred within 400m of each other were more
likely to share the same MO where less time elapsed between events. This
pattern was particularly evident in the analyses for which smaller units
of time were considered. Thus, crimes that occurred within one day and
400m of each other were over three times as likely to share the same MO
as those that occurred over 400m from each other.
Thus, the analyses suggest that crimes committed near to each other both
in space and time are more likely to be conducted in the same way
than other crimes. Critically, this pattern was observed after the separate
influences of the distance between events and the time between events
had been taken into account, thereby demonstrating that it was not simply
an extension of these more general patterns. To elaborate, consider that
if events were purely independent of each other - carried out by completely
different offenders that just happened to choose the same MO - then, as
a consequence of the likely similarity in access opportunities of nearby
homes, we might anticipate a similarity in the way that crimes committed
near to each other were conducted. However, there would be no reason to
suspect that this similarity would vary depending on the time that elapsed
between events.3 Thus, the results represent
an event-related signature.
Therefore, the results appear consistent with the hypothesis that there
is dependency between events that occur close together in both space and
time. Consequently, the next question to answer is in what way are near
repeats dependent on each other? A number of alternatives exist, as follows:
- The same offender has committed the burglaries.
- A group of co-offenders offend in an area together.
- A more general 'offender network' has discovered that good opportunities
exist within a certain area and brief each other about the opportunities
and successful MOs.
- There is a new vulnerability in an area (e.g. a CCTV camera or security
gate is removed) that is manipulated by unrelated offenders within a
short space of time.
- Offenders are displaced or deflected from other nearby areas as a
result of a crime prevention intervention and subsequently move into
the same area simultaneously.
The last of these alternatives is the least plausible. If offenders were
displaced from one area to another, this would clearly impact on the rate
of the crime in the area. There is no reason to believe that such a change
would have a systematic impact on the way in which crimes were
committed in the area, in the way considered here. The fourth explanation
is also fairly unlikely to generate the results observed. For instance,
if the conditions in an area changed in a way that had implications for
the way in which offenders commit offences, this would need to be a temporary
change or otherwise offenders would continue to use the same MO over longer
periods of time. If there were frequent changes within an area this could
cause the results observed, but such changes are unlikely. Consequently,
perhaps the most likely explanation is that an offender or group of offenders
favor a particular MO and use this where appropriate. In support of this,
Everson and Pease (2001) report that for a sample of offenders in West
Yorkshire, crimes committed on the same street were typically committed
by the same offender(s).
In this paper we have used offence data only to examine the relationship
between offences. If information on case status and detections was made
available, future research could examine the extent to which those offences
which are defined as being strongly associated by the methods reported
here are, in fact, actually carried out by a common offender or group
of offenders. This would be a good method of further testing the state
dependency hypothesis.4
The above findings and their interpretation have a number of implications
for crime prevention and detection, some of which will now be discussed.
In a recent paper (Bowers, Johnson and Pease, 2004), we have discussed
how the finding that the risk of crime is communicable can be used to
predict the location of future crimes. Using this method a geographical
map can be generated to indicate future areas of high risk. We have called
this method prospective mapping to distinguish it from retrospective techniques
such as traditional hotspot mapping. In a comparison of the predictive
efficiency of the different techniques, we have shown that prospective
mapping is significantly more accurate than extant methods, correctly
identifying the future locations of between 64 and 80 percent of burglary
events for the period considered (Bowers et al. 2004, Johnson et al. 2004).
The current findings suggest that an extension to the general method may
be warranted. As well as predicting when and where crimes will be committed,
the findings suggest that it should also be possible to anticipate how
they might be committed. The implication of this is that crime reductive
resources may not only be directed to the right places at the right times,
but the most appropriate tactical options may be selected. For instance,
in an area where a series of near repeats occur and the favored MO is
to gain access via a rear window using a brick, installing better locks
on front doors would be unlikely to have a crime reductive effect.
In relation to crime detection, monitoring the constellations of spatial
and temporal patterns of crime could assist with the detection of offenders.
If, as the data suggest, it can be assumed that offenders are likely to
visit houses near burgled homes within short spaces of time, covert surveillance
schemes could be used in areas where near repeats events are particularly
prevalent. Alternatively, silent burglar alarms monitored by the police
could be given to burglary victims and their neighbors for a short period
of time following an initial event. This strategy has been used effectively
in the past to reduce repeat burglary (Anderson et al. 1995), and hence
this suggestion represents an extension of a tried and tested intervention.
To optimize the effectiveness of such strategies, it may be sensible to
adopt these techniques in areas where there is not only evidence of repeat
victimization or near repeats, but also where the consistency in the MO
used to gain access to burgled homes exceeds a certain threshold5,
as this may suggest the increased likelihood of a common offender. Research
concerned with repeat victimisation (Everson and Pease 2001) suggests
that those who commit repeat victimisation tend to be prolific offenders
well established in their criminal careers. If the same is true for those
that commit near repeats then these types of strategies offer the opportunity
to catch offenders quickly and to apprehend those responsible for the
bulk of offences.6
Perhaps a more effective way of linking crimes would be to use physical
evidence such as DNA recovered from domestic burglary crime scenes. In
England and Wales, although scene of crime officers attend around 70 percent
of domestic burglary crime scenes, DNA is only recovered at around 5 percent
of locations, fingerprints at around 27 percent (Williams 2004). This
would clearly limit the utility of such an approach but it would enable
some analyses to be conducted, particularly in police forces where the
recovery rates are highest.
To recapitulate, the results of the current analyses show that near repeats
are committed with greater consistency than other crimes. To the authors,
this suggests a common offender across near repeat event series. Data
from interviews with offenders also suggest that this conclusion is, to
some extent, warranted (Ashton et al. 1998). In any event, the results
have clear implications for operational policing, both in relation to
crime prevention and detection. Future research may determine whether
or not the consistency with which crimes are committed co-varies with
other factors. An examination of the time of day would be particularly
useful given the implications the findings could have for the precise
timing of crime reductive effort (see, for example, Ratcliffe 2002). Equal
consideration could be also given to the area in which crimes are committed,
the type of housing in the area, and the goods stolen.
ENDNOTES
1. The authors would like to thank Ken Pease for comments on an earlier
draft of this paper. Thanks also go to Barry Webb for discussions regarding
the potential use of DNA in crime linking. back
2. However, as pointed out by one of the reviewers, information concerning
the relationship between the victim and offender may help to explain how
the offender gained knowledge of the opportunity. This would be an interesting
issue to pursue in further research. back
3. Unless, of course, there were systematic changes in the area at specific
points in time that made some MOs favourable for a period of time and
others at another. back
4. Thanks go to one of the reviewers for suggesting this. back
5. The authors would like to thank Gisela Bichler for pointing out this
possibility. Various standard statistical techniques could be used to
operationalise this approach. back
6. For instance, it has been estimated that between 5 percent and 19
percent of the population accounts for over one-half of all crime (e.g.
Wolfgang, 1983). back
REFERENCES
Adderley, Richard and Peter Musgrove. 2003. "MO modelling of group
offending: A data mining case study." The International Journal
of Police Science and Management 5(4), 267-276.
Anderson, David, Sylvia Chenery and Ken Pease. 1995. "Biting Back:
Tackling Repeat Burglary and Car Crime." Crime Detection and Prevention
Series Paper No 58. London: Home Office.
Ashton, J., Brown, I., Senior, B., and Pease, K. (1998). "Repeat
Victimisation: Offender Accounts." International Journal of Criminology,
3(4), 269-279.
Bailey, Trevor C. and Anthony C. Gatrell. 1995. Interactive Spatial
Data Analysis, Harlow: Longman.
Bennell, Craig and David Canter. 2002. "Linkning Commercial Burglaries
by Modus Operandi: Tests using Regression and ROC Analysis." Science
and Justice, 42(3), 153-164.
Bowers, Kate J. and Shane D. Johnson. 2005. "Domestic burglary repeats
and space-time clusters: the dimensions of risk." The European
Journal of Criminology, forthcoming.
Bowers, Kate, Johnson, Shane D., and Pease, K. (2004). "Prospective
Hot-Spotting: The Future of Crime Mapping?" The British Journal
of Criminology, 44(5), 641-658.
Budd, Tracey. 1999. "Burglary of Domestic Dwellings: Findings from
the British Crime Survey." Home Office Statistical Bulletin 4/99,
London: Home Office Research and Statistics Directorate.
Ericsson, Ulrica. 1995. "Straight from the Horses Mouth." Forensic
Update, 43, 23-25.
Everson, Steve and Ken Pease. 2001. "Crime Against the Same Person
and Place: Detection, Opportunity and Offender Targeting". In G.
Farrell and K. Pease (Eds.) Crime Prevention Studies Volume 12: Repeat
Victimisation. New York: Criminal Justice Press.
Ewart, Brian. W., Giles C. Oatley, and Kevin Burn. 2004. "Matching
Crime Using Burglars' Modus Operandi: A Test of Three Models." Forthcoming.
Farrell, Graham and Ken Pease. 1993. "Once Bitten, Twice Bitten:
Repeat Victimisation and its Implications for Crime Prevention."
Crime Prevention Unit Series Paper No 46. London: Home Office.
Farrell, Graham and Ken Pease. 2001. Repeat Victimisation. New
York: Criminal Justice Press.
Forrester, David, Mike Chatterton and Ken Pease. 1998. "The Kirkholt
Burglary Prevention Project, Rochdale." Police Research Group
Crime Prevention Series 13, London, Home Office.
Johnson, Shane D., Kate J Bowers and Alex F.G. Hirschfield. 1997. "New
Insights into the Spatial and Temporal Distribution of Repeat Victimisation."
British Journal of Criminology, 37(2), 224-241.
Johnson, Shane D., Kate J. Bowers, Christopher A. Young, and Alex F.G.
Hirschfield. 2001. "Uncovering the True Picture: Evaluating Crime
Reduction Initiatives using Disaggregate Crime Data." Crime Prevention
and Community Safety: An International Journal, 3(4), 7-24.
Johnson, Shane D. and Kate J. Bowers. 2004. "The burglary as a clue
to the future: the beginnings of prospective hot-spotting." The
European Journal of Criminology, 1(2), 237-255
Johnson, Shane D., Kate J. Bowers, K. and Ken Pease. 2004. "Predicting
the Future or Summarising the Past? Crime Mapping as Anticipation."
In Melissa Smith and Nick Tilley (Eds.) Launching Crime Science.
London: Willan.
Kodz, Jenny and Ken Pease. 2003. "Reducing Burglary Initiative: Early
Findings on Burglary Reduction." Home Office Research Findings
204. London: Home Office.
Pease, Ken 1998. "Repeat victimisation: Taking stock." Crime
Detection and Prevention Series, Paper 90. Home Office: London.
Polvi, Natalie, Terah Looman, Charlie Humphries, and Ken Pease. 1990.
"Repeat Break and Enter Victimisation: Time Course and Crime Prevention
Opportunity." Journal of Police Science and Administration,
17(1), 8-11.
Polvi, Natalie, Terah Looman, Charlie Humphries, and Ken Pease. 1991.
"The Time Course of Repeat Burglary Victimisation." British
Journal of Criminology, 31(4), 411-14
Ratcliffe, Jerry H. 2002. "Aoristic Signatures and the Temporal
Analysis of High Volume Crime Patterns." Journal of Quantitative
Criminology, 18(1), 23-43.
Ratcliffe, Jerry and Michael McCullagh. 2001. "Crime Repeat Victimisation
and GIS." in Hirschfield, A. and Bowers, K. (Eds) Mapping and
Analysing Crime Data: Lessons from Research and Practice. Taylor and
Francis: London.
Sherman, Lawrence W., and David Weisburd. 1995. "General Deterrent
effects of Police Patrol in Crime "Hot Spots": A Randomized
Controlled Trial." Justice Quarterly, 12(4), 625-648.
Townsley, Michael, Ross Homel, and Janet Chaseling. 2003. "Infectious
Burglaries: A Test of the Near Repeat Hypothesis." British Journal
of Criminology, 43, 615-633
Williams, R. (2004). "The Management of Crime Scene Examination
in Relation to the Investigation of Burglary and Vehicle Crime."
Home Office online report 24/04. London: Home Office.
Wolfgang, Marvin E. 1983. "Violent juvenile- a Philadelphia profile."
in Kenneth, R. Feinber (ed) Violent Crime in America, 17-24. Washington,
D.C.: National Policy Exchange.
|
