Content uploaded by Wim Bernasco
Author content
All content in this area was uploaded by Wim Bernasco on May 15, 2017
Content may be subject to copyright.
http://euc.sagepub.com
European Journal of Criminology
DOI: 10.1177/1477370808095124
2008; 5; 411 European Journal of Criminology
Wim Bernasco
Burglaries
Them Again?: Same-Offender Involvement in Repeat and Near Repeat
http://euc.sagepub.com/cgi/content/abstract/5/4/411
The online version of this article can be found at:
Published by:
http://www.sagepublications.com
On behalf of:
European Society of Criminology
can be found at:European Journal of Criminology Additional services and information for
http://euc.sagepub.com/cgi/alerts Email Alerts:
http://euc.sagepub.com/subscriptions Subscriptions:
http://www.sagepub.com/journalsReprints.navReprints:
http://www.sagepub.co.uk/journalsPermissions.navPermissions:
http://euc.sagepub.com/cgi/content/refs/5/4/411 Citations
at Universiteitsbibliotheek on November 12, 2008 http://euc.sagepub.comDownloaded from
Volume 5 (4): 411–431: 1477-3708
DOI: 10.1177/1477370808095124
Copyright © 2008 European Society of
Criminology and SAGE Publications
Los Angeles, London, New Delhi, Singapore
and Washington DC
www.sagepublications.com
Them Again?
Same-Offender Involvement in Repeat
and Near Repeat Burglaries
Wim Bernasco
Netherlands Institute for the Study of Crime and
Law Enforcement (NSCR)
ABSTRACT
Burglary victimization is associated with a temporary elevated risk of future
victimization for the same property and nearby properties. Previous research
suggests that often the initial and subsequent burglaries involve the same
offenders. This paper tests this assertion, using data on detected residential
burglaries during the period 1996–2004 in The Hague and its environs, in the
Netherlands. It demonstrates that pairs of detected burglaries occurring in close
proximity in space and time are much more likely to involve the same offenders
than pairs that are not so related. Topics for future research and implications for
the detection of burglaries are addressed.
KEY WORDS
Apprehended Offenders / Burglary / Near Repeat / Repeat Victimization.
When a house is burgled twice within a short span of time, or when nearby
premises are burgled shortly one after another, it is tempting to assume that
the events involve the same perpetrator. Intuitively it seems unlikely that the
offences could be independent, and indeed sometimes there are reasons to
presume that the same offender was involved in both.
How often is this assumption correct? Does the likelihood of same-
offender involvement depend on the time and the distance between the first
and the second burglary? This paper intends to answer these questions.
at Universiteitsbibliotheek on November 12, 2008 http://euc.sagepub.comDownloaded from
The significance of repeat victimization
Since the introduction of victimization surveys in the 1970s, it has become
widely recognized that crime is concentrated among relatively few victims
(Hindelang et al. 1978). A significant number of people become repeat vic-
tims, some of them over and over again. Repeat victimization occurs when
a person or a target becomes the victim of two or more offences of the same
type (e.g. two burglaries or two robberies) within a specified time frame (e.g.
a year). During recent decades several studies have addressed repeat victim-
ization. Many of these studies, especially those conducted in the UK, were
specifically designed to inform crime prevention strategies (Laycock 2001).
It has generally been demonstrated that past victimization is a good
predictor of future victimization, that the likelihood of repeat victimization
varies across individuals and across geographical areas, and that, when
victimization recurs, it tends to do so swiftly (for an overview, see Pease
1998). These findings have been explained by two mechanisms (Tseloni and
Pease 2003). The first is that victimization flags people and places that have
enduring attributes that attract offenders. According to this mechanism,
both the initial crime and the repeated crime reflect the elevated risk asso-
ciated with stable attributes of the target. The second mechanism is that the
initial victimization boosts the likelihood of a repeat victimization. Under
this mechanism, the initial crime alters something about the victim that
increases his or her risk of becoming a crime victim again. One factor that
typically changes about a victim is that he or she is added to the offender’s
list of suitable targets. Thus the ‘boost’ explanation is compatible with the
possibility that a repeat offence against the same person or target involves
the offender who committed the initial offence.
It has recently been suggested that the elevated risk in the aftermath
of victimization may spill over to the social and spatial environment. It was
demonstrated that, in the wake of a domestic burglary, not only the prop-
erty itself but also properties near the victimized property have an elevated
burglary risk (Johnson et al. 2007a), and similar findings have been
reported with respect to shootings (Ratcliffe and Rengert 2008) and vehicle
crime (Johnson et al. 2006). Patterns of risk communication might also
operate in social networks, so that family members, friends, classmates
or colleagues of victims are ‘infected’ with a temporarily elevated risk of
victimization.
Repeat burglary victimization: Return of the same offenders?
What is known about repeat victimization is to a large extent based on
studies of repeated burglary victimization. For example, many studies have
412 European Journal of Criminology 5(4)
at Universiteitsbibliotheek on November 12, 2008 http://euc.sagepub.comDownloaded from
demonstrated that previous burglary victimization is associated with an
elevated risk of future burglary victimization (e.g. Budd 1999; Johnson
et al. 1997). Further, the risk of re-victimization is particularly elevated dur-
ing the first few months after the initial victimization and decreases as time
passes. It has been argued that this temporal pattern in particular suggests
the involvement of the same offender or offender group in both offences
(Polvi et al. 1991). In other words, in line with the ‘boost’ account of repeat
victimization, it is suggested that repeat offenders are responsible for repeat
burglary victimization.
This argument is supported by findings that demonstrate consistency
between the initial and the repeat burglary with respect to the point and
method of entry into the property and with respect to the time of day.
Ratcliffe and McCullagh (1998) examined the similarity of point and
means of entry in repeat burglaries, and found that repeats taking place
swiftly were more similar to the initial burglary than repeats that occurred
later. New research (Sagovsky and Johnson 2007) provides evidence that
repeat offences against the same property, as compared with random pairs
of burglaries, occur significantly more often at the same time of day.
Assuming that many offenders encounter burglary opportunities during
time-structured routine activities, this finding could support the hypothesis
that the same offenders are involved in both the initial and the subsequent
burglary.
These temporal phenomena, however, only indirectly support the
hypothesis that the same offenders are involved in repeat burglary victim-
izations. But is the hypothesis correct? The same phenomena could alterna-
tively signal short-term changes or cyclical patterns in the routine activities
of the burglary victims. For example, a property may lack signs of occu-
pancy when the residents are away on vacation, and these signs may attract
more than one burglar during the short holiday period only. In addition, the
timing consistency could also be caused by victim routines because the time-
structured routine activities of victims may leave the property empty and
vulnerable at certain times of the day or days of the week. Alternatively, an
initial burglary may leave the property vulnerable if broken windows or
locks are not replaced immediately, resulting in an easy entrance for other
burglars. Thus, if a property is burgled more than once during a short
period of time, the offenders might be different individuals on each occa-
sion. To support the claim that it is the same offenders who are involved
in repeat burglaries of the same object, more direct evidence is needed. In
particular, more information is needed about offenders’ repeat targeting
practices.
Such direct support comes from two sources. First, a few studies use
police data on detected cases of burglary to document this assertion
Bernasco Same-offender involvement in burglary 413
at Universiteitsbibliotheek on November 12, 2008 http://euc.sagepub.comDownloaded from
(Everson 2003; Kleemans 2001). (Note that, for the purpose of this paper,
‘detected’, ‘cleared’ or ‘solved’ burglaries indicate cases in which the police
have arrested at least one person to whom the burglary is attributed.)
Accounts by offenders themselves, typically collected through interviews
with prisoners or through ethnographic research among active offenders
‘out on the street’, are a second source of data relevant to the claim that
many burglary repeats can be attributed to the offenders who committed
the initial burglary (Ashton et al. 1998; Ericsson 1995; Hearnden and
Magill 2004; Palmer et al. 2002; Shaw and Pease 2000). As found in these
studies, the percentage of interviewed burglars who admit having gone back
to burgle a previously burgled property ranged between 31 and 76 percent,
probably reflecting differences in the wording of interview questions as well
as differences in the sample frames used for the research.
Although the results confirm the notion that returning to a previously
burgled property is a common strategy for burglars, offenders’ accounts are
not detailed enough to form the basis of a reliable estimate of the extent to
which repeat burglaries at the same address are committed by the same
offenders. For such an estimate, one would at least want to know the tim-
ing of all burglaries and to have an indication for each burglary of whether
it was a repeat burglary against the same property.
‘Near repeat’ burglaries
It has recently been shown that the elevated victimization risk after burglary
not only applies to the victimized property but generalizes to the immediate
environment of that property. In other words, burglary victimization
appears to be contagious. In the wake of a burglary, properties near the vic-
tim’s property run heightened burglary risks as well. The phenomenon was
first established in Beenleigh, a police division near Brisbane in south-east
Queensland, Australia (Townsley et al. 2003), and in Liverpool, UK
(Bowers and Johnson 2004, 2005; Bowers et al. 2004; Johnson and Bowers
2004a, 2004b), and its ubiquity has recently been demonstrated in no fewer
than 10 regions around the world (Johnson et al. 2007a).
When a nearby property is burgled shortly after a burglary, the bur-
glary is called a ‘near repeat’ (Morgan 2001). Near repeats follow a tem-
poral decay pattern similar to repeat burglaries and the explanation of near
repeat victimization more or less parallels the explanation of repeat victim-
ization. That is to say, it is asserted that the same offender(s) who commit-
ted the first burglary will often be responsible for the subsequent burglaries
committed nearby. Returning to the same vicinity would give the burglar
the advantage of knowing the area and the layout of the houses. It has fur-
ther been established that a substantial percentage of near repeats actually
414 European Journal of Criminology 5(4)
at Universiteitsbibliotheek on November 12, 2008 http://euc.sagepub.comDownloaded from
take place during the same overnight or daytime period, suggesting that in
many instances a near repeat does not represent a return to a previously
targeted area but is just the continuation of an ongoing series (Bernasco
2007). In any event, as with repeat victimization, the theoretical claim is
that near repeats involve the same offenders. Evidence that supports this
claim is scarce. In fact, most papers seem merely to take same-offender
involvement in near repeat burglaries for granted.
Bowers and Johnson (2004) provide indirect support for same-
offender involvement in near repeats. In a procedure analogous to a study
referenced above (Ratcliffe and McCullagh 1998), they compare the modus
operandi similarity of near repeat burglary pairs with the modus operandi
similarity of pairs of burglary events that are not close in time and space,
and demonstrate that ‘means of entry’ and ‘point of entry’ are significantly
more congruent for the former than for the latter. If ‘means of entry’ and
‘point of entry’ reflect offender specialization, then this finding supports the
assertion that the same offenders tend to be responsible for both initial and
subsequent burglaries in near repeats. Note that it is essential that the time
between the burglaries is included in this test, because spatial proximity
alone may simply reflect similarity between the properties involved that
could ‘invite’ a method and point of entry. For example, all houses on a par-
ticular street may have the same vulnerable entry point, i.e. a back door
accessible via a back alley that is not gated. Furthermore, generalizing the
findings of Sagovsky and Johnson (2007), it has recently been demonstrated
that, even when pairs of events taking place on the same day are excluded,
near repeats tend to occur at the same time of day (Johnson et al. 2007b).
This could be considered additional indirect evidence for same-offender
involvement in near repeat burglaries.
There has, however, been no direct support for the assertion that the
same offenders are responsible for near repeats. Interviews with burglars
have not explicitly addressed the near repeat phenomenon – which is not
surprising because it has been discovered only recently – and analyses of the
near repeat phenomenon using police data have thus far been based on
reported incidents of burglary exclusively and have not used data on
offenders.
The aim of the present paper is to assess whether we can find direct
evidence of the involvement of the same offenders in near repeat victimiza-
tion. More specifically, the aim is to answer the following questions:
•What percentage of repeat and near repeat burglaries involves the same offender?
•What is the association between spatial and temporal proximity and same-
offender involvement? Does same-offender involvement decrease with larger tem-
poral and spatial distances?
Bernasco Same-offender involvement in burglary 415
at Universiteitsbibliotheek on November 12, 2008 http://euc.sagepub.comDownloaded from
Data and analytical strategy
The structure of the data set used here is simple. It consists of 3624 detected
cases of attempted or completed residential burglary (breaking-and-entering)
recorded by the police during the years 1996–2004 in The Hague and its
environs. This is a metropolitan area with a population of about one mil-
lion comprising the city of The Hague and other cities and towns in the
vicinity. The full data set, including the undetected as well as the detected
burglaries, has been used before to establish patterns of near repeat victim-
ization in the cities of The Hague and Zoetermeer in 2002–3 (Johnson et al.
2007a) and in the cities of The Hague, Delft and Zoetermeer during the
same period as is used here, 1996–2004 (Bernasco 2007).
The record of each of the 3624 detected burglaries consists of:
•a number that identifies the burglary object (residential property) uniquely;
•an XY coordinate pair indicating the location of the object of the burglary;
•the date of the offence;
•the identities (unique registration numbers) of all offenders known to be involved
in the burglary.
Residential burglary is not an offence explicitly distinguished in Dutch
penal law. The police classify an incident as a burglary in the case of an
attempted or completed theft involving illegal entry of a domestic property,
provided it does not involve violence against persons. Accordingly, attempts
include unsuccessful attempts to enter the property illegally, as well as suc-
cessful entry without anything being stolen. In this study we include such
attempts. Illegal entries and theft from gardens, sheds, etc. are, however,
excluded.
A well-documented disadvantage of police data is that they include
only offences reported to the police. From the Police Monitor, a large-scale
victimization survey in the Netherlands, it is known that victims report
about 90 percent of completed burglaries and about 50 percent of
attempted burglaries to the police. Clearly, the use of police data to link two
offences has an additional disadvantage because both offences must have
been reported in order to establish a link. Because we are interested in
offenders, and because in the survey victims are not asked about offenders,
there is no other choice here.
Although differences in definitions and criminal justice procedures
lead to varying detection rates from country to country, the detection rate
for property crimes, including residential burglaries, is universally low, typ-
ically below 10 percent (Smit et al. 2004). Thus, even among the burglaries
that are reported to and recorded by the police, for the vast majority the
416 European Journal of Criminology 5(4)
at Universiteitsbibliotheek on November 12, 2008 http://euc.sagepub.comDownloaded from
perpetrators go undetected. Indeed, the 3624 detected burglaries analysed
here comprise only 6.21 percent of the full data set of 58,395 recorded res-
idential burglaries. It should be pointed out that a sample of detected cases
is likely to be selective with respect to the central issue of the relation
between spatio-temporal proximity and involvement of the same offenders.
Because this issue is more easily addressed after the analytical results have
been presented, it will be taken up below.
The 3624 burglaries involve 2516 offenders. The majority of burgla-
ries (71 percent) involve a single offender, another 20 percent involve two
offenders, 5 percent involve three, and 2 percent involve four. The largest
burglary group involves nine offenders. Most offenders (70 percent) are
involved in only one detected burglary; the average number of detected bur-
glaries per offender was two. These are minimum numbers because prob-
ably some offenders successfully hide the fact that they had accomplices.
There is sometimes uncertainty about the exact date of a burglary (the
victims may have been away for a few days) and often about the exact time
of the crime. For this reason, for reported events many police forces record
the earliest and latest possible date and time of occurrence. Burglaries were
excluded from the analysis if the period between the earliest and latest pos-
sible moment was longer than a week. Because we assume that it is easier
for victims to recollect when they realized they had been burgled than it is
for them to remember the last time when they did not recognize this, the
latest possible date was used as an estimate of the burglary date.
Geo-coding of victimized addresses was done on the basis of city or
town, street name, house number and possible additional specifications.
Geo-coding was unsuccessful in 3.4 percent of the burglaries because of
missing or unknown street names or because house numbers were missing.
The removal of cases with missing house numbers was not because the geo-
coding would be impossible or imprecise but because, without house num-
bers, we cannot distinguish between repeat burglaries and near repeat
burglaries in the same street.
Analysis of the involvement of the same offenders in repeat and near
repeat burglaries requires that burglary dyads or pairs are examined. Together,
the 3624 detected burglaries generate (N(N1)/2), N6,564,876 burglary
pairs.
Findings
Because the existence and significance of the near repeat phenomenon in the
area under study have been established elsewhere (Bernasco 2007; Johnson
et al. 2007a), the present analysis is not concerned with showing that near
Bernasco Same-offender involvement in burglary 417
at Universiteitsbibliotheek on November 12, 2008 http://euc.sagepub.comDownloaded from
repeats occur significantly more often than would be expected under the
alternative hypothesis of space–time independence. Instead, the present
analysis attempts to assess the level of same-offender involvement in repeats
and near repeats. As a first descriptive step, all pairs of detected burglaries
were generated and for each pair it was assessed
•whether the burglaries were offences against the same targets (repeats);
•what distance (in metres) they were apart (obviously zero for repeats, but zero dis-
tance could also occur for flats on floors above or below one another, which does
not count as a repeat but could count as a near repeat if the time between the
events was short);
•how many days separated the burglaries;
•whether or not at least one offender was involved in both burglaries, i.e. was any
one of noffenders involved in both events in a pair?
The police registration used for the analysis cannot link undetected
crimes on the basis of forensic evidence, such as fingerprints or DNA
(Jobling and Gill 2004; Leary and Pease 2003; Townsley et al. 2006). For
example, if fingerprints or DNA stains collected at one burglary scene
match those found at another, there would be evidence that the same
offender was involved in both burglaries, but this would not show up in the
data used for the present investigation. Non-forensic indicators such as
modus operandi or witness accounts are also used by the police to tenta-
tively ‘link’ cases, but they should not be viewed as direct evidence for
same-offender involvement, and accordingly they were not used in the
analysis either. Thus, in the registrations available for the present analysis,
burglaries are linked to other burglaries only through identified offenders.
Any empirical demarcation of repeat and near repeat burglary must
categorize the spatial and temporal distance dimensions to be used in the
analyses. Should repeat burglaries taking place within 2 weeks be distin-
guished from those taking place within 2–4 weeks? Should the thresholds
for defining spatial proximity start with 50 m, or 100 m, and should they
increase by a constant distance? Although there are numerous possibilities
for doing this, it was decided to use a categorization that would accommo-
date the finding that differentiation already occurs at very short spatial and
temporal ranges (Bernasco 2007) and would include repeat burglaries at the
same address as a separate category on the spatial dimension (e.g. Johnson
et al. 2007a). A matrix (shown as Table 1) was generated that shows the
percentage of pairs that involve a common offender for each space–time
combination and shows pairs of burglaries against the same property
(repeats) as a separate category on the spatial dimension. As an example of
how to interpret the table, consider the upper-left cell in the column labelled
418 European Journal of Criminology 5(4)
at Universiteitsbibliotheek on November 12, 2008 http://euc.sagepub.comDownloaded from
Bernasco Same-offender involvement in burglary 419
Table 1 Percentages of detected burglary pairs that involve the same offender, by spatial and temporal distance between the burglaries: The Hague and
environs, 1996–2004
Spatial distance
Temporal
distance Same address 1–100 m 101–200 m 201–300 m 301–400 m 401–1000 m 1001
m
0–7 days 98 89 71 63 50 27 2
8–15 days 83 57 55 35 36 18 1
16–31 days 93 48 33 20 23 12 1
32–62 days 92 23 13 9 7 4 0
63–92 days 70 14 5 4 4 2 0
93days 31 3 2 1 1 1 0
Note: N3624 burglaries generating 6,564,876 burglary pairs.
at Universiteitsbibliotheek on November 12, 2008 http://euc.sagepub.comDownloaded from
‘same address’ and the row labelled ‘0–7 days’. The number 98 means that,
of all the detected repeat burglary pairs that took place at the same address
and 7 days or fewer apart, 98 percent involved the same offender and 2 per-
cent involved different offenders. In the cell below, the number 83 means
that, of all the detected repeat burglary pairs that took place no fewer than
8 and no more than 15 days apart, 83 percent involved the same offenders,
and thus 17 percent involved different offenders. As a final example, in
the cell to the right of the previous cell, in the column ‘1–100 m’ and the
row ‘8–15 days’, the number 57 means that, of the burglary pairs that
occurred at a distance of 1–100 m from each other and 8–15 days apart, 57
percent involved the same offender (and 43 percent involved different
offenders).
With a few exceptions, Table 1 shows a very regular pattern of
decreasing percentages of same-offender involvement along the spatial and
temporal dimensions. Within the 1–100 m bandwidth, the degree of
same-offender involvement decreases from a high of 89 percent for pairs
0–7 days apart down to only 3 percent for pairs that take place more than
three months apart. Similar patterns can be observed at greater spatial dis-
tances. A notable exception to the general pattern is the initial drop (8–15 days)
and subsequent increase (16–62 days) in same-offender involvement for
repeats proper. This is consistent with the interpretation that, after an ini-
tial burglary, some burglars deliberately wait a few weeks before returning
to re-burgle the same property, in the hope that the owners will have
replaced stolen items.
Another way to summarize the level of same-offender involvement by
spatial and temporal proximity is to present the percentages of same-
offender involvement cumulatively, as in Table 2. This table shows, for
example, in the cell where the column ‘200 m’ crosses the row ‘62 days’
that, of all the detected burglary pairs taking place within 62 days and
within 200 m of each other (and including repeat victimization of the same
address within 62 days), 55 percent involved the same offender. Therefore,
in 45 percent of these burglary pairs, the first and the second burglary
involved different offenders.
Note that the column to the far right and the bottom row indicate the
absence of a spatial and temporal restriction respectively, which means that
they are limited only by the spatial and temporal boundaries of the sample:
two burglaries cannot be more than 25 km and more than 9 years apart.
For example, the top entry in the column labelled ‘25 km’ means that, of
all burglary pairs taking place no more than 1 week apart (no matter how
far apart spatially), 5 percent involve the same offender (and 95 percent
involve different offenders). The entry furthest to the left in the
‘9 years’ row means that 62 percent of the burglaries at the same address
420 European Journal of Criminology 5(4)
at Universiteitsbibliotheek on November 12, 2008 http://euc.sagepub.comDownloaded from
Bernasco Same-offender involvement in burglary 421
Table 2 Cumulative percentages of detected burglary pairs that involve the same offender, by spatial and temporal distance between the burglaries: The
Hague and environs, 1996–2004
Spatial distance
Temporal
distance Same address 100 m 200 m 300 m
400 m
1000 m
25 km
7 days 98 90 83 78 72 46 5
15 days 95 85 77 70 64 37 4
31 days 94 80 68 58 52 28 3
62 days 94 72 55 44 37 18 2
92 days 91 64 46 36 30 14 1
9 years 62 13 8 5 4 2 0
Note: N3624 burglaries generating 6,564,876 burglary pairs.
at Universiteitsbibliotheek on November 12, 2008 http://euc.sagepub.comDownloaded from
(no matter how many months apart) involve the same offender (and 38 percent
involve different offenders). As the lower-right cell of Table 2 shows, the
likelihood of a random pair of burglaries involving the same offender is
close to zero. This implies that the percentages in the other cells of Tables
1 and 2 are extraordinarily high. Same-offender involvement in detected
burglaries is thus very strongly associated with spatio-temporal proximity.
The patterns presented in Tables 1 and 2 are quite remarkable. They
indicate that the level of same-offender involvement in detected repeat and
near repeat burglaries is very high, not only in comparison with pairs that
are distant in space and time, but also in comparison with pairs that are
close in time but distant in space or close in space but distant in time. For
example, Table 2 shows that 80 percent of the burglary pairs less than
100 m and 31 days apart involve the same offender, but this is true for only
13 percent of the pairs less than 100 m apart (see the bottom cell in the col-
umn labelled ‘100 m’) and for only 3 percent of the pairs that take place
less than 1 month apart (see the rightmost cell of the row labelled
‘31 days’).
Although the patterns presented in Tables 1 and 2 unequivocally
demonstrate that same-offender involvement is very high for burglary pairs
within a month and within 200 m of each other and decreases with increas-
ing spatial and temporal distances between burglaries, a description in sim-
ple percentages does not involve comparison. What is the level of
same-offender involvement in burglary pairs that did not take place close in
time and space and how does that compare with same-offender involvement
in close pairs? In order to answer these questions, the association between
spatio-temporal distance and same-offender involvement in burglary pairs
is quantified with the help of odds ratios. An odds ratio is a measure of
association between two dichotomous variables and it is calculated as the
ratio of cross-products of the cross-tabulation of these variables. Odds
ratios can take on any non-negative value. A value of 1 indicates that near
burglary pairs are equally likely to involve the same offenders as distant
pairs. Values above 1 imply that near burglaries are more likely to involve
the same offenders and values below 1 that they are less likely to involve
the same offenders than distant burglary pairs.
As an example of how odds ratios are calculated in the analysis
reported here, in Table 3 same-offender involvement is cross-tabulated with
spatio-temporal proximity, here defined as two burglaries taking place
within 31 days and within 200 m of each other. It can be verified that the
cumulative percentage of same-offender involvement in near repeats is 68
percent (printed bold in Table 2), which is calculated as 667 out of 974 (see
first column of Table 3). On the basis of Table 3, the odds ratio is 1183,
422 European Journal of Criminology 5(4)
at Universiteitsbibliotheek on November 12, 2008 http://euc.sagepub.comDownloaded from
Bernasco Same-offender involvement in burglary 423
Table 3 Numbers of burglary pairs by same-offender involvement and spatio-temporal
proximity: The Hague and environs, 1996–2004
Close Not close
(200 m and (200 m or
31 days) 31 days) Total
Same offender 667 12,033 12,700
Not same offender 307 6,551,869 6,552,176
Total 974 6,563,902 6,564,876
Notes: N3624 burglaries generating 6,564,876 burglary pairs.
Percentage same offender involved in close pairs: 100 667 / 974 68.
Odds ratio: 667 6,551,869 / 12,033 307 1183.
which tells us that, for two burglaries taking place within 31 days and
200 m apart, the odds of their involving the same offender are 1183 times
the odds of pairs at larger temporal or spatial distances. Because the odds
ratio is a symmetric measure of association, by the same token the odds of
a pair involving the same offender being spatially and temporally nearby
are 1183 times the odds for a pair involving different offenders.
Whereas Table 3 presented the basis for calculating the odds ratio for
a single threshold combination (31 days and 200 m), Table 4 presents odds
ratios for all combinations of temporal and spatial thresholds of ‘closeness’.
The table is constructed by calculating the odds ratio for every cell sepa-
rately, as was explained for the cell ‘31 days, 200 m’ in Table 3. In Table 4,
consider the value 466 in the column labelled ‘200 m’ and the row
labelled ‘92 days’. This value means that, for a pair of burglaries taking
place within 92 days and within 200 m of each other, the odds of the two
burglaries involving the same offender are 466 times the odds
of two burglaries involving the same offender if they took place more than
200 m or more than 92 days apart. The outcomes show that the association
between same-offender involvement and the spatio-temporal proximity of
detected burglaries is extremely large. Even if near repeats are defined to
include burglary pairs up to 1 km and up to 3 months apart, the odds
that they involve the same offender are still 105 times the odds that two
burglaries that are distant in time and space involve the same offenders.
In sum, then, it seems that all findings point in the same direction,
which is that both repeat burglaries and near repeat burglaries are much
at Universiteitsbibliotheek on November 12, 2008 http://euc.sagepub.comDownloaded from
424 European Journal of Criminology 5(4)
Table 4 Odds ratios of spatio-temporal distance and same-offender involvement, by spatial and temporal distance between the burglaries: The Hague
and environs, 1996–2004
Spatial distance
Temporal distance Same address 100 m 200 m 300 m 400 m 1000 m 25 km
7 days 22779 4935 2560 1891 1396 485 34
15 days 9326 3030 1826 1279 967 349 26
31 days 8818 2095 1183 778 606 234 20
62 days 8203 1350 673 446 341 141 14
92 days 5584 943 466 314 244 105 11
9 years 848 81 46 32 26 13
Note: N3624 burglaries generating 6,564,876 burglary pairs.
at Universiteitsbibliotheek on November 12, 2008 http://euc.sagepub.comDownloaded from
Bernasco Same-offender involvement in burglary 425
more likely to involve the same offender than are spatially or temporally
unrelated burglaries. Before rushing to conclusions, however, some reflec-
tion is necessary concerning the investigative process, the universally low
burglary detection rate, and the consequences for drawing conclusions from
police-recorded burglary data.
Repeats, same-offender involvement and detection: A reflection on
the findings
When police data are analysed without considering the mechanisms by
which they are generated, the conclusions drawn may be seriously biased.
For example, many crimes are never reported to the police and reported
crimes differ from the ones that are not reported (Goudriaan 2006). The
possibility of selectivity also concerns the question of whether, among the
reported burglaries, detected cases are different from undetected cases.
More specifically, the question is whether pairs of repeat and near repeat
burglaries are more likely to be detected and whether this depends on the
same offenders being involved in both offences. If the answers to both ques-
tions are positive, our findings may overestimate the amount of same-
offender involvement. These issues are addressed in this section.
For a random pair of burglaries, it seems reasonable to assume that
the involvement of the same offender in both crimes increases the likelihood
that they will both be detected. The reason is that, when the police have
knowledge of the involvement of an offender in one crime, they will generally
take into account the possibility that this person has been involved before
in similar crimes. This expectation will direct their further investigations
with the likely result that, if the offender has indeed committed other simi-
lar offences, the other offences are more likely to be detected than if the
offender was not involved (Kleemans 2001). For example, the police may
try to obtain a confession by explaining to the offender that, if he or she
was to confess to having been involved in other burglaries, this would
barely affect the severity of the sanction (and would ensure that he or she
could not be charged with those offences in the future), a procedure similar
to the ‘taken into consideration’ (TIC) policy used in the UK.
In addition, it may also be reasonable to assume that, for burglary
pairs committed by the same offender, the likelihood that both are detected
depends on the spatio-temporal proximity of the offences. In the case of a
repeat burglary within a short period of time, it could be routinely assumed
that the police suspect that the same offender (arrested or not) is involved.
‘Them again?’ may be the first thing on the minds of police officers investi-
gating the case.
at Universiteitsbibliotheek on November 12, 2008 http://euc.sagepub.comDownloaded from
It seems likely that the same reasoning applies to pairs of burglaries
that are near repeats. Thus, if a burglar is caught in the act of committing a
burglary, and if a week ago the neighbours of the current victims were bur-
gled, the police may routinely consider the possibility that they have just
arrested the person who also committed the previous burglary. In other
words, if the police already assume that repeat and near repeat burglaries
involve the same offenders, then the likelihood that they will detect two bur-
glaries involving the same offender depends positively on the burglaries tak-
ing place close in space and time. This implies that our results must to some
extent overestimate the involvement of the same offenders in repeat and near
repeat victimization. The extent of this overestimation is addressed in the
discussion, as well as some alternative methods for assessing the amount of
same-offender involvement in repeat and near repeat victimization.
Discussion
Extant research provides considerable evidence to support the claim that a
substantial percentage of repeat burglaries involve the same offender. This
finding was generalized in this paper to near repeat burglaries. For near
repeat burglaries, the results provide evidence that same-offender involve-
ment is directly related to the spatial and temporal distance between bur-
glaries. It is the regularity of the spatial and temporal decay rather than the
absolute percentages of same-offender involvement that support the claim
that same-offender involvement underlies repeats as well as near repeats,
even though part of the relation between same-offender involvement and
spatio-temporal proximity may reflect the investigative focus of the police
detectives.
Nevertheless, addressing the potential influence of this selection effect
on the inferences drawn requires further research, perhaps ethnographic in
nature. Such research could select individuals who have been involved in at
least two burglaries in the recent past (e.g. a year). For every burglary in
which the offender was involved, it would be necessary to establish, as pre-
cisely as possible, when and where the offence took place and whether he
or she was arrested for and charged with that offence. Ideally, such data
should be verified using police data, in part because offenders who are not
caught will generally not know whether the victim reported the burglary to
the police (and unreported burglaries are not recorded and unrecorded bur-
glaries cannot be detected). Even if the validity and reliability of such data
were limited for various reasons, they would provide at least a rough esti-
mate of the true relations between detection, same-offender involvement
and spatio-temporal distance between the burglaries.
426 European Journal of Criminology 5(4)
at Universiteitsbibliotheek on November 12, 2008 http://euc.sagepub.comDownloaded from
Bernasco Same-offender involvement in burglary 427
An alternative approach to solving the analytical puzzle could be the
use of forensic evidence. Forensic evidence can help determine whether
offences involve the same offender, irrespective of whether they are detected
or not. This feature makes forensic evidence very useful for research pur-
poses. If all burglaries were reported to the police, if burglars were so polite
as always to leave traces and if fingerprints and DNA stains were collected
after each burglary, then we would have an excellent measure of same-
offender involvement that was independent of other aspects of the investi-
gation process. Because these conditions are unlikely ever to be fulfilled,
ethnographic research specifically designed to address these questions is
required.
Although the answer to the core question addressed in this paper, ‘Are
the same offenders involved in repeat and near repeat burglaries?’ seems
clear enough, the meaning of ‘are involved in’ has implicitly been translated
as ‘commit’. Thus, we have been attempting to answer the question: ‘Is it the
same offenders who commit repeat and near repeat burglaries?’ There are,
however, ways to be involved in an act without committing it. For example,
an offender who committed the first burglary may tip off an acquaintance
who commits the repeat burglary. Even if the original perpetrator is, for any
reason, unable to return, it is difficult to see why this would be a regular
burglary strategy. First of all, it seems only logical that the original burglar,
making the first risky investment in breaking and entering the property,
would like to reap the fruits by returning to the place himself or herself.
More important, many of the advantages of returning to the same place
(knowledge of where it is, how it can be entered, the internal design and
places where valuables are to be found) are not easily communicated
between individuals. A rational burglar would be advised to return in per-
son and use the prior experience. These arguments are in part supported by
the findings of Hearnden and Magill (2004), who found that only 18 out
of 80 burglars had ever relied on tips or second-hand information to return
to a property previously burgled by someone else. Alternatively, the infor-
mation may flow via a third person, for example a person involved in fenc-
ing stolen items, or even organizing burglaries without committing them.
This may not be an uncommon scenario amongst professional burglars.
The findings reported in this paper have some interesting implica-
tions for burglary investigations, i.e. for the process of identifying who the
offenders are. Before discussing these, it should be emphasized that irre-
spective of who commits burglaries, it is the concepts of repeat victimiza-
tion and near repeat victimization themselves that are central to crime
prevention because they have clear and straightforward implications: if
you want to prevent burglaries, focus on recently burgled properties and
victims as well as on nearby properties and residents, react very quickly,
at Universiteitsbibliotheek on November 12, 2008 http://euc.sagepub.comDownloaded from
and allocate resources elsewhere when the elevated risk has decreased.
A good example of a current development in this area is a recent report
(Johnson et al. 2007b) that evaluates the application of a new prediction
tool based on near repeat burglary patterns in an operational context.
In addition to efforts to prevent future burglaries, the finding in this
paper – that repeat and near repeat burglaries very often involve the same
offenders – may help police investigators to link burglary cases, i.e. attrib-
ute two burglaries to the same offender or offender group. Linking cases is
often a difficult task when the offenders are unknown, but it is also very
important because evidence from one case added to evidence from another
may help to solve both cases instead of leaving both unsolved. A recent
study underlines the importance of spatio-temporal burglary features for
linking burglaries (Goodwill and Alison 2006). Using a series of detected
burglaries, the authors evaluated criteria for linking crimes to the same sus-
pects, and concluded that spatial and temporal information was more effec-
tive in linking crimes than behavioural information (on the offender at the
crime scene) or than the characteristics of the victimized dwelling.
As noted above, ‘Them again?’ is possibly a routine working hypoth-
esis in many police departments in the case of repeat burglaries. However,
this might not be the case for near repeats, especially if the near repeat
character is not too obvious. An ‘obvious’ near repeat is when a property is
burgled one weekend and the neighbouring property is burgled the next
weekend. In this case, ‘Them again?’ is a likely response. But this might not be
true of two burglaries taking place two weeks and 200 m apart (possibly in
different streets and possibly across a division boundary), although the
findings in this paper suggest that more than half of such pairs do involve
the same offender.
In the Netherlands, where repeat victimization has not received as
much attention from practitioners as it has in the UK, Kleemans (2001)
observed that, even in the case of repeated burglaries of the same dwelling,
the police did not always consider the possibility that the same offender had
come back for another burglary. In other words, ‘Them again?’ was not
always their first reaction. It is even more unlikely that it would be if the
two burglaries had taken place in two nearby properties rather than in the
same property. Thus, an increased awareness of the link between spatio-
temporal proximity and involvement of the same offenders might help to
link cases and solve them.
At a more general level, the results underline the importance of bring-
ing the detection rate of residential burglary to a higher level (also see
Coupe and Griffiths 1996). As mentioned above, the detection rate is below
10 percent in many countries. In this respect, it should be stressed that the
offenders who commit repeat burglaries tend to be prolific offenders
428 European Journal of Criminology 5(4)
at Universiteitsbibliotheek on November 12, 2008 http://euc.sagepub.comDownloaded from
Bernasco Same-offender involvement in burglary 429
(Everson 2003; Everson and Pease 2001), so that a focus on repeat victim-
ization would be efficient not only because it targets repeat victims but also
because it targets chronic offenders. The same might be true for near
repeats if the offenders who commit repeats are the same offenders who
commit near repeats. It has been suggested that near repeats are the result
of the successful prevention of repeats (Bernasco 2007). When a burglar
returns to a previously burgled property, finds the residents have taken ade-
quate prevention measures and subsequently targets the neighbours, then a
potential repeat burglary has been displaced and turned into a near repeat
burglary. This scenario suggests that repeat offenders and near repeat
offenders are the same people, and thus that both repeats and near repeats
are mostly committed by prolific offenders. A special focus on repeat and
near repeat burglaries in police investigations may thus pay off because it
is likely to target prolific offenders.
Acknowledgement
This research was supported by a British Academy International Collaborative
Network grant and by additional funding from UCL Futures, NSCR and the Research
Incentive Fund at Temple University. I thank Shane Johnson, Michael Townsley, Kate
Bowers, Jerry Ratcliffe, George Rengert and Henk Elffers for helpful comments.
References
Ashton, J., Brown, I., Senior, B. and Pease, K. (1998). Repeat victimisation:
Offender accounts. International Journal of Risk, Security and Crime
Prevention 3, 269–79.
Bernasco, W. (2007). Is woninginbraak besmettelijk? Tijdschrift voor Criminologie
49, 137–52.
Bowers, K. J. and Johnson, S. D. (2004). Who commits near repeats? A test of the
boost explanation. Western Criminology Review 5, 12–24.
Bowers, K. J. and Johnson, S. D. (2005). Domestic burglary repeats and space-time
clusters: The dimensions of risk. European Journal of Criminology 2, 67–92.
Bowers, K. J., Johnson, S. D. and Pease, K. (2004). Prospective hot-spotting: The
future of crime mapping? British Journal of Criminology 44, 641–58.
Budd, T. (1999). Burglary of domestic dwellings: Findings from the British Crime
Survey, Home Office Statistical Bulletin 4/99. London: Home Office.
Coupe, T. and Griffiths, M. (1996). Solving residential burglary, Crime Detection
and Prevention Series, 77. London: Home Office, Police Research Group.
Ericsson, U. (1995). Straight from the horse’s mouth. Forensic Update 43, 23–5.
Everson, S. (2003). Repeat victimisation and prolific offending: Chance or choice?
International Journal of Police Science and Management 5, 180–94.
Everson, S. and Pease, K. (2001). Crime against the same person and place:
Detection opportunity and offender targeting. In G. Farrell and K. Pease (eds)
Repeat victimization, 199–220. Monsey, NY: Criminal Justice Press.
at Universiteitsbibliotheek on November 12, 2008 http://euc.sagepub.comDownloaded from
430 European Journal of Criminology 5(4)
Goodwill, A. M. and Alison, L. J. (2006). The development of a filter model for pri-
oritising suspects in burglary offences. Psychology, Crime & Law 12,
395–416.
Goudriaan, H. (2006). Reporting crime. Effects of social context on the decision of
victims to notify the police. Veenendaal, The Netherlands: Universal Press.
Hearnden, I. and Magill, C. (2004). Decision-making by house burglars: Offenders’
perspectives, Home Office Research Findings 249. London: Home Office.
Hindelang, M., Gottfredson, M. R. and Garafalo, J. (1978). Victims of personal
crime: An empirical foundation for a theory of personal victimization.
Cambridge, MA: Ballinger.
Jobling, M. A. and Gill, P. (2004). Encoded evidence: DNA in forensic analysis.
Nature Review Genetics 5, 739–51.
Johnson, S. D. and Bowers, K. J. (2004a). The burglary as clue to the future: The
beginnings of prospective hot-spotting. European Journal of Criminology 1,
237–55.
Johnson, S. D. and Bowers, K. J. (2004b). The stability of space-time clusters of bur-
glary. British Journal of Criminology 44, 55–65.
Johnson, S. D., Bowers, K. and Hirschfield, A. (1997). New insights in the spatial
and temporal distribution of repeat victimisation. British Journal of
Criminology 37, 224–41.
Johnson, S. D., Summers, L. and Pease, K. (2006). Vehicle crime: Communicating
spatial and temporal patterns. London: UCL Jill Dando Institute of Crime
Science.
Johnson, S. D., Bernasco, W., Bowers, K., Elffers, H., Ratcliffe, J., Rengert, G. and
Townsley, M. (2007a). Space-time patterns of risk: A cross national assess-
ment of residential burglary victimization. Journal of Quantitative
Criminology 23, 201–19.
Johnson, S. D., Birks, D. J., McLaughlin, L., Bowers, K. J. and Pease, K. (2007b).
Prospective crime mapping in operational context. London: Home Office.
Kleemans, E. R. (2001). Repeat burglary victimization. In G. Farrell and K. Pease
(eds) Repeat victimization, 53–68. Monsey, NY: Criminal Justice Press.
Laycock, G. (2001). Hypothesis-based research: The repeat victimization story.
Journal of Criminology and Criminal Justice 1, 59–82.
Leary, D. and Pease, K. (2003). DNA and the active criminal population. Crime
Prevention and Community Safety: An International Journal 5, 7–12.
Morgan, F. (2001). Repeat burglary in a Perth suburb: Indicator of short-term or
long-term risk? In G. Farrell and K. Pease (eds) Repeat victimization, 83–118.
Monsey, NY: Criminal Justice Press.
Palmer, E. J., Holmes, A. and Hollin, C. R. (2002). Investigating burglars’ decisions:
Factors influencing target choice, method of entry, reasons for offending,
repeat victimisation of a property and victim awareness. Security Journal 15,
7–18.
Pease, K. (1998). Repeat victimisation: Taking stock. London: Police Research
Group, Home Office.
Polvi, N., Looman, T., Humphries, C. and Pease, K. (1991). The time-course of
repeat burglary victimisation. British Journal of Criminology 31, 411–14.
Ratcliffe, J. H. and McCullagh, M. J. (1998). Identifying repeat victimization with
GIS. British Journal of Criminology 38, 651–62.
Ratcliffe, J. H. and Rengert, G. F. (2008). Near repeat patterns in Philadelphia
shootings. Security Journal 21, 58–76.
at Universiteitsbibliotheek on November 12, 2008 http://euc.sagepub.comDownloaded from
Bernasco Same-offender involvement in burglary 431
Sagovsky, A. and Johnson, S. D. (2007). When does repeat victimisation occur?
Australian and New Zealand Journal of Criminology 40, 1–26.
Shaw, M. and Pease, K. (2000). Research on repeat victimisation in Scotland: Final
report. Edinburgh: Scottish Executive Central Research Unit.
Smit, P. R., Meijer, R. F. and Groen, P.-P. F. (2004). Detection rates, an international
comparison. European Journal on Criminal Policy and Research 10, 225–53.
Townsley, M., Homel, R. and Chaseling, J. (2003). Infectious burglaries: A test of
the near repeat hypothesis. British Journal of Criminology 43, 615–33.
Townsley, M., Smith, C. and Pease, K. (2006). First impressions count: Serious
detections arising from Criminal Justice Samples. Genomics, Society and
Policy 2, 28–40.
Tseloni, A. and Pease, K. (2003). Repeat personal victimization. ‘Boosts’ or ‘flags’?
British Journal of Criminology 43, 196–212.
Wim Bernasco
Wim Bernasco is a Senior Researcher at the Netherlands Institute for the
Study of Crime and Law Enforcement (NSCR). He is currently involved in
studies of criminal target choice, crime displacement, geographic offender
profiling, and repeat and near repeat victimization.
wbernasco@nscr.nl
at Universiteitsbibliotheek on November 12, 2008 http://euc.sagepub.comDownloaded from
