Content uploaded by Ray Bull
Author content
All content in this area was uploaded by Ray Bull on Jan 11, 2015
Content may be subject to copyright.
A preliminary examination of crime
analysts’ views and experiences of
comparative case analysis
Amy Burrell†and Ray Bull
(Both) University of Leicester, University Road, Leicester LE1 7RH, UK
†(Corresponding author) Training Consultant, Perpetuity Training, 148 Upper New Walk,
Leicester LE1 7QA, UK. Email: amb58@le.ac.uk or a.burrell@perpetuitygroup.com
Submitted 17 May 2010; revision submitted 14 August 2010
Keywords: comparative case analysis, case linkage, investigative
psychology, crime analysis, behavioural consistency, crime investigation
Amy Burrell
is undertaking a PhD in forensic
psychology at the University of Leicester. Her
thesis will focus on serial robbery and compar-
ative case analysis. Amy has worked for the Jill
Dando Institute of Crime Science and Perpetuity
Research & Consultancy International. She now
works for Perpetuity Training, a company spe-
cialising in security and risk assessment training.
Amy supports the delivery of courses which
include security management (Level 4 BTEC)
and managing security surveys (Level 4 BTEC).
Ray Bull
is Professor of Forensic Psychology at
the University of Leicester. He was part of the
small team commissioned by the Home Office to
write the 2002 government document Achieving
Best Evidence in Criminal Proceedings: Guid-
ance for Vulnerable or Intimidated Witnesses,
Including Children (ABE). In 2002/03 he led the
small team commissioned by the government to
produce an extensive training pack relating to
ABE. In 2005 he received a Commendation from
the London Metropolitan Police for ‘Innovation
and professionalism whilst assisting a complex
rape investigation’.
A
BSTRACT
Comparative Case Analysis (CCA), typically
conducted by crime analysts, uses crime scene
behaviours to try to identify series of crimes
committed by the same offender. Accurate identi-
fication of series of offences allows the police to
pool resources and evidence, thereby boosting the
potential to identify and apprehend the offender.
This paper discusses the results of a survey of a
sample of crime analysts working in two UK
police forces about their views and experiences of
CCA. The results focus on how CCA is con-
ducted, what evidence and information is con-
sidered, and how useful CCA is to criminal
investigation. Suggestions for how CCA might
be developed further are also included.
INTRODUCTION
A small proportion of offenders are respons-
ible for the majority of crime (Clarke &
Eck, 2003; Tilley & Laycock, 2002) and so
it is unsurprising that government policy
frequently puts pressure on the police to
focus their investigative efforts on prolific
offenders (Woodhams & Toye, 2007).
Identifying serial crimes and linking these
offences to individual offenders is a crucial
part of police work (Bennell, Jones, &
Melnyk, 2009; Santtila, Fritzon, &
Tamelander, 2004; Sorochinski & Salfati,
2010; Yokota & Watanabe, 2002) and is used
International Journal of Police
Science and Management,
Vol. 13 No. 1, 2011, pp. 2–15.
DOI: 10.1350/ijps.2011.13.1.212
Page 2
International Journal of Police Science & Management Volume 13 Number 1
to streamline investigative efficiency (Sant-
tila, Junkkila, & Sandnabba, 2005), pool
evidence (Bennell et al., 2009) and/or nar-
row the search for the offender (eg, through
geographic profiling and/or offender profil-
ing (Canter & Larkin, 1993; Santtila et al.,
2004; Santtila, Häkkänen, & Fritzon,
2003)).
Linking offences can be relatively
straightforward if forensic evidence is found
at the crime scene (Grubin, Kelly, & Bruns-
don, 2001). However, forensic evidence is
not always present; for example, Ewart,
Oatley, and Burn (2005) report that such
evidence is often absent at burglary scenes,
and Hazelwood and Warren (2003) note
that no or insufficient DNA evidence is
found in a significant proportion of sexual
offences. Even where forensic evidence is
gathered, a lack of searchable databases of
reference samples can limit the usefulness of
this type of evidence (Cole, 2010). Indeed,
the UK National DNA Database
(NDNAD), which has the highest rate of
profiles per capita (Parliamentary Office of
Science and Technology, 2006), is limited,
as just 1 per cent of recorded crimes con-
tribute DNA profiles to the database
(House of Commons Science and Techno-
logy Committee, 2005). In addition, DNA
evidence might not be processed whilst the
investigation is ongoing and research has
found that in some cases this only occurs
where a legitimate suspect has been identi-
fied (Strom & Hickman, as cited in Beaver,
2010). This means that, although forensic
‘hits’ might be valuable as evidence in
court, they may not be available during the
investigation and so cannot be used in the
search for the offender. When forensic evid-
ence is unavailable, behavioural analysis can
be used to identify a linked series of offen-
ces (Bennell & Jones, 2005; Grubin et al.,
2001; Hazelwood & Warren, 2003; Wood-
hams, Bull, & Hollin, 2007; Woodhams &
Toye, 2007). This is known as case linkage.
Two principles underpin case linkage:
behavioural consistency, and behavioural
distinctiveness (Bennell et al., 2009; Santtila
et al., 2005; Woodhams et al., 2007). Taken
together, these hypotheses assume that an
offender will behave in the same way across
his or her offences (behavioural consistency)
but that this behaviour will be different
from the way in which other offenders
behave (behavioural distinctiveness). This is
important because if offenders are con-
sistent in their behaviour but all behave in a
similar fashion then it would be impossible
to distinguish the crimes of one offender
from those of another (Santtila et al., 2005).
Offenders must therefore commit their
offences in a consistent but distinctive man-
ner for case linkage to be feasible (Wood-
hams et al., 2007). Academic exploration of
case linkage (eg, testing the principle of
behavioural consistency) has traditionally
concentrated on serious offending (such as
rape and murder) (eg, Bateman & Salfati,
2007; Hazelwood & Warren, 2003; Salfati &
Bateman, 2005; Sorochinski & Salfati,
2010), however, there is a growing body of
research exploring how case linkage can be
used to link volume crimes including
vehicle crime (Tonkin, Grant, & Bond,
2008) and burglary (Bennell & Jones, 2005;
Markson, Woodhams, & Bond, 2010). Case
linkage in real-world investigative settings is
typically conducted by crime analysts work-
ing within police forces and other specialist
crime agencies (Woodhams & Toye, 2007).
This is known as Comparative Case Analysis
(CCA).
This paper reports the key findings of a
survey of analysts’ views of CCA. The cur-
rent research is exploratory and aims to
gather analysts’ experiences of CCA to gain
an understanding of how case linkage oper-
ates in a real-world setting. Analysts’
responses to the survey allow for a prelim-
inary examination of how analysts’ know-
ledge and experience of CCA compares
Burrell and Bull
Page 3
with the key findings in the academic liter-
ature on case linkage.
METHODOLOGY
The research gathered the views of analysts
working in two police forces in the UK;
one urban and one rural. These two forces
were selected due to their engagement in a
wider piece of research (currently being
undertaken by the first author) on case
linkage.
Ethical approval to use the survey was
granted by the University, and senior police
staff were approached in each force to seek
permission to disseminate the survey. Once
permission was granted, the invitation to
participate in the research was channelled
through senior analytical staff. As it is chal-
lenging to gain access to online survey tools
on police computers due to security fire-
walls, analysts received the survey as a Word
document via email. Once complete, the
surveys were returned to the first author via
email.
Tool development
The survey tool was developed by the first
author with support and feedback from the
second author. Questions were developed
using both the academic literature on case
linkage (eg, to identify key issues in case
linkage such as the problems of falsely link-
ing crimes that are not committed by the
same offender and/or failing to identify
links between offences) and the first
author’s own practical experience of work-
ing alongside crime analysts (eg, phrasing
questions using language familiar to ana-
lysts). In addition, two academic colleagues
who work in the field of case linkage
commented on the draft survey, and their
feedback was incorporated into the final
tool (which had 23 questions).
As the research focuses on information
gathering, it was essential that the survey
tool was flexible in order to capture the
necessary information. Therefore the survey
primarily consisted of open questions. Tra-
ditionally, closed questions are preferred as
these are quick to answer and easier to code
for analysis (Babbie, 1990; de Vaus, 1996).
However, closed questions rely on a finite
number of responses which might not
encompass all relevant options. This can
lead to misleading results (de Vaus). This
problem was demonstrated clearly in this
research as the list of evidence types pro-
vided in the survey excluded ‘offender
description’ which was popularly cited by
analysts in their responses to the subsequent
open question asking them which other
evidence they use in CCA. Although
written questionnaires do not lend them-
selves to the use of open questions (de Vaus)
and the use of open questions increases the
time it takes to complete the survey, it was
felt that it was important to allow analysts to
express their views freely. The questions
focused on:
●why analysts conduct CCA;
●the process of completing a CCA (eg,
how long it takes to complete a CCA,
decision-making around whether cases
are linked);
●the evidence used to complete a CCA;
●benefits and challenges to CCA.
Sample
The survey was disseminated to an estim-
ated 72 analysts in two UK police forces
between June and September 2009. A quar-
ter of analysts (n = 18) completed and
returned the survey. One of the analysts
who completed the survey reported that
they had not been in post long, however,
they had previously worked for an organisa-
tion specialising in CCA work, and were
therefore able to comment extensively on
CCA. For this reason, their responses were
not excluded from the sample.
It is recognised that the response rate was
low and the sample size is small, and the
Examination of crime analysts’ views and experiences of CCA
Page 4
authors acknowledge that the results of this
research are not necessarily generalisable to
other analysts within the forces surveyed,
other UK forces, or indeed to police
agencies in other countries. However, low
response rates are not uncommon when
surveying police staff. For example, Jamel,
Bull, and Sheridan (2008) reported receiv-
ing just 19 responses to a survey dissemin-
ated to 300 Sexual Offences Investigative
Techniques (SOIT) officers, despite sending
out reminders. Weir and Bangs (2007) also
reported low response rates in their work
surveying crime analysts across England and
Wales about how they use geographical
information systems (GIS). One probable
reason for this is the difficulty in gaining
access to potential participants: demand for
police staff time is high and it can be
difficult to obtain permission to interview
or survey staff as participation in research
impacts on their operational time. How-
ever, Jamel et al. and Weir and Bangs both
reported that, although the sample was
small and therefore not representative of the
total population, their research was useful to
explore the topics under investigation. Sim-
ilarly, the current research aimed to explore
how case linkage operates in a real-world
setting. Therefore, the results seek merely to
open a discussion about analysts’ views and
experiences of CCA rather than form a
comprehensive assessment of the topic. It
was felt that the sample size was adequate
for this purpose.
Qualitative analysis
The survey responses did not lend them-
selves to traditional content analysis coding
in that it was difficult to create discrete and
meaningful categories for analysis. For
example, the survey asks analysts how long
it takes to conduct a CCA. As expected, this
was dependent on a multitude of factors
making it difficult for analysts to provide an
estimate. The discussion of this issue, there-
fore, focuses on the factors that analysts
reported affected the time it takes to com-
plete a CCA rather than the estimated times
provided.
The decision to use open questions
proved to be prudent as the survey
responses provided were very detailed.
Qualitative analysis of the transcripts
revealed common themes and experiences.
These are reported in the results section,
with quotes used to highlight salient points.
Unique identifiers are used to protect par-
ticipant anonymity.
RESULTS
The small sample size negates the possibility
of comparing findings across the two forces.
In fact, the responses were similar. There-
fore, the results are presented in a combined
format.
The analysts worked in a variety of
departments, primarily Basic Command
Units (BCUs) (ie, police operational dis-
tricts usually headed by a Chief Super-
intendent), community safety units (ie,
departments that focus on tackling com-
munity issues such as anti-social behaviour
and crime), and the force intelligence units
(ie, analytical units that concentrate on
force-wide crime issues). A number of spe-
cialist departments were also represented,
for example, the confidential unit (which
works with highly sensitive intelligence),
the counterterrorism unit, and the Auto-
matic Number Plate Recognition (ANPR)
department (which deals with intelligence
about car registration plates). The majority
of the analysts were female (n= 15); three
were male. The majority of the analysts had
at least two years experience of working in
an analytical role (n= 14 out of 18, or 78
per cent) and 2 out of 18 (11 per cent) were
working at a senior analyst level.
All of the analysts routinely work on
serial crime and all have experience of
conducting CCA. CCA is conducted on a
wide range of offences with the offence
Burrell and Bull
Page 5
types typically dictated by the department
in which an analyst works. CCA for serious
offences, such as rape and murder, is likely
to be conducted by force-level analysts and
specialist departments. BCU-level analysts
tend to work on a wider range of offence
types linked to BCU priorities, which vary
by BCU (although burglary, robbery and
car crime were commonly cited). Analysts
rarely considered themselves to specialise in
linking a particular type of offence.
How do you conduct a CCA?
There are two different approaches to con-
ducting a CCA, namely (1) identifying all
the offences committed by a known
offender, and (2) identifying individual
series within an offence type. Analysts,
depending on their preferences and how
they were tasked, used both of these
methodologies. For example, one analyst
commented ‘It depends what [the] starting
point is and the specific tasking. Even if I’ve
been tasked to look at an offence type, I
wouldn't ignore any links to other crime
types that are jumping out’ (participant
N-5). However, there were some concerns
about linking by offender, for example: ‘In
my experience, the majority of CCAs are
completed by matching the crime types and
not the offenders. If you attempt to create a
CCA by matching the offenders you are
assuming they are guilty’ (W-1). The con-
cerns of this analyst are understandable as it
is possible that the investigation could be
impeded if focus is placed on a suspect who
is subsequently found to be innocent.
However, a retrospective CCA can be
beneficial to build a case against a suspect.
It was reported that searches for informa-
tion should start wide but that a clear
rationale is required for each search. All but
one analyst reported using computer soft-
ware to conduct a CCA. The most com-
monly cited software packages were Excel
and i2. An example of a typical process for
conducting a CCA was to retrieve data
from databases using force systems (eg,
Business Objects or Discoverer), and build
and complete a matrix of linking factors in
Excel. Findings are typically displayed using
graphical tools within Excel, i2, and/or
mapping software (such as MapInfo).
The use of a matrix provides an easy
method of recording and visualising the
links between different offences across
themes. The literature suggests that linking
across multiple categories (ie, identifying
unique combinations of behaviours) is
important when linking offences (Hazel-
wood & Warren, 2003; Woodhams et al.,
2007). This research indicates that analysts
share this view. For example, when asked
how they reduce the chances of falsely
identifying cases as linked, W-4 stated
‘. . . by ensuring the identifying factors (ie,
location, description, time of day etc) are
the same/similar in most cases, ie, seven out
of the ten factors match the trend’.
Analysts reported that it is important that
each link is reassessed prior to submitting
the CCA report, and that clear explanations
are included about why cases have or have
not been included in the series. Weighted
links demonstrate the strength of the rela-
tionship between offences and clarify
whether links are strong or tentative. In
some cases, it is useful to include a list of
offences that might be part of the series but
that have fewer links to the rest of the series.
This highlights offences that might be rel-
evant without misleading the investigative
team. This is not favoured by all analysts,
but some felt that they would prefer to
include all offences that might be linked
rather than exclude them from the analysis
completely.
Analysts reported that it is important to
take all available information into account
and analysts should liaise with the officer in
command (OIC) of the investigation to
ensure any new information is received
promptly and fed into the CCA.
Examination of crime analysts’ views and experiences of CCA
Page 6
What evidence do you use in CCA?
The survey provided a list of forensic,
temporal/spatial, and behavioural evidence
that might be used to link offences. Figure
1 shows how frequently these factors were
used to support CCA.
Figure 1 clearly shows that behavioural
evidence is routinely used to link offences.
The majority of analysts use temporal
factors and/or location to link offences.
Forensic evidence is used less frequently
than might be expected. There are a num-
ber of reasons for this. One analyst noted
that ‘CCAs are often commissioned early in
the investigative process prior to forensic
evidence being gathered’ (W-1) and so
forensic evidence might be unavailable. This
analyst also noted that ‘serious offences are
more likely to be attended by forensic spe-
cialists’ (W-1) and so analysts working on
less serious offences are unlikely to be able
to source forensic evidence for their CCA.
Two analysts (both from the same force)
reported that ‘forensic links will be high-
lighted by the forensic team’ (N-5) and that
their force employs ‘specific forensic analysts
to look at DNA/fingerprints etc so does
not really fall under my remit’ (N-2).
Analysts reported a number of other
types of evidence used in CCA. The most
commonly cited types of information (ie,
by 13 or more analysts) were spatial and
temporal trends (eg, the distance between
offence locations) and the offender descrip-
tion. Less commonly cited evidence
included vehicle registration numbers,
methods for committing the offence (eg,
how the offender controlled the victim,
how the offender responded to resistance
from the victim), victim information (eg,
victim demographics, victim statements)
and witness statements. Although informa-
tion from witnesses/victims was considered
Figure 1
Evidence used in the
CCA
Burrell and Bull
Page 7
useful to the CCA process, there were con-
flicting views about which offences were
more/less likely to include such informa-
tion. For example, one analyst stated ‘Burg-
laries and robberies will also often have
numerous witnesses to support the case.
Stranger rapes for example will occur in
isolation and very often the only evidence
will be that of a scared and confuse[d]
victim, this often clouds statements’ (W-1).
However, another analyst reported ‘Burg-
lary and vehicle crime are more difficult
because MOs are so generic . . . and little
information is known about the offence
because the victim is usually not present’
(N-4).
The level of information available about
offences varies by crime type. This is either
because more behaviours can be observed
(eg, in offences where there is more likely
to be a witness) or because more attention
is paid to recording details (eg, in serious
violent crimes): ‘It’s simply the amount of
information recorded against each crime
that makes it easier to identify series. I
would think murder and rapes would be
easier as fewer crimes and a lot more
information captured. Some burglaries sim-
ply state “unknown MO” and have no
statements, a wide time span etc, which is
likely to lead to linked crimes being missed
through the analytical process’ (N-5). A lack
of information and/or receiving generic
crime information therefore makes some
offences more difficult to work with:
‘[t]heft of vehicle is often more difficult
because of a lack of MO, ie, the car has
gone and has not been recovered. More
minor crimes such as breaking into vehicles
are easier to link tentatively because they are
often closely geographically concentrated,
but the simple MOs often used (window
broken, sat-nav taken) means that they are
hard to link definitively’ (W-8). Thus, some
offences are potentially more difficult to
conduct CCA with not only because of the
type of offence, but also due to a lack of
information being gathered about how cer-
tain offences are committed. This emphas-
ises the importance of robust data-recording
practices to ensure high-quality data are
collected for all offence types.
Analysts highlighted the importance of
taking the reliability of data into account
when making judgements about whether
offences are linked. One analyst provided an
example of this: ‘I am wary of using ethni-
city to link robbery offences . . . particularly
when the suspects are wearing balaclavas as
a witness will often confuse a tanned white
male, with a light skinned black male, Asian
male, mixed race etc if their face is partially
obscured’ (W-5). Data reliability has been
consistently highlighted as a barrier to
effective CCA in the academic literature
(eg, Tonkin et al., 2008; Woodhams et al.,
2007) and it is unsurprising that this opin-
ion is shared by frontline analytical staff.
Furthermore, previous research has indi-
cated that analysts commonly cite data qual-
ity as a barrier to effective analysis (eg, Weir
& Bangs (2007) reported that a third of
their respondents stated that the quality of
the data available for mapping crime was
insufficient or very poor).
How long does it take to complete a
CCA?
A single CCA can take anything from
20–30 minutes to several weeks to com-
plete. Typically, this was dependent on cir-
cumstance: for example, a quick CCA for
immediate use where there is little available
information can take less than an hour,
whereas an exhaustive search of all informa-
tion relating to a large number of crimes
and the need for a detailed report could
lead to the CCA taking at least a week to
complete. Analysts commonly reported that
estimating the time it takes to complete a
CCA is difficult as it is dependent on a
variety of factors. The key theme is that any
factor that increases the volume of informa-
tion to be examined will substantially add
Examination of crime analysts’ views and experiences of CCA
Page 8
to the time it takes to complete a CCA; for
example, severe cases, more offenders,
increasing the size of the geographical area
and increasing the time frame under
analysis.
Other factors reported to affect the
length of time it takes to complete a CCA
included:
●The need to source information from
multiple systems. Often information is
stored in different crime systems that all
need to be searched. For example, W-8
noted that if ‘many vehicles/mobiles [are]
used in the commission of the offence
[this] will increase the time, as they will
each need to be checked under different
systems’. In addition, some systems take
longer to access than others.
●The need to source information (eg,
forensic intelligence) from external units
and/or other forces. In these cases, the
analyst is dependent on the speed at
which these units/forces can provide
information.
●The complexity of the behaviours dis-
played by the offender as some behavi-
ours are more difficult to interpret than
others.
●Data quality — eg, incomplete reports,
poor/inaccurate recall by victims/
witnesses, incomplete interviewing. This
can mean the analyst has to spend more
time sourcing additional information.
●Strength of similarities — if cases are
very similar they are easier to link, mak-
ing the CCA process quicker.
●Tasking requirements — requesting high
levels of information and detail in the
CCA increases the time it takes to com-
plete. Poor quality tasking requests can
delay the CCA as the analyst needs to
seek clarification of what is required.
●Timescales — how long analysts have to
work on the CCA and how quickly the
work is needed. For example, where an
offender is in custody the CCA may be
needed very quickly and so the analyst
may spend less time on the CCA in these
circumstances.
●The need to prioritise other work/
workload can reduce the time the analyst
has to spend on the CCA.
The proportion of an analyst’s time spent
conducting a CCA varies widely and this is
largely dependent on tasking. For example,
an analyst might spend 80 per cent of his or
her time on CCA in one week but just 10
per cent the next week. CCA was, how-
ever, a regular task. The two senior analysts
spent less time completing CCA as their
role is managerial; they had, however, spent
substantial proportions of their time on
CCA in the past. Their CCA-related work
now centred on managing teams of analysts
who routinely complete CCA.
Which offences are easier to link and
why?
Volume offences, such as burglary, were
considered easiest to link by some analysts.
For example: ‘There may be more Vehicle
Crime, Burglary or Robbery offences,
making it quicker to identify common fac-
tors’ (W-4). However, the majority felt that
having high volumes of offences to scan
hindered the linking process: ‘I would say
murder/rapes easier as not so many. BDHs
[burglary in a dwelling] robberies and car
theft would be harder as there are more
crimes to trawl through’ (W-2).
For the most part, it was reported that
‘[a]ny crime which involves human contact
is likely to be easier to link as a person’s
behaviour is often very individual and
therefore significant’ (W-10). As N-2 notes:
‘with face to face crimes, ie, personal rob-
bery, violent and sexual offences, the
offender’s description/MO [modus
operandi]/motives/mannerisms are often
key to linking offences but this is often
absent in burglary and thefts’. Thus it is
clear that distinctive behaviour is important
Burrell and Bull
Page 9
when linking crimes: ‘Rape and Burglary I
would say are the easiest to try and link as
there may be more links if offenders have
certain ways of doing things’ (W-9). How-
ever, distinctive behaviours were not always
present in personal crimes: ‘I found it very
difficult to conduct CCA on murders —
particularly where there is very little dis-
tinctive about the scene, the way the victim
was left etc as there is so little known
behavioural information’ (W-12). Where
distinctive behaviours are absent, it can
be difficult to link offences: ‘When, for
example, a vehicle with the same VRM
[vehicle registration mark] has committed a
number of different types of offences, then
they are easy to link. It is often more
difficult to link, say, burglary and vehicle
crime in an area, unless the temporal pat-
terns are unusual’ (W-8).
Overall, distinctive features were high-
lighted as particularly useful to the linking
process, for example, burglars eating food in
the houses they are burgling is rare and so
finding this behaviour across crimes would
suggest that these offences are linked. This is
one of the situations where the local
knowledge and experience of analysts was
cited as key, as they are able to identify
when a behaviour was rare in their area.
These findings indicate that analysts are
considering the theoretical assumption of
behavioural distinctiveness in their CCA
work. It is therefore important that analysts
are able to access a high level of information
for each offence to increase the chances of
identifying distinctive behaviours with
which to link offences.
What are the benefits of CCA?
The analysts were asked if they agreed with
a series of statements about the benefits of
CCA, the results of which are presented in
Table 1.
Most analysts agreed with the provided
statements, but there were some exceptions.
In one case, the analyst explained why he
disagreed with a statement: ‘I don’t think it
develops our understanding of crime prob-
lems as once CCAs are completed at a
regional level they are not reviewed or
debriefed for best practice’ (W-1). This sug-
gests that CCA is not currently boosting
wider knowledge of offending patterns.
Analysts largely felt that CCA is valuable
and independently identified a number of
additional benefits of CCA work. Several
themes emerged including the development
of crime prevention tactics (eg, offender
targeting, advice to potential victims),
understanding patterns of offending (eg,
helping to understand motive, and predict-
ing future offending) and support for the
courts (eg, building a robust evidence base,
and ensuring appropriate sentences are
passed). Identifying strong links between
offences allows the police to charge
offenders with multiple offences once they
are apprehended. CCA also supports work
to encourage offenders to admit to offences
Table 1: Agreement with statements regarding the benefits of CCA
Statement Number
of analysts
% of
analysts
Do you think the identification of series assists with any of the following:
Detection rates/catching offenders 18 100
Prioritising suspects 17 94
Improve the efficiency of investigations 17 94
Developing our understanding of particular crime problems (e.g. burglary) 15 83
Examination of crime analysts’ views and experiences of CCA
Page 10
where there may be insufficient evidence
for a formal charge (a process known as
‘taken into consideration’ (TIC)). CCA
provides a key opportunity to alert other
police forces, or BCUs within the same
force, of crime series that they might not be
aware of. CCA can also be used as a public
awareness tool to boost public confidence
in the police by highlighting where
offenders have been apprehended for mul-
tiple offences.
The practical benefits of CCA were also
highlighted. The CCA not only provides a
concise account of a large amount of data
and information, but also an easy-to-
interpret version of accounts for investiga-
tive teams. CCA has the additional benefit
of being a living document that can be
refreshed as the investigation proceeds.
What are the challenges for CCA?
Data availability, accessibility and quality
Analysts reported that the standard of a
CCA is dependent on data availability,
accessibility and quality. As discussed above,
difficulties accessing information from
external agencies and/or multiple computer
systems, along with poor data quality, sig-
nificantly impact on the time it takes to
complete a CCA. These difficulties also
impact on the reliability of the CCA, and
therefore the usefulness of the CCA to the
investigative team. It was widely agreed that
the reliability of a CCA is wholly depend-
ent on the quality and amount of informa-
tion that analysts have to work with and
analysts highlighted the importance of tak-
ing the reliability of data into account when
making judgements about whether offences
are linked. These findings are unsurprising
given that the literature has repeatedly high-
lighted that police data can be of poor
quality and that this hinders analysis (e.g.
Santtila et al., 2005; Tonkin et al., 2008;
Weir & Bangs, 2007; Woodhams et al.,
2007).
Offence type
It was reported that some offences are easier
to link than others. However, there were
mixed — and often conflicting — opinions
about which offences are easier to work on.
These differences are probably linked to
data quality. For example, one analyst
reported that ‘robbery is probably most dif-
ficult as the offences tend to be opportun-
istic’ (N-1) suggesting limited distinctive
information was available about these cases,
but another stated ‘[r]obbery gives you the
most factors to analyse with analysing the
IP [injured party] and offender/weapon
seen/speech used etc’ (W-2). Interestingly
these statements are from analysts in differ-
ent forces. This could explain the discrep-
ancy, particularly if more information is
routinely collected in robbery cases in one
force than the other (eg, if robbery is a
priority in one force it is possible that
officers are encouraged to gather more
data). Alternatively, the differences could be
based on the individual analysts’ differing
experiences of conducting CCA with rob-
bery: W-2 has been working in an analytical
role for five to ten years, compared with
one to two years for N-1, which might
mean that W-2 has more experience of
completing CCA on robbery and/or has a
more developed understanding of what
information to source for the CCA. Finally,
the difference could be because robbery
offences have different features in different
areas; perhaps robberies are more organised
and/or distinctive in W-2’s force area and
therefore there is more behavioural
information to gather about these offences.
Linking across offence types
One analyst outlined the difficulties of link-
ing across offence types: ‘In my experience
a CCA will only work when comparing the
same crime type. It’s harder to compare a
crime series incorporating different types of
offences’ (W-5). However, another noted
the importance of considering the range of
Burrell and Bull
Page 11
offences which a single offender commits:
‘We look at all crime types that each indi-
vidual may have committed, for example
due to the fact that sex offenders generally
have some history of violence, we also take
into account violent offenders when invest-
igating rapes’ (W-10). It is suggested that
this discrepancy centres on the approach to
CCA used, with the first analyst concen-
trating on identifying series from within
offence groups and the second using an
offender-centred approach.
Being aware of caveats
One of the biggest challenges in CCA is
minimising false positives (ie, inaccurately
identifying offences as linked) and false neg-
atives (ie, failing to link offences that are
committed by the same offender) and aca-
demic studies have explored how to max-
imise the number of accurate links whilst
minimising false alarms (eg, Bennell et al.,
2009). Analysts are aware of this issue and
proactively take steps to improve accuracy.
It was emphasised that it is important to
keep an open mind and not make assump-
tions about people, places and/or circum-
stances. The analyst must remain objective
in order to ensure that CCA work is not
subject to bias; a task that can be challeng-
ing if the tasking officer has preconceptions
about the case in question. Developing and
testing hypotheses was considered to be a
useful boost to linking accuracy. It is reas-
suring that analysts are aware of the risk of
false positives and false negatives and apply
caution when interpreting links to ensure
the CCA is reliably presented.
As alluded to already, analysts stressed
that it is important to highlight caveats
where these exist. For example: ‘Also, the
receiver must be aware of the human error
factor. The CCA method is not finite and is
always subject to the possibility of human
error or false interpretation’ (W-1). This
allows the investigative team to apply cau-
tion where appropriate when using the
CCA information. A potential solution to
minimise the impact of human error is to
have an analyst colleague review and/or dip
sample the evidence. Seeking a second
opinion provides some reassurance that
information has been interpreted appro-
priately, particularly where links are tentat-
ive. Keeping an open dialogue with
colleagues is therefore key to the CCA
process.
Recommendations for future
directions in CCA
Analysts made a number of recommenda-
tions for the further development of CCA.
Most of these recommendations centred on
improving data collation and recording.
Analysts stressed the importance of collect-
ing as much information about offences as
possible, and ensuring this information is
captured accurately in the crime reports.
The development, and consistent use, of a
more robust data coding system for modus
operandi behaviours was also recommended
to boost the accuracy of linkage work.
From a practical perspective it was high-
lighted that having dedicated time to con-
duct CCA would improve the quality of the
work. More timely access to data would
facilitate the CCA process; in particular it
was suggested that investing in more soph-
isticated computer software could substan-
tially reduce the time spent searching for
information. Building and maintaining
effective communications between BCUs,
and between police forces, would allow
quicker access to information for CCA, and
might also yield supplementary datasets that
could be included in CCA work.
One analyst highlighted the need for
more research: ‘What would also help
would be a continuing research particularly
around behavioural consistency (what parts
are more stable and what parts aren’t)’
(W-12). This is an encouraging finding for
Examination of crime analysts’ views and experiences of CCA
Page 12
academics working in the field of case link-
age. The focus of applied research is to
ensure findings can be of practical use, and
so it is reassuring that ongoing research on
behavioural consistency would be con-
sidered useful for analysts working on the
frontline.
It is clear that analysts share the view of
the academic community that data access
and quality needs to be improved to
enhance the ability to link offences accur-
ately. However, analysts also identified more
practical issues (such as the negative impact
of poor quality tasking and the importance
of being allowed adequate time to complete
the CCA) that are less likely to be identified
as barriers to CCA by the academic com-
munity. This demonstrates the value of
surveying analytical staff, alongside the
quantitative research, when developing
CCA techniques.
DISCUSSION
The CCA provides a concise account of a
large amount of data and information and
analysts believe these to be beneficial to
criminal investigations. The research clearly
indicates that CCA forms a key part of the
analyst’s role, and that a range of informa-
tion, including temporal and spatial trends,
forensic evidence, MO behaviour, and
offender/victim characteristics, are used to
link offences committed by the same
offender. Analysts emphasised the import-
ance of linking across multiple factors and
the value of using distinctive behaviours to
build strong links between cases.
However, analysts highlighted a number
of challenges in CCA that need to be
addressed to improve the quality and value
of the CCA product. Most notably, analysts
expressed concerns about data availability,
accessibility and quality. This is a view
shared by the academic community, mem-
bers of which have reported that the quan-
tity and quality of information varies widely
from case to case (Santtila et al., 2005) and
expressed concerns that selective question-
ing by police officers can result in details
they deem to be irrelevant being ignored or
omitted from crime reports (Canter &
Alison, 2003, cited in Woodhams et al.,
2007; Tonkin et al., 2008). There are a
number of ways in which data quality could
be improved, including interviewing vic-
tims more skilfully (Milne & Bull, 1999)
and developing a robust data coding system
for MO behaviours. There might also be
opportunities to incorporate more forensic
evidence into CCA as police forces increase
the number of crime types attended by
forensic teams. For example, local docu-
mentation (dated February 2010) from one
of the forces indicates that 100 per cent of
burglary dwelling scenes are now attended
by the Scenes of Crime Officers (SOCO)
to maximise forensic recoveries and obtain
evidence to detect these crimes. The devel-
opment of computer systems that allow data
to be input, updated and stored more effi-
ciently, and strengthening communication
networks with external organisations,
would facilitate faster access to information.
This would boost the level and quality of
information available for analysis.
This research has indicated that analysts’
concerns about CCA commonly mirror
those of academics. For example, W-1
reported that a scared and confused victim
(in rape cases) might not be able to provide
much information. This supports concerns
expressed by Woodhams et al. (2007) that
the trauma experienced by the victim dur-
ing the offence can adversely affect memory
recall. However, analysts also outlined more
practical problems (such as lack of clarity in
tasking reports and the need to be able to
dedicate time to CCA) that are not high-
lighted in existing academic research. This
clearly demonstrates the value of gathering
the views of frontline staff when researching
case linkage.
Burrell and Bull
Page 13
CONCLUSION
This paper has discussed the findings of a
survey of 18 analysts, working in a range of
departments across two police forces, about
their views and experiences of CCA. It is
acknowledged that these findings are indic-
ative rather than conclusive; however, they
do highlight that analysts are routinely con-
ducting case linkage on a variety of offence
types (ranging from car theft to murder) to
support the investigative process, and that
they are using behaviour to identify series
of offences. It is important to continue
research in this area as failure to link
offences accurately can waste police
resources, delay the investigative process
and/or reduce the chances of apprehending
the offender (Bennell, Bloomfield, Snook,
Taylor, & Barnes, 2010; Grubin et al.,
2001). It is, therefore, important that the
academic community continues to support
case linkage work. It is suggested that focus-
ing on continuing to build the evidence
base for behavioural consistency, and identi-
fying which behaviours are the most reli-
able indicators that cases are linked, will be
beneficial to analysts.
Expanding this research, using more in-
depth data-gathering techniques (such as
one-to-one interviews or focus groups) to
gather the views and experiences of a larger
sample of analysts would also be beneficial
to identify opportunities to support the
development of operational CCA work.
Interviews with police officers who request
CCA reports would also be valuable; for
example, to identify examples of scenarios
when an officer will request a CCA, what
they hope to achieve by requesting a CCA
and feedback about how the CCA was used
to provide leads for the investigation.
Finally, it would be useful to review a
sample of CCA reports to examine the
specific types of information used and how
links are formed between cases. This would
be a challenging task as the content of CCA
reports will be restricted due to the like-
lihood of containing personal information
about offenders and/or victims. However, if
adequate security clearance could be
secured and permission granted from police
forces to review this material, this would be
greatly beneficial.
R
EFERENCES
Babbie, E. (1990). Survey research methods (2nd
ed.). California: Wadsworth.
Bateman, A. L., & Salfati, C. G. (2007). An
examination of behavioral consistency using
individual behaviours or groups of
behaviours in serial homicide. Behavioral
Sciences and the Law, 25, 527–544.
Beaver, K. M. (2010). The promises and pitfalls
of forensic evidence in unsolved crimes.
Criminology & Public Policy, 9, 405–410.
Bennell, C., Bloomfield, S., Snook, B.,
Taylor, P., & Barnes, C. (2010). Linkage
analysis in cases of serial burglary:
comparing the performance of university
students, police professions and a logistic
regression model. Psychology, Crime & Law,
16, 507–524.
Bennell, C., & Jones, N. J. (2005). Between a
ROC and a hard place: a method for
linking serial burglaries by modus operandi.
Journal of Investigative Psychology and Offender
Profiling, 2, 23–41.
Bennell, C., Jones, N. J., & Melnyk, T. (2009).
Addressing problems with traditional crime
linking methods using receiver operating
characteristic analysis. Legal and
Criminological Psychology, 14, 293–310.
Canter, D., & Larkin, P. (1993). The
environmental range of serial rapists. Journal
of Environmental Psychology, 13, 63–69.
Clarke, R. V. & Eck, J. (2003). Become a
problem solving crime analyst in 55 small steps.
London: Jill Dando Institute of Crime
Science.
Cole, S. A. (2010). Forensic identification
evidence: utility without infallibility.
Criminology & Public Policy, 9, 375–379.
de Vaus, D. A. (1996). Surveys in social research
(4th ed.). London: UCL Press.
Ewart, B. W., Oatley, G. C., & Burn, K.
Examination of crime analysts’ views and experiences of CCA
Page 14
(2005). Matching crime using burglars’
modus operandi: a test of three models.
International Journal of Police Science &
Management, 7.3, 160–174.
Grubin, D., Kelly, P., & Brunsdon, C. (2001).
Linking serious sexual assaults through behaviour
(Home Office Research Study 215).
London: Home Office.
Hazelwood, R. R., & Warren, J. I. (2003).
Linkage analysis: modus operandi, ritual,
and signature in serial sexual crime.
Aggression and Violent Behavior, 8, 587–598.
House of Commons Science and Technology
Committee. (2005). Forensic Science on Trial.
London: The Stationery Office.
Jamel, J., Bull, R., & Sheridan, L. (2008). An
investigation of the specialist police service
provided to male rape survivors. International
Journal of Police Science and Management,
10(4), 486–508.
Markson, L., Woodhams, J., & Bond, J. (2010).
Linking serial residential burglary:
comparing the utility of modus operandi
behaviours, geographical proximity, and
temporal proximity. Journal of Investigative
Psychology & Offender Profiling, 7, 91–107.
Milne, R., & Bull, R. (1999). Investigative
interviewing: psychology and practice.
Chichester: Wiley.
Parliamentary Office of Science and
Technology. (2006). Postnote: the national
DNA database. London: Author.
Salfati, C. G., & Bateman, A. L. (2005). Serial
homicide: an investigation of behavioural
consistency. Journal of Investigative Psychology
and Offender Profiling, 2, 121–144.
Santtila, P., Fritzon, K., & Tamelander, A.
(2004). Linking arson incidents on the basis
of crime scene behavior. Journal of Police and
Criminal Psychology, 19, 1–16.
Santtila, P., Häkkänen, H., & Fritzon, K.
(2003). Inferring the characteristics of an
arsonist from crime scene behaviour: a case
study in offender profiling. International
Journal of Police Science and Management, 5.1,
1–15.
Santtila, P., Junkkila, J., & Sandnabba, N. K.
(2005). Behavioural linking of stranger
rapes. Journal of Investigative Psychology and
Offender Profiling, 2, 87–103.
Sorochinski, M., & Salfati, C. G. (2010). The
consistency of inconsistency in serial
homicide: patterns of behavioural change
across series. Journal of Investigative Psychology
& Offender Profiling, 7, 109–136.
Tilley, N. & Laycock, G. (2002). Working out
what to do: evidence-based crime reduction
(Crime Reduction Research Paper 11).
London: Home Office.
Tonkin, M., Grant, T., & Bond, J. W. (2008).
To link or not to link: a test of the case
linkage principles using serial car theft data.
Journal of Investigative Psychology and Offender
Profiling, 5, 59–77.
Weir, R., & Bangs, M. (2007). The use of
geographic information systems by crime analysts
in England and Wales. London: Home
Office.
Woodhams, J., Bull, R., & Hollin, C. R.
(2007). Case linkage: identifying crimes
committed by the same offender. In
R. N. Kocsis (Ed.), Criminal profiling:
international theory, research, and practice
(pp. 117–133). Totowa, NJ: Humana Press.
Woodhams, J., & Toye, K. (2007). An
empirical test of the assumptions of case
linkage and offender profiling with serial
commercial robberies. Psychology, Public
Policy and Law, 13, 59–85.
Yokota, K., & Watanabe, S. (2002). Computer-
based retrieval of suspects using similarity of
modus operandi. International Journal of Police
Science & Management, 4.1, 5–15.
Burrell and Bull
Page 15
