Fire Safety Journal 62 (2013) 30–36

Contents lists available at ScienceDirect

Fire Safety Journal

journal homepage: www.elsevier.com/locate/firesaf

Using the vulnerable localities index to identify priority areas for

targeting fire safety services
Spencer Chainey n
University College London, Department of Security and Crime Science, 35 Tavistock Square, London WC1H 9EZ, United Kingdom

art ic l e i nf o a b s t r a c t

Available online 21 May 2013 The Vulnerable Localities Index (VLI) is a popular analytical method used by police agencies in England
Keywords: and Wales to identify residential neighbourhoods that should be prioritised for attention. The VLI uses
Vulnerable Localities Index data on crime, deprivation and socio-demography to form a single composite index value for small
Priority neighbourhoods geographic areas, which when mapped can assist in determining where to target neighbourhood policing
Analysis and crime prevention initiatives. Its use has also helped to prompt other public sector partners such as
Deliberate fires local municipal government agencies to recognise the collaborative role that they could play with the
Malicious false alarms police in helping to address mutual issues of public safety. Fire safety agencies have also expressed
Fire prevention interest in the VLI, but to date this has mainly involved offering a supporting role to the police and local
community safety partnerships on crime specific matters of public safety, with a focus on fire safety often
on the periphery. In part this is due to limited analysis into whether the priority areas identified by the
VLI are also areas of key concern to fire services. This research explores this relationship, identifying that
the priority neighbourhoods of interest to the police are also areas where there are significantly higher
than expected levels of deliberate fires and malicious false alarm calls to the fire service. These findings
suggest that in the spirit of partnership working, potential opportunities exist in achieving mutually
beneficial gains in improving public safety through the collaboration of fire, police and other local public
service delivery.
& 2013 Elsevier Ltd. All rights reserved.

1. Introduction Hotspot analysis has become a popular method for identifying

An intelligence-led focus in policing and law enforcement has
been a contemporary paradigm that has placed significant empha-
sis on an analytically driven approach for targeting resources [19].
This has included using a range of analytical methods for identify-
ing and understanding features relating to offending behaviour,
victimisation and vulnerability, and when and where crime
problems take place. In contrast, while prevention is often at the
heart of fire service delivery there has been very little focus in
practice on the use of similar intelligence frameworks and the use
of analysis to support this philosophy. The data features of
incidents of crime are similar in content to fire incidents: someone
commits the act, someone is on the receiving end of the act (e.g. is
personally victimised or requires an emergency service response
that addresses a matter of wider public safety concern, such as
extinguishing a refuse fire), with the incident occurring at some
place at some time. This suggests that analysis techniques used in
policing can be replicated for fire prevention and support a more
analytically driven approach to fire safety.
n
Tel.: +44 20 3108 3203.
E-mail address: s.chainey@ucl.ac.uk
where crime incidents concentrate and assisting in directing the
targeting of police responses to tackle these issues [7,20]. Whilst
hotspot analysis is an effective tool for helping to determine where
incidents may occur in the future [5], it is though singular in
dimension in how it aims to identify areas that require attention.
That is, it only uses data on previous incidents to determine where
these future incidents may occur, rather than considering other
variables. This can mean that within a landscape of increasing
partnership collaboration between public agencies that work
together to tackle public safety issues, a singular view of the
problem using data from only one agency can limit engagement
from a disparate range of agencies [8,16,13].
In England and Wales many police agencies use an analytical
technique called the Vulnerable Localities Index (VLI) to identify
neighbourhoods that require priority attention [3]. The VLI uses
data on crime, deprivation and socio-demographics to form
composite index values for small geographic areas, which when
mapped can assist in identifying these priority neighbourhoods. Its
use has not only supported the targeting of police resources, but
also other local partners responsible for improving public safety,
including the local municipal government and the probation/
corrections service [4,21]. Its use has also been considered for
supporting the fire service in its resource targeting (for example

0379-7112/$ - see front matter & 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.firesaf.2013.03.013
 S. Chainey / Fire Safety Journal 62 (2013) 30–36
see [1]), but to date its value has yet to be statistically tested with
any rigour.
In this article we explore the value of the VLI in identifying
areas that are of interest to fire safety services by exploring the
relationship between the locations of a range of fire incidents with
the priority areas identified using the VLI. Specifically, we test
three hypotheses:

The distribution of deliberate fires is not related to differences
in VLI scores

The distribution of malicious false alarms is not related to
differences in VLI scores

The distribution of accidental dwelling fires is not related to
differences in VLI scores.
We begin by describing the VLI, including a discussion on the
data it uses and how these are combined to form a single
composite index. This is supported with examples to help illus-
trate its popular use for a wide variety of community safety
interests. Section 3 then describes the method that was used to
analyse comparisons between VLI scores for each small area and
the spatial distribution of a range of fire incidents. Section 4
presents the results from the analysis, followed by a discussion
and conclusions on the value of the VLI for supporting the
targeting of fire safety initiatives.
2. The Vulnerable Localities Index
The VLI was created in 2004 as a means for helping police
agencies in England and Wales to identify residential neighbour-
hoods that required priority attention. This was in response to a
series of riots in Bradford, Oldham, Wrexham and Burnley in 2001
and the UK Government reviews that followed. These reviews
identified the requirement for community cohesion to be a central
aim of government, reflected in all policy making, and requiring
closer working between local government, police and community
leaders in order to counter the rise in social, economic and cultural
polarisation [10].
The VLI uses six variables: burglary to a dwelling, criminal
damage to a dwelling, income deprivation, employment depriva-
tion, population of young people and educational attainment.
These data were chosen following research into variables that
would help to identify residential areas where there was likely to
be issues of community breakdown, driven by an undercurrent of
crime and poor socio-economic conditions [3]. This included
drawing from the UK Government reviews of the riots in 2001
[10–12,14] and a history of research into social disorganisation
[2,26,6,27,15,17,18,29,25]. That is, which variables could accurately
identify neighbourhoods that would warrant further prioritised
attention by police agencies? The findings from the riots that took
place in England in 2011 [22] also validate the choice of variables
that form the VLI.
It is for similar reasons that the VLI has gained interest by Fire
and Rescue Services in the UK for identifying areas for prioritised
attention. That is, rather than working solely in a response
manner, and using retrospective data on fire incidents, can the
VLI help fire services identify areas of strategic interest for
directing fire safety and prevention programmes, particularly to
residential communities? And also, are the areas the VLI identifies
that are of strategic interest to the police and other local agencies
also of strategic interest to the fire service? Shaw and MacKay [26]
and more recently Sampson [25] have helped spirit a research
agenda that has shown that the areas that draw heavily from fire
response services (and other emergency services) are also places
of social disadvantage and social disorganisation. This paper aims
31
to identify whether the VLI, built on several of the principles
illustrated by Shaw and McKay [26] and others, is of practical use
to fire services and whether it offers an easy to apply analytical
tool that could help promote strategic fire safety and prevention
programmes.
The selection of the six VLI variables also had to fit a criteria
that would ensure a consistent national approach to determining
vulnerable localities, including consistency in the definition of
each variable (e.g. the police practice for recording a burglary to a
dwelling needed to be consistent in all police areas in England and
Wales), the data were available for small areas (including the
requirement for any crime data to be accurately geocoded and not
suffer significantly from issues of under-reporting), and the data
were easy to access by the agency that was calibrating the VLI. In
addition, the focus of the VLI to identify residential areas rather
than city centres and shopping malls meant that the crime data
associated with retail, entertainment and commercial centres
were not included as this would bias the identification of these
non-residential areas. The two crime variables included in the VLI
are available from police crime recording systems, the two
deprivation variables can be sourced from the UK's Index of
Deprivation, and educational attainment and the population of
young people can be sourced from the UK's Census. More details
on each data and their specific sources can be found in Chainey [3]
and even though it was developed for application in England and
Wales, the VLI has been replicated in several other countries
where similar crime recording, deprivation measurement and a
census of the population exist.
Output Areas (OAs)1 form the smallest standard form of
geographical unit in England and Wales for which population
and administrative data are collected, and in many cases form the
building blocks to determine neighbourhood areas. Each of the
two Census variables for each OA in England and Wales are
available to freely download from the UK's Office for National
Statistics (ONS) online Neighbourhood Statistics Service (http://
www.neighbourhood.statistics.gov.uk). The Census in England and
Wales is conducted every 10 years, and while there have been
some concerns that the accuracy of these data decays as time
passes between Census dates, the evidence suggests that the
characteristics of the population in these areas remain broadly
similar as time passes and that the Census data continues to be the
best and most legitimate measure of socio-demography at the
small area level [3]. Deprivation data are recorded at Super Output
Area level (SOA)2, are available for free download from the ONS
Neighbourhood Statistics Service, but require these data values to
be assigned to the OAs within each of their respective SOAs. Crime
data are available as geocoded point data, therefore require an
operation in a geographical information system that aggregates
these data as a count of crime for each OA. Once these data have
been sourced they can be loaded into a freely available VLI
Microsoft Excel template for calculation [28].
In order to aggregate these six variables to form the VLI, each
variable has to be converted into some normalised form. This
requires the crime counts for each OA to be converted to a crime
rate, and educational attainment and young population counts to
be converted into percentage representations. To do this requires
data on the number of households in each OA and data on the total
resident population in each OA to also be sourced from the Census
and included in the VLI Excel template to calculate these rate and
proportion values. Deprivation data in the UK is made available as
index values and do not require this normalisation treatment.
1
An Output Area has an average of 125 households. There are 175,434 OAs in
England (165,665) and Wales (9,769).
2
Super Output Areas have an average population of 1500. SOAs are built from
groups of OAs (typically five)
32 S. Chainey / Fire Safety Journal 62 (2013) 30–36
The average of each variable across the OAs in the study area
(including only those OAs where this value is greater than zero) is
then calculated and acts as the benchmark value against which
each OA value is converted into an index form [3,4]. The following
example helps to demonstrate this calculation:

There were four burglaries to dwellings in an OA over a 12-
month period. The number of households in this OA is 110. The
burglary dwelling rate for this OA is therefore 36.4 per 1000
households per annum. There are 1000 OAs in the study area.
In 90 of these OAs there was no crime. These 90 OAs are
excluded from the calculation that determines the average
crime rate across the OAs in the study area. The average
burglary rate across the remaining 910 OAs in the study area
is 20.3 burglaries per 1000 households per annum. The average
(20.3) acts as the benchmark. Index values for each OA are
calculated using the formula: (Rate in OA/average rate across all
OAs in the study area)100. The OA that experienced four
burglaries therefore has a Burglary Dwelling Index score of
179 [i.e. (36.4/20.3)100].
This process is repeated for each variable to determine indivi-
dual standardised index values for each variable and for each OA.
These six indexed values are then summed and divided by six to
determine a single composite Vulnerable Localities Index value for
each OA. The results can then be mapped and visualised using
choropleth mapping.
Fig. 1 shows an example of the VLI output for the district of
Middlesbrough, England. OAs with an index value close to 100 are
considered to be average for the area. The higher the VLI value, the
more vulnerable the locality. In this example the VLI helped
the Middlesbrough Community Safety Partnership3 (CSP) identify
areas that required strategic and prioritised attention for addres-
sing crime and other community safety issues, and subsequently
has resulted in good reductions in crime and disorder in these
areas, including a narrowing of the difference between these
priority neighbourhoods and the average level of crime across
Middlesbrough. The VLI is used in a similar way (at the time of
print) by over 100 CSPs and twenty police forces in England and
Wales, and has been used since 2011 by the Royal New Zealand
Police to determine the areas for targeting its national neighbour-
hood policing strategy [23].
Merseyside Police is one of the English police forces that use
the VLI, identifying in 2011 29 strategic priority areas (across six
local authority districts) for police and local public partnership
intervention, with these areas representing only 2% of the land
mass of Merseyside [21]. Merseyside Police have also qualified the
value of using the VLI by identifying the areas it selected for
prioritised attention were places that experienced many other
issues. For example, in the 29 priority areas (in comparison to the
Merseyside average) they found there to be seven times the levels
of serious violence, three times the level of knife crime, 12% of all
gun crime offenders, 9% of all robbery, burglary and vehicle crime
offenders, higher levels of child obesity, and poorer life expectancy
[21]. Greater Manchester Police and its local partners, including
Greater Manchester Fire and Rescue Service have also made use of
the VLI, and in particular have illustrated its potential for
3
Community Safety Partnerships were formalised in England and Wales in
1998 under the Crime and Disorder Act. These Partnerships are made up of
representatives from the police, local council, and the fire, probation and health
services. CSPs in England and Wales are organised around local district, and county
jurisdictional areas. The Partnership works to develop and implement strategies for
reducing crime and improving community safety. They also work with local
community and voluntary groups. In 2012 there were 310 CSPs in England and
22 in Wales.
specifically assisting in the targeting of fire prevention work by
establishing that a third of all recorded incidents of arson took
place in the 20 most vulnerable wards of Greater Manchester [9].
3. Research methodology
Fire incident data was provided by Northumberland Fire and
Rescue Service for the Newcastle-upon-Tyne district in North East
England. Newcastle has a population of 278,000. Fire incident data
was provided for the period October, 2010–September, 2011. The
incident data were split for analysis into the categories shown in
Table 1. Deliberate fires can also be defined as acts of arson.
Data on burglary to a dwelling and criminal damage to a
dwelling for the period October, 2010–September 2011 were
provided by Northumberland Police. These data were aggregated
as crime counts to the Output Areas for Newcastle (OA n ¼888)
using ESRI ArcGIS version 10. These crime counts were then
exported from ArcGIS and imported into the VLI Excel template
[28]. Data for each OA on the number of households, the resident
population, the population of young people between the ages of
15–24, and educational attainment were sourced from the 2001
UK Census from the ONS Neighbourhood Statistics website (http://
www.neighbourhood.statistics.gov.uk) and loaded into the VLI
Excel template. Data on income and employment deprivation
were also sourced from the ONS Neighbourhood Statistics website
but were first imported into ArcGIS where deprivation scores for
SOAs were matched to their respective OAs. The two deprivation
variables were then imported from ArcGIS into the VLI Excel
template. The two VLI crime variables were converted to rates,
educational attainment and young people population were con-
verted to proportions, and each of the six VLI variables were then
standardised into an index form from which a VLI value for each
OA was calculated following the methodology described by
Chainey [4,3].
A chi-square goodness-of-fit test was used to compare whether
the distribution of fire incidents was related to differences in VLI
values. The following null hypotheses were tested:

The distribution of deliberate fires is not related to differences
in VLI scores

The distribution of malicious false alarms is not related to
differences in VLI scores

The distribution of accidental dwelling fires is not related to
differences in VLI scores
That is, each category of fire incidents was tested in turn in
order to identify any differences in relationships.
The VLI results were imported into ArcGIS and mapped
thematically using two methods:

VLI thematic classes: OAs were grouped into one of five classes,
following the thematic class groupings as advised by Chainey
[3]. Group 1 contained OAs with VLI values between 0–79,
group 2—VLI values between 80–119; group 3—VLI values
between 120–159; group 4—VLI values between 160–199; and
group 5—VLI value greater than or equal to 200.

VLI decile classes: Each decile class contained 88–89 OAs. The
OAs with the lowest VLI values were grouped into decile 1,
through to decile 10 which contained the OAs with the highest
VLI values.
These two methods were used in order to draw conclusions for
the standard thematic classing of VLI values that practitioners use
and to compare it to a more standard statistical method of
grouping categories for hypothesis testing. For each grouping
 S. Chainey / Fire Safety Journal 62 (2013) 30–36
Fig. 1. Priority neighbourhoods in Middlesbrough calculated using the VLI. An area with an index value of 100 is similar to the district average. The higher the index value,
the more vulnerable the locality
Source: [24]
Table 1
Fire incident record types and the number of incidents for the district of Newcastle-
upon-Tyne (October, 2010–September, 2011).
Fire incident record type
Deliberate fires
Malicious/fake false alarm call to the fire service
Accidental dwelling fire
method, and for each class, ArcGIS was used to count the number
of fire incidents for each incident type. The chi-squared analysis
was performed to determine if the observed frequencies (i.e. the
volume of incidents for each VLI class) were significantly different
to those that were expected.
4. Results
Fig. 2 shows the VLI results for Newcastle. Fig. 2a shows
differences in VLI values using the five standard thematic classes,
and Fig. 2b shows the VLI differences using decile classes. VLI
values ranged between 16.29 and 284.22, with the two maps in
Fig. 2 showing a number of areas that could be determined as
priority areas of interest (based on high VLI scores) for targeting
strategic policing and crime prevention initiatives. An inspection
of these data did not reveal any spatial outliers.
Table 2 shows the results of the chi-square analysis for the
observed number of fire incidents in relation to VLI values grouped
33
into the standard thematic classes. Only 12 out of 888 OAs were in
the top thematic class of VLI values of 200 or greater. The
chi-square statistic (x²) for deliberate fire incidents was significant
(x²¼ 433.8; df ¼4; p ¼0.001). Further inspection showed that for
N
the groups of OAs in the top three VLI thematic classes (VLI
649 120–159; 160–199; ≥200), the difference between the observed
86 number of deliberate fires and the expected number was signifi-
85 cantly higher than expected (p ¼0.001). The number of deliberate
fires in the lowest VLI thematic class (values between 0 and 79)
was significantly lower than expected (p4 0.01). We can therefore
reject the null hypothesis that the distribution of deliberate fires
was not related to differences in VLI scores.
The chi-square statistic (x²) for malicious false alarms was
significant (x² ¼65.5; p ¼0.001). Malicious calls to the fire service
were significantly higher than expected for the top two VLI
thematic classes (p¼ 0.001). Thus the null hypothesis that the
distribution of malicious false alarms was not related to differ-
ences in VLI scores is also rejected.
The chi-square statistic for accidental dwelling fires was not
significant (x²¼ 3.8). This is shown by further inspection of the
table showing there to be no statistical difference between the
number of incidents that were expected and the number of
accidental dwelling fire incidents observed across the range of
VLI thematic class values.
Table 3 shows the results of the chi-square analysis for the
observed number of fire incidents in each VLI decile class. For
deliberate fires, the differences between the observed number of
deliberate fire incidents and the expected number across the VLI
decile classes was significant (x²¼ 475.0; p ¼0.001). This analysis
34 S. Chainey / Fire Safety Journal 62 (2013) 30–36

Fig. 2. Vulnerable Localities Index maps for Newcastle-upon-Tyne (a) using standard thematic classes as advised by [3] and (b) decile classes.

revealed that for deciles 9 and 10, the number of deliberate fires
were more than expected (p¼ 0.01 for decile 9; p ¼0.001 for decile
10), and were lower than expected for deciles 1, 2, 3 and 4
(p ¼0.001 for deciles 1, 2 and 3; p ¼0.01 for decile 4). This further
supports the previous analysis, rejecting the null hypothesis that
the distribution of deliberate fires is not related to differences in
VLI scores.
Similarly, there was a significant difference between the
number of malicious false alarms that were observed in compar-
ison to what were expected across the range of VLI decile classes
(x²¼ 42.6; p ¼0.001). Malicious false alarm calls to the fire service
were seen to be significantly higher than expected in VLI decile 10
areas (p ¼0.001). In all other VLI decile classes, malicious false
alarms were not significantly different to what was expected. That
is, areas with the highest VLI values were seen to be the areas
where there was a significantly higher level of malicious false
alarm calls to the fire service. The difference between the observed
number of accidental dwelling fire incidents and the expected
number was not significant (x² ¼5.9). These two results for mal-
icious fires and accidental fires further support the conclusions
from testing their respective null hypotheses against the VLI
thematic classes.
 S. Chainey / Fire Safety Journal 62 (2013) 30–36 35

Table 2
Chi-square results for the observed number of fire incidents in Newcastle-upon-Tyne for VLI thematic classes (as advised by [3]). Results in bold indicate those values that
were significantly higher than expected; results in italics indicate values that were significantly lower than expected.

Types of fire incidents VLI thematic class

0–79 80–119 120–159 160–199 200+ x²

Output Areas per thematic class 443 263 144 26 12

Deliberate fires 130 210 207 59 43
Expected deliberate fires 323.77 192.22 105.24 19.00 8.77
nnn nnn nnn nnn nnn
(O−E)²/E 115.97 1.65 98.39 84.19 133.60 433.79
Malicious false alarms 23 24 22 10 7
Expected malicious false alarms 42.90 25.47 13.95 2.52 1.16
nnn nnn nnn
(O−E)²/E 9.23 0.08 4.65 22.23 29.32 65.5
Accidental dwelling fires 39 24 16 3 3
Expected accidental dwelling fires 42.40 25.17 13.78 2.49 1.15
(O−E)²/E 0.27 0.05 0.36 0.11 2.98 3.8

n
p: 0.05. Results in bold indicate those values that were significantly higher than expected; results in italics indicate values that were significantly lower than expected.
nn
p: 0.01. Results in bold indicate those values that were significantly higher than expected; results in italics indicate values that were significantly lower than expected.
nnn
p: 0.001. Results in bold indicate those values that were significantly higher than expected; results in italics indicate values that were significantly lower than expected.

Table 3
Chi-square results for the observed number of fire incidents in Newcastle for VLI decile classes.

Types of fire incidents VLI deciles

1 2 3 4 5 6 7 8 9 10 x²

Deliberate fires 19 17 22 28 44 51 85 72 102 209

Expected deliberate fires 64.9 64.9 64.9 64.9 64.9 64.9 64.9 64.9 64.9 64.9
nnn nnn nnn n n nnn nnn
(O−E)²/E 32.46 35.35 28.36 20.98 6.73 2.98 6.23 0.78 21.21 319.95 475.0
Malicious false alarms 4 5 3 3 6 5 12 13 12 23
Expected malicious false alarms 8.6 8.6 8.6 8.6 8.6 8.6 8.6 8.6 8.6 8.6
nn nnn
(O−E)²/E 2.46 1.51 3.65 3.65 0.79 1.51 1.34 2.25 1.34 24.11 42.6
Accidental dwelling fires 5 5 9 10 6 9 10 10 12 9
Expected accidental dwelling fires 9 9 9 9 9 9 9 9 9 9
(O−E)²/E 1.44 1.44 0.029 0.27 0.74 0.03 0.27 0.27 1.44 0.03 5.9

n
p: 0.05. Results in bold indicate those values that were significantly higher than expected; results in italics indicate values that were significantly lower than expected.
nn
p: 0.01. Results in bold indicate those values that were significantly higher than expected; results in italics indicate values that were significantly lower than expected.
nnn
p: 0.001. Results in bold indicate those values that were significantly higher than expected; results in italics indicate values that were significantly lower than expected.

5. Discussion a highly vulnerable locality or an area that is not considered to be

The Vulnerable Localities Index is a simple, yet effective
method used by many police agencies in England and Wales to
help identify residential areas for prioritised attention, and
although its origins follow the national UK agenda of improving
community cohesion following a series of riots in 2001, its use is
now common for supporting the strategic targeting of neighbour-
hood policing and crime prevention resources. To date, its primary
relevance has been towards tackling crime issues, and although
many of the UK's Community Safety Partnerships (that include
local fire and rescue services) have extensively used the VLI for
considering wider issues of public safety, its relevance as a tool to
specifically support fire safety has not been tested with any rigour.
This research has shown, using the example of Newcastle-
upon-Tyne, that those areas with high VLI values are places where
there is a significantly higher than expected volume of deliberate
fires and malicious false alarm calls to the fire service. That is,
using a mix of crime, deprivation and socio-demographic data,
areas that are identified as being vulnerable localities and should
therefore require prioritised attention from police agencies coin-
cide with many of the areas where there are issues with deliberate
fire setting and malicious false alarm calls to the fire service: the
police and fire services are responding to issues in the same places.
This suggests there to be potential opportunities for strategic
collaboration between these agencies. In contrast, the volume of
accidental dwelling fires was spread more evenly across the range
of VLI values. This suggests that regardless of whether one lives in
vulnerable (as measured by the VLI), there is no difference in risk
of being a victim of an accidental dwelling fire. These results do
though only refer to one area – Newcastle-upon-Tyne – and
although we encourage further research that explores the relation-
ship between the distribution of fire incidents and VLI scores, we
believe Newcastle offers a representative indication of the poten-
tial we have identified. Further research should also consider the
use of more robust statistical methods that test for spatial
relationships e.g. Poisson regression.
Many practitioners have suggested that a fire-specific VLI could be
developed. This would involve replacing the crime variables the VLI
uses with fire incident data (for example, replacing the burglary and
criminal damage data with refuse fires and malicious false alarm
calls). While this argument does have some merit, there is some
reluctance to do this. The VLI is already a popular analytical tool in
England and Wales that a range of local partner agencies use to help
identify priority areas for mutually beneficial collaboration [3,1,21],
with the combination of deprivation and socio-demographic data,
alongside crime data helping to lessen a purely police-centric,
singular crime dimension approach to strategic partnership focusing
and resource targeting. The creation of a fire-specific VLI, while of
benefit perhaps to a single fire agency, could undermine the
collaborative partnership focus that the current VLI helps to activate
and inhibit opportunities for strategic partnership working. That is, a
fire-specific VLI could be viewed by local partners as a tool for
identifying priority neighbourhoods that are of value only to the fire
service, and has little value for their strategic resource targeting.
36 S. Chainey / Fire Safety Journal 62 (2013) 30–36
It could also be argued that at least one fire variable should be
included in the VLI. This research has shown that the existing VLI
is already effective in identifying areas that are of potential
strategic interest to the fire service, therefore questioning the
addition of another variable to a process that already works, and
which may also prompt requests from other agencies for data on
health, anti-social behaviour, ambulance calls for service and
traffic accidents to be included. In addition, deliberate fires and
malicious calls appear from our study to be highly correlated to
the existing variables used in the VLI, therefore the addition of
these data would offer very little in addition. This in part was
illustrated by Bullen [1] in an attempt to combine additional
variables into the standard VLI, by creating an ‘alternative’ and
more data-heavy VLI. The results identified priority areas that
were no different to the areas the standard VLI identified. One of
the strengths of the VLI is its simplicity, using six variables, rather
than seven, 10 or 20 variables to identify neighbourhoods for
prioritised attention, and the addition of more data adds complex-
ity to the process, without necessarily adding value. The VLI is
already in common use in England and Wales, with many agencies
updating the content of the crime data it uses annually in order to
identify and monitor changes in the priority areas. The inclusion of
other variables such as incidents of deliberate fires would there-
fore not permit like-for-like comparisons over time. Any changes
to the standard VLI would also be difficult to communicate to all
agencies in England and Wales, therefore break the current
consistency in the VLI methodology.
It is also worth reflecting that the VLI is just one tool that offers
a different analytical dimension to hotspot analysis and other
spatial analysis techniques, rather than replacing these other tools.
Indeed, the VLI is purely designed to help identify areas that
require attention, from which further analysis should then follow,
helping to confirm and explore in more detail the reasons why the
areas it identifies require this prioritised focus, and opportunities
for local agency collaboration. Previous research has shown the
validity of the identification of the priority areas in relation to a
wide range of other socio-economic factors, with this study
indicating these areas also to be relevant for fire safety and fire
prevention resource targeting.
6. Conclusions
The VLI is a popular analytical tool used by police agencies to
identify residential neighbourhoods that require prioritised atten-
tion. Its use has also helped to prompt other local public agencies
working to improve community safety to recognise the localities
the VLI identifies are areas that of strategic interest to them, and
the opportunities that therefore exist for partnership collabora-
tion. This research has suggested there is value in using the VLI for
identifying areas of interest to the fire service, illustrating that
areas with high VLI values are also areas where there is a
significantly higher than expected volume of deliberate fires and
malicious false alarm incidents. This provides the potential oppor-
tunity for fire services to not only use the VLI for assisting in their
own fire safety resource targeting, but through the identification
of areas that are also of common interest to other local partner
agencies, helps to prompt opportunities for improving fire safety
in collaboration with the activities of other local partners.
References
[1] I. Bullen, Priority neighbourhoods and the vulnerable localities index in Wigan
—a strategic partnership approach to crime reduction, in: S.P. Chainey,
L. Tompson (Eds.), Crime Mapping Case Studies: Practice and Research, Wiley,
London, 2008.
[2] E.W. Burgess, The growth of the city: an introduction to a research project, in:
R.E. Park, E.W. Burgess, R.D. McKenzie (Eds.), The City, The University of
Chicago Press, Chicago, 1925, pp. 47–62.
[3] S.P. Chainey, Identifying priority neighbourhoods using the Vulnerable Local-
ities Index, Policing: J. Policy Pract. 2 (2) (2008) 196–209.
[4] S.P. Chainey, Identifying priority neighbourhoods using the Vulnerable Local-
ities Index, in: Class Presented at the UK National Crime Mapping Conference,
Manchester, England, 2010, 〈http://www.ucl.ac.uk/jdi/events/mapping-conf/
conf-2010/conf-pres2010/pres16〉.
[5] S.P. Chainey, L. Tompson, S. Uhlig, The utility of hotspot mapping for predicting
spatial patterns of crime, Secur. J. 21 (2008) 4–28.
[6] J.S. Coleman, T. Hoffer, Public and Private High Schools: The Impact of
Communities, Basic Books, New York, 1987.
[7] J. Eck, S.P. Chainey, J. Cameron, R. Wilson, Mapping Crime: Understanding
Hotspots, National Institute of Justice, Washington DC, 2005.
[8] H. Goldstein, Problem-Oriented Policing, McGraw-Hill Inc., New York, 1990.
[9] GMAC, Greater Manchester Strategic Threat Assessment, Manchester: Greater
Manchester Against Crime, 2005.
[10] Home Office, Building Cohesive Communities: A report of the Ministerial
Group on Public Order and Community Cohesion, Home Office, London, 2001.
[11] Home Office, Community Cohesion: A report of the Independent Review Team
chaired by Ted Cantle, Home Office, London, 2001.
[12] Home Office (2003). Building a Picture of Community Cohesion. Home Office
Community Cohesion Unit, London.
[13] Home Office, Information Sharing for Community Safety: Guidance and
Practice Advice, Home Office, London, 2010.
[14] Local Government Association, Guidance on community cohesion, LGA Pub-
lications, London, 2002.
[15] K. Marjoribanks, Y. Kwok, Family capital and Hong Kong adolescents' academic
achievement, Psychol. Rep. 83 (1998) 99–105.
[16] J. Morgan, Safer Communities: The local Delivery of crime Prevention Through
the Partnership Approach, Home Office, London, 1991.
[17] A. Portes, Social Capital: its origins and applications in modern sociology,
Annu. Rev. Sociol. 24 (1998) 1–24.
[18] R.D. Putnam, Bowling Alone: The Collapse and Revival of American Commu-
nity, Simon and Schuster, New York, 2000.
[19] J.H. Ratcliffe, Intelligence-Led Policing, Willan Publishing, Cullompton, 2008.
[20] J.H. Ratcliffe, T. Taniguchi, E. Groff, J. Wood, The philadelphia foot patrol
experiment: a randomized controlled trial of police patrol effectiveness in
violent crime hotspots, Criminology 49 (3) (2011) 795–831.
[21] R. Reese-Smith, S. Kirby, Exploring the effectiveness of the Vulnerable
Localities Index in the context of multi-agency community safety initiatives,
Policing: J. Policy Pract. (2013).
[22] The Riots Communities and Victims Panel, After the Riots, London, The Riots
Communities and Victims Panel, 2012.
[23] RNZP, Annual Report for the year ended 30 June 2010, Royal New Zealand
Police, Wellington, 2011.
[24] Safer Middlesbrough Partnership, Safer Middlesbrough Partnership Strategic
Assessment, AIM Unit, Middlesbrough, 2007.
[25] R.J. Sampson, Great American City: Chicago and the Enduring Neighborhood
Effect, University of Chicago Press, Chicago, 2012.
[26] C.R. Shaw, H.D. McKay, Juvenile Delinquency and Urban Areas, The University
of Chicago Press, Chicago, 1942.
[27] J.D. Teachman, K. Paasch, K. Carver, Social capital and dropping out of school
early, J. Marriage Fam. 58 (1996) 773–783.
[28] UCL DSCS, VLI Excel template, 〈http://www.ucl.ac.uk/scs/research-consul
tancy/geographical-analysis/VLI〉, 2012 (accessed 01.08.2012).
[29] J. Young, Crime and Social Exclusion, in: M. Maguire, R. Morgan, R. Reiner
(Eds.), The Oxford Handbook of Criminology, 3rd edition,Oxford University
Press, Oxford, 2002.