Guidance for creating a violent crime problem profile
analysis – April 2009
1.1 About this guidance
This guidance covers:
; an introduction that discusses the reason a problem profile might be
needed and what makes a problem profile ‘fit for purpose’;
; the overall structure of the problem profile;
; defining and categorising violence;
; reviewing data and accessing data sources;
; analysis of violent crime including understanding locations, offenders and
victims as well as understanding the underlying drivers of violence;
; interpretation of data; and
; a reading list.
This guidance is not intended to be prescriptive or exhaustive. Instead it provides suggestions as to what aspects of the problem should be considered in the analysis, whilst still allowing flexibility in the final content, to reflect the local situation. This guidance may at times appear to be stating the obvious; it does so to ensure that all points are covered.
1.2 What is a problem profile
Problem profiles provide a clear picture of a problem; show evidence of information analysis; make recommendations based on the analysis; and enable managers to justify actions and allocate appropriate resources.
The most important point when developing the problem profile is to keep in mind the uses to which it will be put. The analysis should therefore move from the general to the specific, discarding that which is superfluous to developing responses and focusing on the detail that can help the implementation process.
Analysis which is to be of most benefit to local policy-makers and practitioners may not be immediately obvious to the analyst. This highlights the importance of maintaining a dialogue with local policy-makers and practitioners and suggests that an iterative approach, in which analysis is developed in consultation with policy-makers and practitioners may be as beneficial than the production of the ‘finished article’. One way to do this is to start the process of developing a problem profile with a meeting between analysts and the local policy-makers and practitioners who will use the problem profile. The aims of the meeting are to identify gaps in knowledge and ascertain how much impact gaps in knowledge
have on the development of local policy and practice. Questions to discuss at this meeting would include:
; What is already known with certainty?
; What gaps exist in current knowledge?
; What working hypotheses do local policy-makers and practitioners rely on,
that have yet to be tested?
; What is assumed because of national research and analysis, but has yet
to be tested locally?
; Where gaps in knowledge exist, what would the benefit be to local policy
and practice if the knowledge gap was closed?
Through such discussion the terms of reference for the problem profile can be
developed and these can help shape the structure of the profile and the work that needs to be undertaken.
1.3 How does a problem profile differ from performance management
Data will already be gathered at a local level in relation to violence as part of ongoing performance management arrangements. Of particular relevance will be data gathered in relation to Most Serious Violence (NI15) and Assault with Less Serious Injury (NI20). However, data gathered for performance management processes will be restricted to that required by agreed definitions of the types of violence in question. Performance management will focus on gathering information to assess how areas are progressing towards specific targets or objectives rather than exploring underlying causes of violent crime or types of crime that are not included within the agreed definitions. In contrast a problem profile will collect all relevant information on a violence problem and synthesise this to consider patterns, drivers and possible solutions.
1.4 How does a problem profile differ from ongoing analysis undertaken
as part of NIM?
In order to support routine ‘tasking’ of front-line staff, the police and other
organisations within the CDRP will regularly undertake analysis of a range of crime and anti-social behaviour issues. In the case of the police this analysis will follow National Intelligence Model (NIM) requirements. A problem profile differs from this routine analysis in its scope and depth and also in its focus on longer-term causal factors although NIM analysis may well feed into the overall problem profile.
2 Overall structure of the problem profile
The problem profile should include the following: an introduction, describing the problem and reasons for conducting the analysis; a detailed analysis of the
problem, using the appropriate analytical tools; and finally recommendations for further actions, based on the evidence emerging from the analysis.
This should include:
; Aims and purpose of the problem profile – outline why the problem profile
is being prepared, who the target audiences are, and the terms of reference
for the problem profile.
; Strategic Drivers – state how the problem profile fits with CDRP strategies
and targets and how the interventions that will be implemented as a result of
the problem profile will fit in with this. Specify what measures of success are
being used, such as decreases in violence with injury (victim reported rather
than police reported), and increases in reported DV (with associated
decreases in RV)
; Methodology - specify what data sources have been accessed, over what
time period. The methodology should specify exactly how the data has been
extracted to allow for comparisons with previous problem profiles, or others
prepared in the future.
; Limitations of the report – make sure appropriate caveats are noted that
should be taken into account whilst reading the report. This might include
difficulties with accessing the data, limitations due to incomplete data sets or
missing data, differences between recorded crime and incident data that need
to be taken into account, and the impact this will have on the interpretation of
the data and subsequent recommendations. It should also include information
about anything that could be done differently to overcome these problems
(such as collecting data in a different way, better compliance with
requirements to use special interest markers etc) – these should also be
reiterated in the recommendations for organisational change. Any changes in
counting rules or crime classifications that will affect the results of the analysis
should also be taken into account.
The analysis section should show a clear understanding of the problem using standard analytical products and techniques as appropriate. The analysis should draw out clear inferences from the analysis, in understanding why problems exist in certain places at certain times, identify interactions between different factors (e.g. time, place victim/offender relationship) and identify key pinch points where interventions can be most effective. This is discussed in more detail in Section 5.
2.3 Recommendations for further action
Recommendations resulting from the problem profile and analysis should be based on the evidence presented. This could cover intelligence collection, investigation/enforcement, prevention priorities, recommendations for
reassurance and organisational issues (such as data recording). The actions suggested should relate directly to the analysis and understanding of the problem. 3 Defining and categorising violence
3.1 Why is it important to define and categorise violence
The starting point for any analysis of violence is to understand the types of crime covered within this broad heading. There will be little long term value to be gained in analysing total violent crime. In most cases segmenting violence will aid the process of understanding the nature of violent offence as violence involves different types of behaviour and potentially, different types of motivation. This will be especially important when one begins to drill down further into the data, as the characteristics of particular types of offence are likely to vary. In addition, the segmentation of violence may help in developing appropriate interventions
3.2 Definitions of violence
Any analysis should clearly explain the definitions and Home Office offence codes used to extract the data within the problem profile.
For administrative purposes, such as monitoring the delivery of specific targets, different types of violence have been grouped in different ways based on the classification found in Home Office counting rules (as found on iQuanta), the framework for which is based loosely around the levels of injury involved. These include:
; Most serious violence
; Most serious violence against the person
; Assault with less serious injury
; Assault without injury
; Serious wounding
; Other wounding
; Wounding (serious and other)
; Other violence
; Other violence against the person
Details of the most current National Indicators related to violent crime, NI15 (Most Serious Violence) and NI20 (Assault with less serious injury) can be found in Appendix II.
From an analytical perspective, while administrative classifications may be useful for benchmarking against other areas and for performance management, the key will be to understand the different forms of behaviour that make up violent offending. This may involve analysing the individual groups of offences which share similar behavioural, relationship or contextual characteristics rather than the classifications that are provided on iQuanta. For instance useful categories to
consider might include: domestic violence, sexual violence; violence linked to the night-time economy, racial violence or gang-related violence. However even these can be explored in greater detail; for instance, rape is just one element of sexual violence. The profiles of victims and offenders within rape might be very different to those involved in trafficking for sexual exploitation. If considered together this may mask a pattern that would be apparent were these two categories considered separately.
4 Reviewing data and accessing data sources
4.1 Why we include a range of data sources
Problem profiles should draw on data from a range of sources which are relevant to the understanding of the nature of the specific problem under analysis. This is partly because less than half of violent crimes are reported to the police (Walker et al, 2006); and therefore over-reliance on police data will lead to a partial analysis of the problem. Drawing on data from partner agencies may help to provide a more complete picture of the problem but it is not without its difficulties.
Before you begin the process of analysis it may be helpful to consider the following questions:
Is the category you are considering reliable?
In some cases data may not be consistently collected (i.e. if the data were collected again using the same methods similar results may not be obtained). This will affect the reliability of the data. For instance police data will often carry a marker to indicate whether alcohol was a contributory factor although use of this can often depend on the judgement of the arresting officer.
Is the category sizeable enough for meaningful analysis?
If categories are broken down too far, offence numbers can become too small for meaningful analysis. It is therefore important to consider what affect your chosen categories will have on the numbers of offences being analysed. This is particularly important when considering serious violence categories, as many of the offence categories involve small numbers and breaking these down further can often result in groupings which are too small for analysis.
What can be done to look at the inter-relationship between different factors? While it is important to consider distributions of characteristics (variables) associated with offences (e.g. age of offender) it is often helpful to consider how particular characteristics relate to each other to see if this highlights additional patterns. For instance, considering age and time of offence may highlight a pattern of school children committing offences on certain days of the week which can then be linked to school activity.
4.2 Police incident and crime data
Police offence data are the most readily available violence data that are provided at a local level. They will provide an indication of incidents reported to the police, either by victims or witnesses (e.g. members of the public, or the CCTV control room). This may be particularly useful for estimating the number of violent incidents where injury is not sustained, bearing in mind that there may be more than one report of a single incident. However, police data is likely to under-record some types of incidents, especially domestic violence (see below).
Under-reporting to the police is a particular issue for most crime types (with the exception of homicide). For example, the below table indicates the extent of under-reporting for different crime types based on the British Crime Survey 1(BCS).
In 2004/05 the BCS recorded that 46% of all BCS violent crime was reported to the police. For a table demonstrating the variations on reporting across offence types (including relationship between victim and offender) see Appendix III.
The national picture
For a large number of offences, analysis of the nature and pattern of offending 2has been undertaken by various groups including the Home Office and
academic institutions. While they will not provide a local picture, they may be of
1 More information can be found at http://www.homeoffice.gov.uk/rds/pdfs08/hosb0708.pdf 2 More information on Home Office research can be found at http://www.homeoffice.gov.uk/rds/
use in helping to gain a better understanding of the questions that you would want to ask of your local data and may provide an insight into the likely nature or 3 characteristics of victims and offenders.
4.3 Non police sources of violence data
For reasons outlined above (difficulties of definition, under-reporting), reliance on police data may only provide a partial picture of violence in a locality and it may be sensible to draw on other data sources to provide a more comprehensive picture. Here we are not advocating ignoring police data, but adding to it with other sources of data.
When requesting data from partners it is worth considering what format the data should be provided in. Some raw data will provide too much information or require specialist knowledge to analyse. Conversely pre-analysed data may not allow you to drill down to an appropriate level. Initial discussions with data providers may assist in deciding which format may be most suitable for your specific requirements.
The following paragraphs describe data sources that may be available and some things you may want to take into consideration when requesting additional data.
Ambulance data can provide information on the nature of the incidents attended by ambulance staff. Generally speaking ambulance data will focus on the more serious types of violent incident given that they require some form of medical attention. This source of data can provide details on the location of ambulance ‘pick-ups’ and the time of incidents. It may also provide details of the type of
incident, although this can be quite broad, such as ‘assaults’ and ‘stabbings’.
In some cases data will be collected on reports to the ambulance service from members of the public, as well as the cases themselves. This is an important distinction as the former will significantly inflate the figures. This highlights the importance of understanding the characteristics of the data you are working with. In addition, it is possible that incidents are coded as violence-related (e.g. ‘assaults’) but are not actually the result of a violent incident (e.g. it could be that injury was sustained as a result of walking on broken glass). Ambulance data may also provide alcohol markers, although this will relate to the effect of alcohol on treatment of a patient rather than the cause of an incident.
3 For example see Feist, A. et al. (2007) Investigating and detecting recorded offences of rape. Home Office Online Report 18/07. London: Home Office.
Smith, J. (2003) The nature of personal robbery Home Office Research Study 254. London:
Home Office. http://www.homeoffice.gov.uk/rds/pdfs2/hors254.pdf
Accident and emergency data
Many hospitals now collect some form of data on violent incidents, following the Cardiff model (see http://www.homeoffice.gov.uk/rds/pdfs2/hors265.pdf). This
involves collecting a number of items of information, either at the point of reception, or at the point of treatment.
This source of data may provide information on times and the nature of injuries. It may also provide information on locations of incidents – such as the licensed
premises or street in which an incident occurred.
Domestic violence forums
Some areas will have multiple reporting centres for domestic violence. These provide a mechanism for different partner agencies to record incidents of domestic violence reported to them. These may include cases where the victim is unwilling to visit a police station to report the incident ‘officially’. It is important to note that many areas are making efforts to increase the levels of reporting of domestic violence. This clearly has implications for examining trends in domestic violence as increases in recorded incidents may actually reflect increases in reporting, rather than underlying increases in the actual number of incidents occurring.
Multi Agency Risk Assessment Conference (MARAC) data
MARACs provide an opportunity for local partner agencies to share information on high risk domestic violence cases. The benefit of this source of information is that it draws together intelligence from different sources. This may provide a means of identifying the number of domestic violence victims that are currently viewed by professionals to be at high risk of repeat incidents
Other sources of data on violence
Other sources of data can help to understand problems of violence, particularly related to alcohol and the night time economy. These can include council licensing information, reports of disorder on public transport, reports from community forums (e.g. PACTs), and local authority footfall counts in town centres. Data may also be available through PPO/Probation and Prison Service that assists in further developing an understanding of the nature of offending.
4.4 Intelligence gaps / missing data
The problem profile should include a missing data section. This should list data that was sought or requested but that was not available. It is important that the reason for this data being sought or requested is included. For example, if the data had been available, what analysis would have been performed? Following on from this any weaknesses in the analysis resulting from missing data should be highlighted. Recommendations on how to obtain missing data for future problem profiles should also be included. Such recommendations should be as specific as possible, detailing particular actions that will need to be taken and suggesting which organisation is best placed to take them.
4.5 Linking different data sources
It is important to recognise that drawing on data from different sources is not without its problems. Firstly, different data sources will apply different definitions and collect data for very different purposes. This can mean that each data source will provide only a partial insight into a violence problem. For this reason, each data source should be weighed up in its own right and you should seek to understand the strengths and weaknesses of each data source.
Secondly, there is a very high likelihood of multiple counting of offences when examining different sources. For example a serious injury resulting from a pub fight may be recorded by the police, but it may also be recorded by the ambulance service when they take the injured party off to hospital and by the hospital A&E department on arrival. This argues against simply merging data sets; rather comparing datasets will be of more help in identifying patterns emerging from different data sources and identifying areas that need further analysis or research.
If one does attempt to merge data sets, perhaps in order to get a ‘true’ number of violent incidents, it is important to examine individual case characteristics in order to identify (where possible) potential duplicate cases although, in practice, this may be hard to achieve.
5 Analysis of violent crime
5.1 Analytical methods
When undertaking analysis it is important to decide on the most appropriate analytical methods to use for the question, taking into account the data available, the aims of the problem profile, and the depth of information required. Analysis should ensure that enough information is provided to demonstrate what the problem is and to determine if the interventions being implemented to address violent crime are appropriate for the local problem.
The analysis should look at trends in violent crime, counts and percentage changes over time for specific offence types and, where possible, rates of offences. The denominator for determining rates should reflect the type of violent crime and the location in which it is occurring – using household or population
data for violent crime in town centres will give misleading results, as it will not take account of the influx of people due to the night-time economy. For example, the average monthly footfall count in a town centre can give some indication of the change in the number of people out in the town centre and how this relates to the level of violent crime. For domestic violence, rates of victimisation can be determined using sociodemographic data, to ascertain whether the victim and offender profiles are typical of the local population, or if issues of under-reporting may be identified for particular population groups.
5.2 Understanding locations
The geographical distribution of violent crime should be analysed, to identify the most problematic areas within the CRP. The distribution will depend largely on the nature of the problem: violence associated with the night-time economy will be concentrated in town centre areas; domestic violence may be more widely dispersed, across residential areas. Some things to consider when looking at location are:
; The profile of violent crime may differ in different areas within the CDRP to
reflect the concentration and closing times of pubs and clubs, and the ease of
dispersal of customers at the end of the night. A separate detailed analysis
should therefore be carried out for each area.
; The analysis for violent crime should look at hot areas, hot lines (streets) and
hot dots (licensed premises and also transport terminals, take aways, taxi
ranks), and how these have changed over time. Local knowledge and
additional information should be used in conjunction with this to ascertain why
clusters of crime occur in these areas, and identify possible action points to
reduce violence. For example, linking the information on violence in licensed
premises to capacities, closing times and other factors may help to explain
why more offences are associated with some venues more than others. Hot
spot areas on which more detailed analysis is to be focused should be clearly
defined (e.g. using beat identifiers, bounding streets, etc.) so that resources
can be tracked, a results analysis carried out, and further comparable
analysis can be conducted at a later stage if necessary.
; Geographical hot spot analysis, and GIS techniques are suitable for location
based offences, although not so appropriate for DV. The analysis of DV
should look at the proportion of incidents that occur in private dwellings
compared to public places, and where possible, look in more detail at the
public locations. This may provide information to help target publicity material
at both victims and offenders, although care must be taken to ensure that
areas with low reporting rates are not excluded on this basis.
; Other non-violent crimes may be affected by initiatives to reduce violence,
such as criminal damage. It may be worth considering these in the analysis,
and subsequent evaluation, to identify possible displacement or diffusion of
Temporal analysis should be used to help recognise patterns and changes in offences, to ensure that adequate resources are available for deployment at higher risk times, and that additional opportunities for crime reduction can be identified.