Building an evidence base on alcohol supply points

Note: This report was funded and/or written by our predecessor organisation, the Alcohol Education Research Council (AERC).

This report presents the findings of a twelve month study conducted by the Applied Criminology Centre, University of Huddersfield. This pilot project sought to generate intelligence for managing areas with licensed premises by building an evidence base on alcohol supply points (ASPs). For the purposes of this project, ASPs are considered to be any licensed premises that can sell alcohol for consumption on and/or off the premise, for example public houses, bars, nightclubs, restaurants, corner shops, off licenses, hotels, cinemas, and social clubs.

Despite substantial efforts towards multi-partnership working, data collection and intelligence sharing to tackle alcohol-related crime and disorder in areas with ASPs, the capture of data on ASPs is fragmented. Intelligence gathering often occurs in isolation except for sporadic multi-agency enforcement visits to premises. This impairs attempts to gain a strategic overview of the timing and location of the availability of alcohol, the proximity of the various outlets to each other, and their relationship to crime and disorder. This evidence base is a necessary foundation for making informed decisions about the management of areas with licensed premises, including: the granting and renewal of licensing applications; the development of local crime prevention and harm reduction strategies; and targeting policing and other enforcement activities.

The aim of this project was to address this deficiency by piloting the creation of an ASPs database in three case study areas, and to explore possible relationships between alcohol supply and crime and disorder in local contexts. There was an identified need for consistent and reliable data on ASPs that could be shared between relevant organisations, and for local areas to develop intelligence on alcohol supply tailored to their needs. Any system developed should be simple, user friendly, relevant, and add value to what currently exists.

One of the intentions of this project was to work closely with relevant agencies, experts and relevant data providers. To achieve this a Project User Group (PUG) was established for the duration of this project, comprising those who were considered key stakeholders in efforts to prevent and reduce alcohol related crime and disorder. Their expertise and knowledge was critical in providing advice and steer to the project and in facilitating access to practitioners and data (see acknowledgements at end of this report for details of membership).

This study was carried out in three phases, each of which involved consultation with the PUG. The first phase of this research consisted of a total of 28 semi-structured interviews conducted in the three case study areas with a number of stakeholder organisations, including representatives from police forces, local authorities, Primary Care Trusts (PCTs), hospital Accident and Emergency (A&E) departments, ambulance trusts, trading standards, and other relevant organisations. The purposes of phase one were to ascertain which data sets were collected currently, for what purpose, how these data were stored, analysed and shared, how these data sets were used to inform decision making at the local level, and how this process might be improved through the creation of a single, consistent, multi-purpose database.

The second phase of this research set out the specification and functions of the single database, and identified core datasets, based on the findings of the interviews with key stakeholders, a review of relevant literature, and discussions with the PUG. Once the minimum requirements of a single database were established, data sharing agreements were devised and negotiations were made for access to these data (for the purposes of research) for each of the three case study areas.

The final phase of this research was to test the usability of the pilot databases and to demonstrate the potential added value of such a system for the management of areas with licensed premises. A number of research questions were drawn up in consultation with the PUG in order to examine the benefits and insights that a single database might offer. Due to difficulties in accessing the minimum datasets required in each case study area, only two case study areas were analysed for the final phase of this research. A Geographical Information Systems (GIS) was used to allow the team to match individual ASPs to a number of different spatial units including electoral wards and Output Areas (AOs) from the 2001 Population Census. These were then used to answer a number of the research questions.

This section discusses the findings from each of the three phases of the research:

Phase One

Key findings from the phase one interviews were as follows:

A number of actions were identified to improve the quality of information available to decision-makers at the local level, and the standards of data analysis. These included:

• a need for more comprehensive and consistent data collection;
• increased sharing of data between partner agencies;
• the nature of information sharing was generally ad hoc and relied on individuals, although occasionally at partnership meetings aggregated/analysed information was shared regularly;
• improved standards of analysis to inform policy enforcement, implementation of prevention strategies, deployment of resources, and for monitoring and evaluation, and;
• improved access, suggested by several stakeholders, to information from hospital attendances and ambulances responses.

A number of stakeholders agreed that a single multi-purpose database would be highly beneficial. However, a number of concerns and obstacles to achieving this were identified both organisationally and in terms of resources. The following concerns were identified:

Organisational concerns

• agencies collect data for a variety of reasons other than for the management of areas with licensed premises;
• comprehensive capture of these data would be resource intensive;
• current systems do not easily allow data to be exported,
• there are limitations in current data collection techniques;
• some organisations store their data on more than one system, therefore obtaining information relevant to ASPs from their systems would be far from straightforward and time consuming;
• there are legislative and cultural barriers to sharing individual level data (data protection).

Resource constraints:

• the cost needed to develop such a system;
• the extent and level of training required;
• a lack of time and resources to interrogate the data; and
• the task would be too complicated and might not contribute sufficient added value to justify the effort

Phase Two

Key findings from phase two of the research included the following:

• The key sources of data identified for the construction of a single database/system were licensed premise data, police recorded crime data, trading standards data, A&E data, and ambulance data

A number of functions were identified for the development of a single database, and these can be classified as short-term operational responses, mid to longer-term strategic policy decision making, and research functions. Key functions of the database identified were:

• to administer licensing applications;
• to monitor individual premises, individual persons (both irresponsible managers and repeat offenders), and areas with high concentrations of premises;
• to compile evidence for licensing hearings and reviews;
• to identify, prioritise and carry out targeted enforcement activity;
• to corroborate and share knowledge; and
• to remove duplication of effort.

There were a number of difficulties encountered during the creation of the pilot databases. As stated previously, it was not possible to acquire all data sources for each of the three case study areas as some organisations were unwilling or unable to share disaggregate data (even with personal information removed). Indeed, only two case study areas could be used for the final analysis and the key obstacles faced here were that:

• A&E data and ambulance data could not be acquired for any of the three case study areas during the time frame of this research (reflecting the concerns expressed by practitioners in phase one about the sharing of health data). This is particularly important when considering the known the under reporting of crime to police; and
• In addition to this, several of the datasets required time-consuming manual processing to prepare them for analysis due to the format in which they were currently produced. This stage of the process added considerable time (several weeks) to the creation of the pilot databases.

Phase Three

In order to test the usability of the pilot database a number of research questions were generated in conjunction with the PUG. These included an examination of:

• the spatial relationship between ASPs, trading hours and crime;
• the relationship between ASP density and crime;
• the spatial relationship between ASP density by type and crime;
• the extent to which specific combinations of licensed premises explain the variations in the different types of crime; and
• the extent local enforcement (trading standards) matched concentrations of licensed premises and crime.

Key findings from this analysis were as follows.

Concentrations of ASPs:

• ASPs are spatially concentrated (in one of the wards in the case study areas the density was found to be 7 households per ASP).
• The number of ASPs and levels of crime in these areas of concentrated drinking were disproportionately higher than their share of the residential population.
• Therefore, the residential population (currently used as the denominator to construct crime rates) may not be the most appropriate measure; for example in the case of violent crime, the ‘total number of licensed premises’ or ‘land area in hectares’ might be better denominators to use.

The relationship between ASPs and crime:

• Correlation analyses were used to produce a more systematic examination of the relationships between crime and ASPs.
• In both case study areas, higher numbers of ASPs (taking into account both the densities of ASPs in a ward, and the population rate) were associated with higher crime rates (supporting the findings of previous studies).
• The strongest correlations revealed that higher levels of violent crime were statistically more likely in the areas with higher numbers of ASPs and longer trading hours.
• Indeed, in the two case study areas, the correlation between ASPs and violence against the persons was 0.905 and 0.775.

The relationship between ASP type and crime

• The overall mix of premise types (based on all ASP types in each ward) appeared not to be related to the ward’s crime rate; that is wards with an equal share of ASPs in each category did not register higher crime than elsewhere.
• However, regression analyses were used to explore how far specific combinations of ASP types explained variations in crime rates.
• This analysis suggested that certain combinations of ASPs accounted for a large proportion of the variation in crime rates.
• Pubs, bars and nightclubs were the strongest predictors of variations in crime
• The only other ASP categories to predict variations in crime were ‘Takeaways’ in one case study area and ‘Stores and off-licences’, and ‘Members/social clubs’ in the other.
• Therefore, neither restaurants, supermarkets, nor other types of licensed premise were strong enough to be predictors of crime in either case study area.

A number of policy recommendations were formulated on the basis of these findings. Some of the key suggestions are:

• There is a need for improved intelligence for the management of licensed premises. A preliminary step is to improve current databases and systems used by the organisations.
• Core datasets should include information from licensing authorities, the police, trading standards, accident and emergency departments, and ambulance trusts.
• Some of the key barriers that exist to developing a single multi-purpose system at present are resource issues (time, cost and training), problems with exporting data between proprietary systems (that are not designed to export information in a consistent fashion), cultural barriers to sharing information, concern about data protection, and cultural barriers to data sharing.
• Licensed premise data should be stored electronically in a system that is easily updateable and can easily be queried. Key features to be captured are: unique identifiers for each premise (that can be linked to other organisations datasets); a geocoded consistent address; opening hours and permitted trading hours by time of day and day of week; capacity (for on licensed premises); type of premise (a standardised classification should be devised for this); information on permitted activities, licensing conditions and temporary events.
• Further analysis is required by local areas to better understand the relationships between ASP types, ASP density, trading hours, and levels of crime.