The measurement and recording of alcohol-related violence and disorder
A report of research commissioned by the Portman Group
As we emphasised at the start of this report, there clearly exists an association between the consumption of alcohol and violent and disorderly behaviour, particularly in town and city centres on Friday and Saturday nights. The evidence is before one´s eyes. It does not require social scientists to demonstrate the existence of what, in our society, is the modern manifestation of a timeless and enduring problem.
It is, however, clearly the case that the large majority of people who go out drinking in the many pubs, bars and clubs that now constitute the night-time economy do not get into fights; they do not engage in rowdy and anti-social behaviour; they do not commit acts of vandalism and criminal damage. Instead, they enjoy themselves without causing offence to others.
We have avoided detailed discussion of the theoretical perspectives concerning the nature of the relationship between alcohol and violent behaviour. These issues are dealt with in depth elsewhere both by SIRC/MCM Research and by many others. Our concern in this relatively modest 6 month programme of research has been first, with the size of the relationship between alcohol and violence, rather than the complex elements which create the connection and with how it can best be measured. Secondly, our concern has been with ways in which initiatives taken to attenuate the relationship and ameliorate the problems can accurately be measured and tested in order that the most effective strategies can be identified. In short, we can see that there is a problem. How big is it? What can be done about it?
Sections 2 and 3 of this report have clearly shown, on the basis of both qualitative and quantitative evidence, that we are still some distance from being able to answer either of these questions with any degree of certainty. We simply do not have data of sufficient quality, even at local levels, to do more than hazard a few informed guesses. Since the true scale of the problem is ill-determined, the true impact of crime reduction strategies in this area remains closed to objective scrutiny since changes in the scale of alcohol-related offending are similarly ill-defined. If our society is serious about its determination to reduce problems of drunken violence and if it is to commit very substantial financial budgets to such an end, we will need to make decisions less on the basis of the speculation which currently surrounds the phenomenon and much more on the basis of reliable, objective evidence and unequivocal measures.
We have identified in this report quite clearly the lack of consistency and robustness in much of what currently passes for data on alcohol-related crime and disorder. Police figures, which are perhaps the most commonly quoted sources of supposed evidence in this area, are rarely without both conceptual and methodological flaws. We have noted, of course, in Section 3.1.1 that over one quarter of local police forces do not maintain any records of alcohol-related offences at all and, of these, nearly 70% have no intention of doing so in the future. Of those forces that do collate information of this type, it is difficult to find any two that do so in the same, consistent manner measures of alcohol consumption vary substantially, along with both recording and collating practices. In summary, the following limitations of police data are evident:
1. The measure of an offender’s level of alcohol consumption is often not determined. The judgement that alcohol is ‘related’ to the offence might be based on the fact that he or she seems drunk or partially intoxicated, reports having consumed alcohol, was in or near a licensed premises at the time of the offence or was involved in an incident in which others were intoxicated.
2. The form of recording such consumption is often unsystematic, even within a single local police force. Some data are based on reports to incident rooms, while others relate to offenders who have been charged with a crime. The recording of the presence of alcohol might be in the form of a simple tick-box, a written note, a reference on a charge sheet or an item on a custody record.
3. The collation of alcohol-related data, in whatever form they are recorded, is often not undertaken. Some computer systems permit the entry of alcohol-related data but have no analysis tools. Only 30% of police forces present their data in reports. Where the data are collated, the responsibility for the procedure is often not assigned to a single individual, leading to further sources of inconsistency.
4. The manner in which the data are recorded and collated varies to such an extent between police forces that only 10% are able to compare directly their figures with those obtained elsewhere.
5. Collated data often do not distinguish between offenders and victims. The number of alcohol-related offences includes both those who have committed violence and disorderly acts while intoxicated and those who, while they were the target of a drunk assailant, might have been unaffected by drink themselves.
6. Data relating to reports of offences in the vicinity of licensed premises are routinely classified as ‘alcohol-related’ even though there may be no direct evidence regarding consumption of alcohol by those involved.
Accident and emergency data
1. The occasions on which alcohol-related data are recorded vary from ‘routinely’ to ‘only when required for clinical assessment’.
2. Assessments of alcohol consumption vary from breath testing to questionnaires and physical inspection.
3. The manner in which the data are recorded varies from a prescribed form entry to clinical notes.
4. The retrieval and collation of alcohol-related data is often hampered by both lack of resources and outdated computer systems.
5. Comparability across A&E departments is severely restricted by variations in methods and recording practices.
6. Integration of A&E data with that obtained by the police is strongly hampered by incompatible databases and data protection concerns.
7. Where data are shared it is often difficult to identify overlapping cases - i.e. those which have been recorded by both the police and the A&E department.
8. A&E data relate primarily to victims of assault and wounding, rather than to alcohol-related offending.
Improving data collection methods
The limitations of current data sources noted above are recognised by both police offices and A&E consultants themselves. Most are of the view that more accurate and reliable measures are required, but point to the lack of resources and expertise with which to achieve such improvements. They also point to the lack of guidance and direction from the relevant government departments on this issue. Where there is no formal requirement to collect reliable alcohol-related data in the first place and where there is no set of nationally agreed, formal procedures for doing so, the current pattern of inconsistency and inaccuracy is likely to remain.
The current situation, however, is unlikely to be remedied simply by the imposition on police forces and A&E departments of untried and potentially problematic procedures. While there are examples of extant practices which appear to be substantially better, in terms of rationale and consistency, than others, there is no single, clear best practice´ model which emerges from our research. We therefore make the following recommendation.
One or more model recording and reporting procedures should be developed for both police and A&E departments that seek to overcome the limitations noted above. They should demonstrate objective assessment, systematic recording and ease of retrieval and analysis. They should also enable meaningful and ethical integration of data from the two sources. In this context we recommend that attention be paid to the work conducted by Project Jupiter (see Section 184.108.40.206).
Such models should be tested in pilot schemes in a number of regions of the UK with the help of guidance from external specialists and appropriate resources. Clear benchmarks for evaluating the viability, reliability and validity of the models should be established from the outset.
Revised procedures based on the results of experimental trials, should, after detailed consultation with relevant stakeholders, be considered by the relevant government departments for implementation nationwide.
Crime reduction partnerships
We have noted in Sections 2.3 and 3.3 the difficulties faced by crime reduction partnerships (CRPs) in evaluating their contributions to reducing alcohol-related violence and disorder. In the absence of reliable measures of alcohol-related crime, which might allow quite modest changes and variations to be detected, it is difficult to see how this aspect of their work can be assessed with any degree of empirical rigour.
The potential benefit of the many CRPs that have been established all over the country is the identification of effective best practices perhaps quite simple initiatives which appear to yield good results for relatively small financial and resource investment. Some of these are already evident without the need for sophisticated evaluation tools. In many areas, however, identification of such successes´ is severely restrained both by a lack of reliable data and by inadequate programme evaluation procedures within the partnerships. We have noted in Section 3.4.5 and elsewhere that nearly one quarter of all partnerships appear to have no formal documents outlining their aims and objectives and only 19% rely on statistical data in their internal evaluations.
Very clear models for the evaluation of activities such as those routinely initiated by CRPs are currently available. The work of Michael Hough and Nick Tilley14 at the Home Office, for example, while concerned primarily with the evaluation of police crime prevention activity, contains both a cogent critique of the limitations of current evaluation methods and clear direction on what good procedures should look like. The work of Ray Pawson and Nick Tilley15 is an even more comprehensive guide to the application of sound social science in evaluation procedures.
Given the quality of this material we do not suggest that we should try to reinvent the wheel´ in this context by examining further how evaluation of CRPs might be made more objective. Rather, we make the following recommendation:
In parallel with the actions proposed in recommendation 1 and in the same geographical areas, evaluation models should be developed with regard to the principles identified by Tilley, Hough, Pawson and others and negotiated with the major stakeholders involved in partnership schemes which have, as part of their aim, the reduction of alcohol-related crime.
Appropriate professional expertise and resources should be provided for this purpose and it would desirable for an external evaluator to attend partnership meetings.
Sources of data deriving from recommendation 1 might profitably be used as dependent measures in the evaluations.
A ‘meta-evaluation’ - i.e. an evaluation of the practicality and ease of implementation of the evaluation procedures themselves - should be conducted after an appropriate time interval.
The results of these activities should be analysed in order to identify model evaluation procedures for implementation, where appropriate and with regard to local concerns, at a national level. Inclusion of such models, in addition to the guidance already provided, in the Crime Reduction Toolkits might be a suitable medium for dissemination.
It is also our view that central funding of CRPs should be dependent upon the existence of clear, objective evaluation procedures.
Alcohol and violence - cause and effect
As we have noted earlier (see, for example, Section 2.2.5 and Section 3.1.2) the term alcohol-related´ is imprecise. Its use varies from the simple observation that an offender or patient suffering from an injury had previously consumed alcohol to the belief that the consumption of alcohol had directly caused the person to commit the offence or sustain the injury. A fundamental problem with existing data and, indeed, with data that might be provided by improved recording procedures is that it is impossible to identify the level of causality that can be attributed to alcohol. This is because the necessary comparison and control samples are almost entirely unavailable.
We know that of people drinking on Friday and Saturday nights in town centres, even those consuming excessive amounts´, only a small proportion commit acts of violence. In addition, a substantial proportion of those who do commit such offences are unaffected by alcohol. The question, therefore, is to what extent are people who have consumed alcohol over-represented in the category of violent offenders or injury victims? To answer this and subsequently establish the causal role that alcohol might play, we need information about the levels of consumption of non-offenders in the same contexts.
One of the few researchers to address this issue in the past has been Professor Shepherd in Bristol and then only in the context of victims of assault. Shepherd concluded:
This finding indicates that young male assault victims may not be distinguishable from the young male population at large solely on the basis of alcohol consumption, and therefore suggest(s) that other risk factors are involved (Shepherd, J.P., 1989b:1050).
While some of the methods used by Shepherd, including the choice of his comparison sample, might be seen as having some weaknesses, he is clearly correct in identifying the need for such comparisons if the level of risk presented by alcohol is ever to be discovered.
In the real world it is difficult to find very many people in town and city centres late on weekend evenings who have not been drinking. That is, primarily, why most of them have gone there. The fact that a large proportion of those arrested for violent or disorderly offences at these times have consumed alcohol is not, therefore, very surprising. Nor is it too surprising to find that the number of violent incidents increases at these times compared with other days of the week. Given the significantly expanded population of people on Friday and Saturday nights, a majority in the age and socio-economic categories most likely to commit offences whether drunk or sober, an increase of some level would be expected. In this context, therefore, we make the following recommendation:
Research should be conducted which directly aims to measure the degree to which alcohol is a risk factor in violent crime and injury. A pilot project should initially be developed in a major urban setting that compares the consumption patterns and levels of known offenders with those of a non-offending comparison sample. Similar comparisons should be made with regard to injuries.
BAC levels should be established for an anonymised sample of offenders apprehended for relevant offences by the police on Friday and Saturday nights. The same measures should be obtained from a random and equally anonymised sample of non-offending visitors during the same times. This might be achieved using disposable alcometers which could be dropped by testees into a sealed box. Additional questionnaire data might also be obtained from both samples. Estimates of the total number of visitors in the potential risk category to the town centre at various times of the week would be the final data element required.
Data from such a pilot study would immediately provide a much better indication of the true role played by alcohol in offending and the degree to which it is a risk factor. It would, probably for the first time, allow an estimation of the true proportion of crimes and disturbances which would not have occurred but for the consumption of alcohol.
The study would also identify the age and other characteristics of those most likely to be influenced by alcohol in this way. (See the studies by Holt and STAG noted in Section 2.2.11 which suggest that it is men in the age category 40-49 years who were over-represented in attendances for alcohol-related injuries.) By enabling ‘at-risk’ groups to be more clearly defined it would have direct implications for the targeting of initiatives aimed at reducing alcohol- related problems.
The uses and abuses of data
In the way that our tabloid newspapers routinely fill our lives with the bad news rather than the good, so too there is a tendency for statistics to be made public only when they show that things are getting worse, or when they are employed in the pursuit of political and other agendas. This appears to be particularly the case when we encounter published data concerned with alcohol-related violence and disorder. On occasions we have found, for example, that local police data have only been collated and tabulated when there was a need to support their objection to a proposed new licensed premises. Elsewhere, as we noted in Section 2.1.7, it is a determination to crack down´ on certain pubs and clubs, or to make a case for more police resources that often motivates the data- publishing process.
In order to preserve the perceived neutrality´ of data on alcohol-related crime we therefore make the following recommendation.
All data on alcohol-related violence and disorder, whether derived from existing or improved procedures, should be collated and published at fixed intervals in consistent formats. Such data should be available to all relevant stakeholders on request as a matter of course.
The role of the drinks and entertainment industry
Few people would argue with the suggestion that the quality of management and style of operation of pubs, bars, clubs and other venues has a significant impact on levels of violence and disorder, both on the premises and subsequently in the immediate vicinity. With increasing competition in the night-time economy there are clearly occasions when the professionalism of this area of the retail trade slips´ a little. Irresponsible promotions and discounting, which lead to excessive or acute patterns of alcohol consumption and the serving of underage customers are just two of the many indicators of such management failings.
From our discussions with trade bodies and organisations it is clear that the industry is very aware of the need to maintain high levels of professionalism and integrity among those at its retail front line and the activities of the BII and BLRA are of particular significance in this context. So too are programmes developed by the Portman Group, such as those which seek to reduce underage drinking and the educational campaigns targeted at secondary school students.
There is, however, an indication that the role of the alcohol industry at a local level is not sufficiently defined (see Section 2.6.4). We make no specific recommendations in this context, but clearly the increased involvement of representatives of the major trade bodies in local crime reduction and prevention initiatives which have a particular focus on alcohol-related issues is both desirable and expedient.