Preventing Repeat Residential Burglary:

A meta-evaluation of two Australian demonstration projects

Contents | Acknowledgements | Executive Summary | Chapter 1: About repeat victimisation | Chapter 2: The demonstration projects | Chapter 3: Meta-Evaluation | Chapter 4: Programme guide and resource tools | Appendix 1: Summary of research study outcomes | Appendix 2: Process evatuation framework | References

CHAPTER 3

Meta-evaluation

This chapter discusses meta-evaluation issues, drawing on both the specific results of the two demonstration projects and on other Australian research and international literature to identify implications for policy and practice. It presents a limited comparative analysis of the outcomes (to the extent possible within the different evaluation methodologies and indicators used), identifies common key success factors, and discusses implications for policy and practice. The term 'break and enter' is used consistently throughout this chapter, even when referring to the outcomes of the one jurisdictional evaluation report that used the word 'burglary'.

Comparison of key programme features

The two demonstration projects delivered some common intervention components, but using a different strategic approach. Both developed a home security assessment process that provided the basis for security advice to residents. Both provided property marking equipment or services and both involved informing near neighbours. The elements were generally delivered in combination, so that it is difficult to analyse the different effects of individual core elements (for example, assessing the benefits of property marking separate from security advice) across the two projects.

The major difference between the two demonstration projects was the overall strategic approach adopted. The South Australian model used volunteers to deliver the intervention. The same package of responses was available to a victimised household, regardless of prior victimisation history at that address. Police were responsible for delivering the intervention in the Queensland project, although volunteers assisted police in one programme element. Identified repeat victims received a more intensive level of intervention. Key features of the two programmes are summarised below.

TABLE 14
Features of the two demonstration projects 

	Queensland	South Australia
Strategic approach	differentiated police response	community-based victim assistance
	graded response (more intensive intervention for identified repeats)	base level response (same intervention response regardless of prior history)
	delivered by police (with some assistance from volunteers)	delivered by community volunteers
Intervention elements	home security assessment	home security audit
	security advice	security advice
	property marking equipment and services provided	property marking referral made by the volunteer
	advice to near neighbours	advice to near neighbours
	loan of portable alarms or similar equipment (where relevant)	provision and installation of locks where relevant (limited to one intervention area)
	additional police patrol activity	informal victim support
	Hot Spot area-level response	referral to victim support services and other community-based services (where appropriate)

The similarity in the types of intervention elements developed and implemented in the two demonstration projects indicates there are commonly acknowledged good practice features likely to be adopted in any break and enter crime prevention programme. These include traditional situational crime prevention measures, such as target hardening victimised residences by identifying and addressing vulnerable features of these dwellings, particularly inadequate locks on windows and doors. They also relate to methods traditionally used to increase risk of offenders being detected, for example by encouraging immediate neighbours to be more vigilant. Processes to reduce rewards to the offender were also introduced, for example, by marking items so that stolen goods are less able to be sold legally and more likely to be recovered.

Comparative outcomes of the demonstration projects

The Queensland and South Australian evaluations selected different measures to analyse change over time and differed in the time period and methodology used to assess repeat victimisation. This limits direct comparability of results between the two projects. It precludes combining data across the two jurisdictions for more detailed analysis than those provided in the individual jurisdictional evaluations. The results of the analyses below need to be interpreted in this context.

Effect on repeat break and enters

Results combined across the two intervention areas in South Australia and the different control areas in each jurisdiction give a simple comparison of the effects of each programme in each jurisdiction. The three-month measure was used for the South Australian data in this analysis, as this includes a larger number of residences in the analysis than the six-month measure. Therefore, it provides more comprehensive coverage.

There was a 16 per cent reduction in the number of repeat victims in the intervention area, compared to an 11 per cent increase in the two control sites in the Queensland demonstration project. In South Australia, there was a two per cent reduction in the combined intervention areas compared to a 25 per cent increase over all control sites.

Although Queensland appears to show a larger absolute reduction, both jurisdictions show a similar magnitude of differences relative to change in the control sites. Given the different methodologies, the more appropriate comparison is the extent of intra-jurisdictional change (that is, the difference between change in each individual jurisdiction's intervention areas relative to control sites) rather than comparing the level of reduction in repeats between the two jurisdictions.

Based on these comparisons, both demonstration projects are associated with modest successes in preventing repeat incidents in the intervention locations, relative to residences in areas where those particular programmes were not operating.

Effect on total break and enters

There was a 19 per cent increase in the total number of break and enter incidents in the Queensland intervention area, compared to a 12 per cent decline in the two control sites. In the South Australian project, there was a 31 per cent increase in the combined intervention areas compared to a 17 per cent increase across all control sites.

Both jurisdictions showed increases in overall break and enter victimisation in the areas that the programmes were operating, in contrast to reductions or lower levels of increase in the control areas.

The project effects on preventing repeat break and enters do not extend to reducing overall break and enter victimisations. In fact, the increased incidence in both jurisdictions runs counter to the expected programme effects. On the basis of the available data, there is no obvious explanation to account for these findings in both jurisdictions. Some possible factors include:

• Target displacement may be operating, that is, the programme's effect on repeat victimisation is displacing offender activity from what are now more difficult targets to more vulnerable dwellings within the area. This is discussed in the section on unintended consequences below.
  
• Given the relatively small numbers involved in some locations, single atypical events can have a significant effect. Chance variation can act to skew comparisons in a manner that would not be as apparent with larger numbers of baseline incidents, where random fluctuations are more likely to be overshadowed by underlying trends. For example, the Queensland evaluation report noted a sharp two-month increase in the intervention area during the programme period. This was believed to reflect the activities of a single prolific offender operating in the area, estimated to have committed more than 130 break and enters during those two months (representing 18 per cent of all offences reported in the intervention area over the whole year of programme operation). An adjusted analysis, using only 10 months of data (excluding June and July data from each time period to remove the potential effect of the single prolific offender from the comparison) showed only a four per cent increase compared to the unadjusted analysis of a 19 per cent increase in the intervention area.
  
• Long-term trends in each area, as well as wider trends in break and enter offences in the jurisdiction as a whole, also may have consequences for outcome comparisons. For example, there is a tendency for sharp increases and decreases to revert to the mean over time, that is, to return to average levels. Analysis of 10-year trends in the Queensland evaluation report shows differences in the both the fluctuation of annual figures over time and in baseline levels between the intervention and control areas⁵⁴. SA shows more volatility in annual rates in the two intervention sites than in the three control areas over the five-year period 1993-97⁵⁵. The different direction of change in the total number of reported break and enters between the pre-programme and programme periods may reflect, at least in part, the continuing effects of these longer-term trends in each area, within the wider context of state-wide trends⁵⁶.
  
• As with most operational research, there is far less opportunity to control conditions than with laboratory experiments. Extraneous factors can affect the purity of intervention and control group comparisons. Local and regional police may make decisions in line with their operational priorities that have unintended implications for programme evaluation. For example, new crime prevention initiatives were introduced in one of the Queensland control sites during the period of the programme's operation. South Australian police were subject to a major structural reorganisation during the programme's operation that could affect policing priorities and activities in relation to break and enter responses in the intervention and control areas.
  
• The operation of the programme, and associated public awareness about its operation, may have affected reporting rates in the intervention areas. For example, victims' expectations about police action may have been raised by publicity about the trial programme, so that some victims who previously may not have reported an incident have been encouraged to report the offence to police. The 1998 ABS Crime and Safety Survey⁵⁷ shows non-reporting rates for break and enter were 28 per cent in Queensland and 22 per cent in South Australia, while non-reporting rates for attempted break and enters were 76 per cent and 72 per cent respectively. Given these Statewide proportions, there is potential for increased reporting in the intervention area to have had a noticeable effect. However, the magnitude of any results cannot be determined from the available data.
  
• Methodological issues also may be affecting outcome comparisons. For example, there are some differences between the geographic boundaries used by the programme to determine intervention eligibility in the South Australian project and those applied in the jurisdictional outcome evaluation (the latter due to the use of a refined data set as discussed in the South Australian outcome evaluation report⁵⁸).

Continuity of programme effects

Both programmes show a decline in reported break and enters in the six to eight month follow-up period after the projects concluded that was not matched by changes in the control sites. This raises the possibility that the real effects of programme intervention on overall break and enter rates may be realised in the long-term rather than the short-term. That is, effects occur only after the programme has had the opportunity to operate fully for a reasonable time to overcome implementation difficulties that arise in the early months of any trial programme; and possibly also after any effects on the rate of reporting incidents to police, triggered by the introduction of, and the publicity surrounding, a new programme have stabilised.

Assessment over a longer follow-up period would be useful, but this was outside the scope of the current evaluations.

Unintended consequences

Displacement is often described as an unintended consequence of crime prevention programmes. This commonly involves spatial displacement, in which offender activity is diverted to another area where the programme is not operating. It also includes crime type displacement to other offences, for example, offenders move from break and enters to car theft or robberies. Other types include target displacement (for example, from repeatedly victimised locations to residences not previously victimised), tactical displacement, where the same crime type is committed using a different method, and temporal displacement, where the same crime occurs at a later date.

Spatial displacement was assessed in both jurisdictional evaluations by comparing trends between the intervention area relative to an adjacent control site and a non-contiguous control site. If large-scale spatial displacement were operating, the adjacent control site would be expected to show increased victimisation relative to the intervention site that is not reflected in the non-contiguous site.

In Queensland, repeat victimisation increased by 86 per cent in the adjacent site compared to a fall of 18 per cent in the non-contiguous area. There was also a much higher ratio of increased risk in the first control site (see Table 7). However, the South Australian evaluation reports the opposite trend, with the adjacent area showing a decline in repeats while the non-contiguous area showed an increase. There is therefore no consistent evidence of spatial displacement within the two jurisdictions.

The consistent finding of increased total break and enters associated with decreased repeats in both jurisdictions raises the possibility of target displacement, that is, offenders changing their focus from repeatedly victimised to previously unvictimised residences. However, if this were a consistent feature of repeat victimisation programmes, then the extent of displacement should be proportional to the degree of repeat victimisation prevention. The pattern of results within the two jurisdictions does not support this.

The two demonstration projects do not show consistent evidence of spatial displacement as assessed by comparisons with adjacent control sites. This is consistent with reviews of international research⁵⁹, which conclude that displacement has a limited effect and is not as problematic as perceived to be in earlier studies. Target displacement to previously unvictimised residences could be operating, although this does not explain the magnitude of increase in total break and enter numbers. The results reported in individual jurisdiction evaluations do not provide evidence about other forms of displacement.

Victim response to intervention advice

In both programmes, follow-up surveys showed victims generally acted on the security advice given by the police or volunteers. In the Queensland intervention area, 77 per cent of victims took some action to improve their home security, as did 62 per cent of victims in the South Australian programme. Property marking action was the most common response taken by the Queensland victims (42 per cent) and installing window locks (41 per cent) by South Australian victims, particularly in the intervention area where the free lock installation service operated.

This take-up rate is somewhat lower than those reported in international research. For example, UK Home Office research⁶⁰ found takeup rates of 72 per cent for property marking, 35-65 per cent for burglar alarms, and 84-90 per cent for door locks, ranging from 35-97 per cent across different security measures and target groups. However, differences between countries in baseline levels of security need to be considered.

The take-up rates for particular security measures in both programmes are comparable with, and in some cases appear to be higher than, figures reported in the Australian Bureau of Statistics (ABS) 1998 Crime and Safety Survey⁶¹. This survey reports on the percentage of break and enter victims that added security measures in the last 12 months in each jurisdiction. For example, 20 per cent of South Australian break and enter victims in the ABS survey reported adding deadlocks on doors compared to 37 per cent of victims in the South Australian programme intervention areas that had installed door locks. The figures for installing a burglar alarm were 13 per cent and 19 per cent respectively.

In Queensland, the percentage of victims installing burglar alarms and fitting security screen doors were the same between the ABS survey and the intervention area results. However, in the ABS survey, 17 per cent reported adding deadlocks on doors and 21 per cent fitting locks, bars or grilles on windows. In comparison, 39 per cent of victims in the programme intervention areas fitted new locks to doors or windows and 15 per cent fitted window screens or grilles.

The rates at which security advice was taken up are broadly comparable between the two demonstration projects. Victims receiving the programme intervention appear to be more likely to improve their home security than reported among other victims of break and enter in victim surveys of the whole jurisdiction.

Comparison of common intervention elements

Both demonstration projects included various intervention elements, but these were implemented somewhat differently. Target hardening (particularly improving locks and bolts on windows and doors) was a core feature of both programmes. All households receiving the intervention in both projects received a security audit/assessment and advice on appropriate security action to reduce risk of future victimisation.

In one of the two South Australian intervention sites, a lock installation service was offered where a local handyperson installed locks to the value of $200 free to victimised residences on advice of the volunteer conducting the security audit. The intervention site providing this service reported a higher rate of installing door locks (46 per cent of residents took this action compared to 31 per cent in the other site) and window locks (50 per cent and 35 per cent respectively).

Property marking was also a standard feature of both interventions. In the Queensland project, all victimised households were provided with a property marking kit for residents to use, as part of the first-tier Stopbreak component of the programme. In the South Australian project, volunteers made immediate referral to one of two free services for the victim to access an engraver to mark their property. The extent to which victimised residents reported marking their property is similar in the two studies, with 40 per cent of South Australian victims and 42 per cent of Queensland victims marking their property.

Advising near neighbours was the responsibility of the patrol officer in the Queensland project, either by direct contact or leaving an advice card in neighbours' letterboxes. This occurred in only an estimated one-third of cases. It is not known what proportion of victims made personal contact with near neighbours about the incident and were given advice on preventive action that could be taken by neighbours. In the South Australian project, victims were asked to make contact with their neighbours or agree to the volunteer making the contact. Neighbours were spoken to in 84 per cent of cases and provided with an information kit. Increased contact with neighbours was reported by surveyed victims in 49 per cent of instances.

It appears that an approach of encouraging victims to make personal contact and providing them with resource information to give to neighbours may be more successful in ensuring neighbours receive this information than making first response police officers responsible for this function.

The other common intervention feature is the security audit or assessment that has been discussed in the previous section.

Cost effectiveness

The Queensland evaluation report gives costs of $42 per reported residential break and enter in the intervention area for the expenses associated with the programme's running cost. This excludes police salaries and related costs and expenses specific to the evaluation component of the project. It covers cost of equipment, such as the portable alarms and property marking kits, crime prevention resource material, and motor vehicle lease and running costs for the delivery of the Hot Dot and Hot Spot interventions. The information kit component of the Stopbreak response accounts for approximately $10 per incident.

The South Australian project reports a cost of $37 per intervention. This covers the cost of training the volunteers, items of equipment (such as mobile phones), and ongoing support meeting costs. It excludes salaries and associated costs of the project team or other participating agencies, as well as the follow-up evaluation survey. The cost calculated to $8.18 per hour of volunteer time, compared to the average hourly salary of a police constable at $21.35. Volunteers contributed a total of 4,438 hours of their time (an average of 97 hours per volunteer) to the project.

The different programme elements and differences in related programme costs subsumed in the auspicing agency budget means that the unit costs between the two demonstration projects are not directly comparable. However, they do indicate that an effective repeat victimisation programme can be introduced at a relatively modest cost compared to the costs of an incident to the victim and the community.

In both jurisdictions, unit costs would decrease if the programme operated over a longer timeframe or a wider area, given overhead costs would be distributed across a larger number of interventions.

Other benefits

Both projects report a number of benefits consistent with the type of strategic approach adopted. In the Queensland police model, there were identified benefits for police practice and competency acquisition among operational officers that would extend beyond the programme. The South Australian evaluation found 80 per cent of the 30 volunteers active at the end of the project continued on in other Local Crime Prevention Committee projects. They required minimal additional training for these roles because of the strong skill base established during the original work. The evaluation report also noted improved local collaboration between police and the relevant Local Crime Prevention Committees.

Return to the top

Key success factors across projects

The experience of the two demonstration projects highlighted a number of features that were particularly important for the success of the programme. Each of these is described below and addressed in the programme guide presented later in this report.

Planning and project management

Both projects spent considerable time and effort planning and designing the relevant programmes, with obvious benefits for programme operation and acceptance when implemented. Critical issues to address at planning and development stage include mechanisms for identifying repeat incidents, processes for notifying service deliverers of locations to receive intervention, and design of a range of effective prevention responses appropriate to the strategic approach adopted and to local circumstances. These are discussed in more detail in the programme guide.

Establishing clear and achievable objectives and selecting intervention areas consistent with those programme objectives is particularly relevant for a trial programme. For example, (as described in more detail in the implications for policy and practice section of this chapter), selecting an area with a high rate of repeats is important if the objective of the repeat prevention programme is to reduce total burglary rates.

Strong project management, including continuous quality monitoring, is critical to the development and evaluation of a successful trial programme.

Information systems

Both projects identified issues concerning the accuracy, consistency, and comprehensiveness of police crime report databases. This required time and effort to correct or resolve in order to meet the requirements for a repeat victimisation project. The Queensland evaluation report noted that the need to manually check and correct address errors and inconsistencies before electronic data was suitable for analysis and mapping required substantial resources and caused considerable delay to the project. In the South Australian demonstration project, a decision was made to use the date that the incident was reported to police, rather than the date of occurrence as the data field for identifying repeats, because of the potential for missing or conflicting dates.

Both projects considered the adequacy of police data was an important enough issue to take into account in deciding on where the intervention or control sites would be located. For example, the most appropriate contiguous control site for one South Australian intervention area (the one with the longest common boundary) was rejected because the rate of inaccuracy in address records was much higher than other areas considered. It would have required considerable additional work to check address records to meet project requirements. One of the reasons that the Queensland intervention site was selected was that there had already been some data analysis and record verification as part of a previous research project.

Even when there is only a small proportion of inaccurate or duplicate records, identifying and excluding these from a larger database can be costl y in both time and resources. For example, to remove the less than one per cent of duplicate records from the South Australian database involved analysis of more than 28,000 reported incidents. Given the relatively small number of repeats as a proportion of total break and enters identified in both projects, not excluding duplicates or correcting other address-based inconsistencies in police records could either markedly over-inflate or underestimate the real rate of repeats.

The difficulty in using police records to identify repeat incidents has been one of the most commonly identified problems in the international research⁶². The experience of the two demonstration projects reinforces the importance of establishing and maintaining accurate address information if police records are to be used to identify intervention targets and as the basis for information systems that evaluate programme success.

Designing effective intervention responses

Developing a range of intervention elements that can be applied as appropriate to suit the particular circumstances of the victimised location is a key component of delivering an effective prevention response. The two demonstration projects established similar intervention elements, although these were implemented somewhat differently. As described in earlier sections of this report, with the exception of near neighbour contact, there were broadly similar take-up rates for each comparable component.

Each demonstration project also introduced unique features relevant to the strategic approach adopted. For example, the Queensland model included an area-wide Hot Spot strategy and increased police patrols as part of the Hot Dot level intervention for some locations. The South Australian project included informal victim support and, where appropriate, referral to victim support agencies and other community-based services as an intervention component.

Both demonstration projects applied a problem-solving element to the intervention, in that security advice given to each individual victim was targeted to the circumstances of that household, as assessed by a detailed security audit. In line with a problem-solving approach, individual locations also had the potential to receive some different intervention elements. For example, a Hot Spot response in two identified areas, additional police patrols in some locations, loans of portable alarms in particular circumstances, and victim referral in appropriate instances. However, generally all households were intended to receive core elements of security advice on target hardening, property marking equipment or access to engraving services, and near neighbour contact and advice.

The research and practice literature indicates that a problem-solving approach, where the response is tailored to individual circumstances, is a particularly effective crime prevention approach. To achieve maximum effectiveness, this would include a range of responses including different strategies identified in the crime prevention literature as increasing the effort required by offenders, increasing the risk of detection, and reducing rewards to offenders, as well as strategies that target offenders. As noted in the Queensland evaluation report, the project had a restricted problem-solving focus. Applying a problem-solving approach in the full sense would involve a wider range of options and strategies including, in some instances, focusing on offenders rather than locations.

Including an area-wide intervention (the Hot Spot response) in the Queensland demonstration project expands the scope of the intervention and extends the problem-solving approach beyond individual households. The lesson from international research⁶³ is that small area burglary interventions need a high level of intensity and coverage to be successful. The Queensland project reported difficulty in mobilising resident involvement in the hot spot interventions. The design and implementation of this sort of response therefore needs particular attention.

The experience of the two demonstration projects supports the effectiveness of the selected intervention responses and applied approach to reducing repeat burglary. The practice literature indicates that applying an extended range of responses within a wider problem-solving approach, such as integrating the situational crime prevention measures used in these projects with an offender focus, may have additional benefits.

Choosing an appropriate service delivery model

Another important factor is selecting the most appropriate service delivery model, given the location. The two demonstration projects have confirmed that effective prevention programmes can be delivered either within a policing model, where police officers are responsible for delivering the intervention elements, or a community-based structure under which volunteers deliver the prevention services.

Based on the experience of the demonstration projects, a key success factor is the capacity of the organisational or community-based infrastructure to deliver appropriate management, coordination, training, quality monitoring, and other professional support necessary for staff or volunteers to deliver a high and consistent standard of intervention services. The ability of the programme's management structure to assure both key stakeholders and the wider community that victim confidentiality will be maintained is particularly important for a community-based model. These issues are discussed in the programme guide, together with strategies to address them.

The different models have varying capacity to establish and maintain effective links with other sections of the organisation or other agencies. This has implications for the degree of effort that needs to be spent in maintaining those links. For example, the South Australian project experienced low notification rates, in that project staff were only notified by police of approximately one-third of victims eligible to receive the intervention. If the service model operates within the police structure, those problems can be overcome by direct access to police records, or capacity to use internal monitoring and accountability mechanisms to change operational police practices. An externally managed programme needs to expend considerable effort in establishing and maintaining access or notification procedures.

On the other hand, a community-based project is likely to have existing networks and links with other community agencies that can facilitate intervention services and programme credibility within the local community. A police-managed project may need to spend time and effort to identify, establish links, negotiate service protocols, and build credibility with those community agencies.

The experiences of the demonstration projects suggest a strong partnership approach, where police and community share responsibility for programme development and management, may be particularly productive and overcome some of the difficulties identified in the programmes implemented in each jurisdiction.

Involving the right people

A comprehensive problem-solving approach to crime prevention means involving a range of public sector and community agencies to participate in developing and delivering the intervention. For example, the South Australian project involved Volunteer SA in the recruitment, selection and training of volunteers and in providing advice on volunteer management. The Queensland project involved a representative from the government department responsible for public housing, given the number of public rental homes in the area.

Selecting suitable volunteers is a key success factor where programme delivery involves volunteers rather than paid officers, given the time and skill demands associated with delivering crime prevention services at the victim's home. Providing a quality service also means officers or volunteers are aware of, and sensitive to, the effects of victimisation, and are able to respond appropriately to victims when delivering the prevention services.

As noted earlier, strong project management is an important success factor, and it is critical to appoint a suitably skilled and motivated project manager or coordinator.

Training and competency development

Both demonstration projects stressed the importance of adequate training of officers and volunteers delivering services to victims. Good practice features for training content and processes identified in both demonstration projects are described in the programme guide.

Effective management and monitoring

This refers to both managing and monitoring the programme and individual performance of service deliverers. For example, programme monitoring identified the low notification rates experienced by the South Australian demonstration project and resulted in new procedures being established between the police and the project team. Similarly, in Queensland early programme monitoring identified individual officers not complying with the project requirements and resulted in introduction of a supervision process by the Officer in Charge that prevented any further incidents of non-compliance. Quality monitoring of volunteer performance identified two South Australian volunteers whose involvement with the project was terminated.

Strong programme management, including overall programme and individual performance monitoring, from the beginning of the programme and operating throughout its life are critical.

Organisational support

Both demonstration projects identified the need for support and ongoing commitment by police management for effective operation of the programme.

Communication

Both demonstration projects reinforced the importance of regular ongoing communication between all participants, particularly the project team, officers responsible for identifying households falling within the intervention parameters, and the officers or volunteers delivering services to victims.

Resources

The Queensland and South Australian projects had access to sufficient resources to be able to allocate dedicated project staff, develop quality crime prevention resources for distribution to victims, provide adequate logistic and training support to volunteers, and even fund security equipment for particular households. The Queenslanddemonstration project reinforced the importance of allocating sufficient time and resources to identification of repeats from police records, particularly in determining hot spots. The South Australian project showed that the coordination and management of volunteers, particularly allocating tasks, imposed considerable demands on programme teams' time.

Effective planning and resource allocation and management are critical to programme success, and relate to skills and competencies as much as to financial and physical resources.

Sustainability

Both demonstration projects designed models intended to be able to be adopted in other locations. The Queensland programme is easily built into mainstream policing. The South Australian model, given its heavy demands on infrastructure support and coordination, would need to be carefully planned before being implemented in other community locations to ensure that suitable infrastructure is available.

Implications for policy and practice

Key questions in considering the policy and practice implications of repeat residential break and enter prevention are:

• To what extent is repeat residential break and enter preventable?
  
• Is targeting repeats a useful strategy for reducing break and enter rates?
  
• Are there other compelling reasons for introducing a repeat victimisation focus?
  
• Are there demonstrably superior models of repeat break and enter prevention?
  
• Does targeting repeats provide a resource effective approach to crime prevention?
  
• Is targeting repeats a sustainable crime response in different locations?
  
• Are there associated unintended consequences that need to be addressed?

Capacity to prevent repeat burglary

The outcomes of the Queensland and South Australian demonstration projects confirm the findings from international research that repeat break and enter is preventable. However, the Australian projects report a much more modest effect on repeat victimisation than has been reported in the international literature.

The smaller effects reported in the Australian projects should not be interpreted as evidence of less effective programmes. Differences in overall levels of repeats and pre-existing standards of home security may be responsible. There is a greater capacity for an intervention programme to have demonstrable benefits where the target offence occurs at high levels and pre-existing preventive action is low. For example, introducing target hardening at a residence with a history of recent break and enters and without any security measures is more likely to show a positive effect than introducing initiatives at a dwelling that already has adequate security.

The Queensland and South Australian intervention areas experienced relatively low levels of repeat victimisation immediately before the programme, compared to those British studies which reported large reductions. For example, the most commonly cited UK study (the Kirkholt project) found repeat rates of almost 50 per cent before the intervention. Both Australian projects reported rates fewer than 10 per cent in the immediate pre-programme period.

The more successful repeat prevention programmes in the UK were conducted in housing estates that had very high crime levels and social disadvantage⁶⁴. For example, a programme review⁶⁵ notes the Kirkholt project had a burglary rate of 25 per cent immediately before the programme, compared to a national average of two per cent reported to police and five per cent reported in crime surveys. The second highest UK area had a rate almost half of the Kirkholt level, at 13 per cent.

In the Kirkholt project simple target hardening measures, such as replacing coin-fed fuel meters, had a dramatic effect. Existing levels of household security will affect the capacity of programmes offering simple target hardening measures to have a demonstrable benefit. Australian households report⁶⁶ reasonably high levels of security measures. For example, 15 per cent of Australian residences have a burglar alarm and 62 per cent have deadlocks on some or all doors. In fact, 13 per cent of victims in the Queensland intervention area stated they did not add any additional security because they already had most of the relevant measures in place. More sophisticated types of intervention are needed where pre-existing household security measures are at an adequate or relatively high standard.

Overall, repeat break and enter is preventable, although the Australian experience shows more modest outcomes than some of the international research. This needs to be interpreted in the context of lower repeat levels overall and possible differences in pre-existing household security measures. An implication for practice is the importance of selecting appropriate intervention sites. Break and enter prevention programmes with a repeat victimisation focus are not only more relevant, but also more likely to be demonstrably effective, in areas with high pre-existing repeat rates.

Return to the top

Utility of targeting repeats to reduce break and enter rates

Both Australian demonstration projects reported an increase in overall break and enter rates, counter to the expected programme effects. Based on the evidence of these two programmes, targeting repeat victimisation does not appear to be a demonstrably effective strategy for reducing overall break and enter rates in the Australian context.

Part of the reason may be the relatively low contribution of repeats to total break and enter rates described earlier. Repeats need to make up a significant proportion of total offences in order to have any discernible effect. For example, to reduce total break and enters by 10 per cent in an area where repeats constitute one-fifth of offences, the intervention strategy would need to reduce repeats by half to meet this overall target. However, if repeats constitute only 10 per cent of all break and enters, then the strategy would have to prevent every repeat in order to achieve only a 10 per cent decrease in total incidents.

In small intervention areas, random and seasonal fluctuations in the number of break and enters can overshadow real programme effects. For example, it is not unusual for crime rates to fluctuate by five to 10 per cent or more in such areas from year to year. Local programmes introduced and evaluated rather than state-wide programmes, would have greater difficulty in demonstrating notable programme effects above random fluctuation if the area has low levels of repeat victimisation.

For example, if every one of the repeat incidents reported in the pre-programme period of the Queensland project were prevented, this would have reduced the number of overall break and enters by nine per cent. This is in the context of annual fluctuations of between +18 per cent and -25 per cent for the intervention area over the previous three years.

Another possible contribution to the increase in overall break and enters may be displacement at local area level, which is discussed in the section on unintended consequences below. Other possible reasons for the outcomes for total break and enter have been discussed earlier in the report. Regardless of reasons underlying the findings of the demonstration projects, there are implications for policy and practice, particularly for selection of intervention areas and programme objectives. There would appear to be three essential preconditions before a repeat victimisation focus is likely to lead to a significant reduction in residential break and enter rates. They are:

• repeat incidents constitute a substantial proportion of total incidents in the intervention area, so that preventing repeats has the capacity to have a demonstrable effect on overall rates
  
• the intervention responses introduced are effective in preventing a significant number of repeat incidents
  
• the repeat prevention programme does not displace offences to previously unvictimised residences, (i.e. offenders simply changing their focus of activity to other easier targets in the locality).

Alternatively, a repeat victimisation focus may be introduced for reasons other than reducing overall break and enter rates, as discussed below.

Reasons for targeting repeats

The implications and preconditions for targeting repeats where the overall goal is break and enter reduction have been discussed above. However, there are other policy and practice reasons for a repeat victimisation focus, as described earlier in this report.

They include:

• to address the disproportionate impact of victimisation experienced by repeat victims
  
• as a strategy for apprehending prolific offenders
  
• to provide a cost effective response to crime by directing scarce resources at known high risk locations.

Evidence base on effectiveness of different models

The previous chapter provided a comparative analysis of the outcom es of the two approaches used in the Australian demonstration projects. The evidence provided in the programme evaluations does not support either model being clearly superior in delivering prevention outcomes. Both achieved modest successes in addressing repeats, were unsuccessful in reducing total break and enters, experienced various implementation difficulties, and reported other benefits associated with the programmes.

The implication for policy and practice is that there is no single model to adopt in reducing repeat break and enter offences. As with any crime prevention programme, success will be determined by the fit between the programme introduced and the targeted problem and by the capacity of the local situation to effectively deliver the programme.

A partnership approach that optimises the information sources and expertise of police and other relevant criminal justice agencies, with the resources and commitment available within the community, would appear to be ideal. Whether this means that the most appropriate model is for police or another criminal justice agency to take a lead agency role, or for the programme to be managed by local government or the community sector with appropriate criminal justice agency participation, will depend on the existence and effectiveness of relevant structures, stakeholder relationships, approaches and past practices.

The experiences of both projects identified several common success factors in the two models, which form the basis of the good practice features summarised later in this chapter.

Resource effectiveness

Both projects have demonstrated that an effective and credible trial programme can be introduced and evaluated within a reasonably modest budget. The unit costs reported in the programme evaluations ($42 and $37) would decrease if the programme were expanded (i.e. involving a larger number of cases) or extended over a longer period of time, because programme overheads therefore would be distributed more widely.

These unit costs do not represent the full economic cost of delivering the programmes (police, project staff, and other organisational costs were excluded). However, they provide an indication of the cost of project specific materials and expenses associated with the use of volunteers that would need to be budgeted for or derived from other sources (for example, sponsorship, in-kind donation, loan of equipment, etc).

Based on the figures provided, the programmes appear to provide a cost effective response to repeat break and enter prevention.

Sustainability and wider application

Both demonstration projects had designated timeframes over which the programmes were funded, operated and evaluated. The ongoing sustainability of trial programmes, particularly where there is a time-limited external funding contribution, is a major policy and practice issue for any crime prevention programme.

In Queensland, the model developed in the demonstration project has provided the basis for The At Risk Premises (TARP) project, which is being implemented Statewide by the Queensland Police Service. Elements of the trial programme have therefore been placed in the mainstream of organisational activities.

The South Australian demonstration project concluded its operational activities in January 2001, and the evaluation outcomes are still to be widely promulgated. Therefore, it is too early to determine whether the model will be adopted in other locations. However, the South Australian programme evaluation[⁶⁷ found that most stakeholders believed that Neighbourhood Watch or community groups could undertake the intervention programme after the project concluded, subject to two essential preconditions:

• a manager or coordinator, generally perceived as needing to be a professional or paid manager
  
• appropriate training (assuming volunteers had been adequately screened when recruited).

Specific issues that would need to be resolved if either transferring responsibility to a community-based group or establishing an intervention programme managed directly by a community-based group were identified by stakeholders as:

• source of funding to enable training and programme management
  
• ways of protecting victim confidentiality and privacy
  
• methods for recruiting sufficient numbers of suitable volunteers
  
• adequate volunteer screening processes
  
• establishing appropriate referral mechanisms
  
• extent of police and council cooperation.

As indicated above, the Queensland experience shows that the programme is sustainable and can be incorporated into the mainstream of State prevention activity. Although the South Australian trial project has concluded, key stakeholders interviewed support the sustainability of the model.

Unintended consequences

A significant issue for crime prevention is displacement; that is, intervention does not prevent crime but simply transfers the problem to another location or target. There was no consistent evidence of spatial displacement across the two projects, that is, of repeat incidents reducing in the intervention area but increasing in the adjacent control site because offenders transfer their activity to another geographic area.

The evaluation results do not rule out target displacement effects, that is, offenders directing their activity from repeat victims to previously unvictimised households in the local neighbourhood. Analysis of the pattern of results shows direct displacement could not account for the level of increase. Even if every repeat incident were displaced to a previously unvictimised residence, the effect on repeat rates would be well below the actual increases experienced in the intervention sites.

However, target displacement is an issue that should be considered in programme design until there is a stronger body of Australian evidence on the issue. For example, integrating a repeat victimisation focus within a broader break and enter reduction strategy that also addresses previously unvictimised homes may be a useful approach.

To some extent, both demonstration projects incorporated an element of wider break and enter prevention. The contact with near neighbours included some crime prevention advice for those residences. However, in the Queensland project this element was poorly implemented. The Hot Spot response also provided an intervention response to non-victims living in the hot spot area. Again, there were implementation difficulties identified in the Queensland evaluation report, particularly the difficulty of mobilising local residents to be involved.

Both projects reported positive unintended consequences. These have been described in the previous chapter.

The major policy and practice implications relevant to unintended consequences relate to preventing potential displacement (noting that there is not yet a clear body of evidence from Australian research as to whether displacement is a major issue for programmes of this type). Integrating repeat victimisation initiatives within wider break and enter prevention strategies would mitigate possible displacement effects.

Recommendations for future policy directions

The Queensland and South Australian demonstration projects were designed to test the effects of two different approaches to burglary reduction focusing on preventing repeat break and enters. The results showed repeats can be prevented (albeit to a modest degree), but doing so did not have a positive effect on total break and enter rates. Before addressing the ultimate policy question of whether targeting repeat victimisation is a worthwhile crime response, a final evaluation issue needs to be considered: whether the modest results for repeats and the contrary to expectation findings for total break and enter are due to:

• theory failure, that is, the basis of the approach is unsound
  
• implementation failure, that is, the approach is sound but not properly put into effect
  
• measurement failure, that is, inability to adequately detect change and measure the extent of a programme's effects.

The two demonstration projects do not provide enough evidence to make a definitive decision on whether a repeat victimisation focus to break and enter should be a preferred strategy for Australian crime prevention. Implementation issues detailed earlier in the report, such as the low notification and participation rates, can markedly affect project outcomes in directions that cannot be determined from the available evaluation data. The timeframes over which the projects operated (12 months to develop and implement strategies) and the relatively small numbers of incidents and low repeat rates introduce measurement issues. The effects of implementation and measurement failure cannot be disentangled from theory failure in the evaluation data available from the two projects.

To definitively answer the question of whether targeting repeat victimisation is a worthwhile crime response that should be widely implemented requires more evidence than can be provided by the two Australian studies. However, the outcomes reported in the demonstration projects indicate that the approach has promise. Repeat break and enter prevention programmes would appear to be able to provide appropriate, effective, and cost-efficient responses to such crimes. But given the results of the two projects, the goals of such programmes are more appropriately focused on preventing repeats for reasons other than overall break and enter reduction, unless specific preconditions are met.

Based on the experiences of the Queensland and South Australian programmes, and the conclusions of the international research and practice literature, ideal preconditions and good practice features for an effective repeat break and enter prevention programme would appear to be:

• high pre-existing levels of repeats in the target area(s)
  
• programme objectives that relate to repeat prevention rather than total break and enter reduction
  
• programme development using a suitable planning and problem-solving approach that recognises the scope of the problem and develops locally appropriate solutions
  
• the programme is developed and managed through a partnership between government and the community to optimise community resources with expertise available from the criminal justice sector, police and crime detection responses
  
• strong programme management (preferably with an appropriately qualified, full-time coordinator)
  
• appropriate training (initial and ongoing) in service delivery, extending over sufficient time to develop requisite practical knowledge and competencies, and involving mentored practice in real situations
  
• a monitoring process that addresses both procedural compliance and competency development from the beginning of the programme and operating throughout its life
  
• information systems supporting programme operation and evaluation accurately and readily identify repeat incidents, and wherever practicable, include mechanisms for taking unreported incidents into account
  
• effective procedures, continuously monitored, for transferring information between those areas responsible for recording crime, for identifying repeat locations, for allocating tasks to officers and volunteers to provide services, and for delivery of the intervention response
  
• intervention measures, combining crime prevention measures at individual residences to increase offender effort and risk while reducing rewards, intervention responses that include near neighbours of victims, area-level approaches to identified hot spots, enhanced police investigation and offender detection processes, and strategies to target offenders and disrupt stolen goods distribution channels
  
• a repeat prevention programme which is integrated or coordinated with a wider burglary reduction strategy to prevent potential target displacement
  
• hot spot interventions which are designed and adequately resourced to provide high intensity and high coverage of prevention activities
  
• a timeframe for operating and evaluating any trial programme that is long enough to establish and assess long-term outcomes of the intervention
  
• ensuring that any trial programme can be sustained, by taking into account local context and capacity issues, so that if the programme is evaluated as successful, the model could be continued at that location or used elsewhere.

Chapter summary

The two demonstration projects achieved modest effects on repeats relative to control sites, were unsuccessful in decreasing total break and enters, experienced various implementation difficulties, and produced other project benefits consistent with the strategic approach adopted.

Key success factors in the projects related to: planning and project management; information systems; designing effective intervention responses; choosing an appropriate service delivery model; involving the right people; training and competency development; effective management and monitoring; organisational support; communication; resources; and sustainability.

Implications for policy and practice relate to the feasibility of preventing repeats at meaningful levels, the utility of a repeat strategy for reducing overall break and enter rates, other reasons for introducing a repeat victimisation focus, cost effectiveness, sustainability, and unintended consequences; as well as whether there are demonstrably superior models to apply in the Australian context.

The meta-evaluation concluded that repeat break a nd enter is preventable, but that programmes should be introduced for reasons other than overall break and enter reduction unless specific preconditions are met: that the programmes appear to provide a cost effective response to repeat break and enter; that the models trialed are sustainable; and that there is no definite evidence of spatial displacement - but that programme design should include strategies to minimise the potential for target displacement. It also concluded that there is no single model to adopt in reducing repeat break and enter. Instead, the strategic approach and the management and implementation of the programme should be tailored to local circumstances.

The results of the two demonstration projects do not provide enough evidence to make a definitive decision on whether a repeat victimisation approach to break and enter should be widely adopted in Australia. However, the outcomes reported indicate the approach has promise. Good practice features were described, based on the experience of the Australian demonstration projects and the international crime prevention literature.

Return to the top