Australian Institute of Criminology

Skip to content


In this section, an outline of the rationale for the project will firstly be presented. Second, the findings of the project will be summarised in the context of previous findings. Third, the implications for policy arising from the project will be discussed. Next, the limitations of the research will be reported. The section will conclude by outlining directions for future research.

Rationale for project

This project aimed to assess whether communities that generated chronic offenders and carried substantial cost burdens associated with offending could be identified. If such communities could be identified, costly interventions may be targeted towards these locations to reduce offending, crime, victimisation and Indigenous overrepresentation. The project drew on methods and findings from research focused on offender trajectories and crime and place. Trajectory research finds that there is a small group of chronic offenders who account for a disproportionate amount of offending and costs (Piquero 2008). While this group can be retrospectively identified, research has not adequately been able to prospectively identify individuals who may be on this trajectory based on risk and protective factors. Findings from crime and place research suggest that these offenders are not randomly distributed geographically and highlight the importance of understanding the temporal aspects of locational data such as offender residential mobility (Gabor & Gottheil 1984; Oberwittler 2004; Sabol, Coulton & Korbin 2004; Schwartz 2010; Wiles & Costello 2000). Given these findings, the project firstly explored the number of offender trajectories, their nature and their cost. The project then focused on individuals in the moderate and chronic offender groups and explored how individuals and costs were geographically distributed. The six research questions addressed by the study were:

  • How many distinct offender trajectories can be identified?
  • What are the demographic, offence and criminal justice system event characteristics associated with trajectory group membership?
  • What are the costs of offender trajectories?
  • Are some communities more likely to generate chronic offenders than others?
  • How residentially mobile are chronic offenders?
  • Which communities carry the cost burden of chronic offenders?

Summary of findings

Consistent with Piquero’s (2008) review of trajectory research, five offender trajectory groups were identified. The offending patterns of these groups were similar to those found by prior research. There was an adolescent-peaked group who offended at low levels (29.3% of cohort; 12.5% of offences) and two groups who offended at chronics—early-onset chronic offenders (3.0% of cohort; 28.1% of offences) and adolescent-onset chronic offenders (2.2% of cohort; 15.6% of offences). Additionally, there was an adult-onset group who offended at low levels (54.9% of cohort; 20.5% of offences) and an adolescent onset group who offended at moderate levels (10.5% of cohort; 23.4% of offences). About one-tenth of the two low offending groups were Indigenous, while between one-third and one-half of the moderate and chronic groups were Indigenous. Therefore, targeting offenders in these three groups is likely to be a useful approach for reducing Indigenous overrepresentation. Chronic offenders were more likely to have committed unlawful entry offences and theft and related offences. They were also more likely to have been subjected to each of the criminal justice system events that were examined and found to account for a disproportionate number of days sentenced to detention/incarceration and community-based supervision.

Costs were applied to the five offender trajectory groups and findings were consistent with previous research, with chronic offender trajectory groups found to account for a disproportionate amount of costs. Early onset (chronic) and adolescent onset (chronic) offenders were 5.2 percent of the cohort, but these two types of offenders combined accounted for 47.2 percent of total costs. On average, each chronic offender cost over $220,000 by the time they turned 21 years of age. Approximately one-tenth (10.5%) of the cohort were in the adolescent onset (moderate) group, but 22.4 percent of the costs were accrued by members of this group. Each adolescent onset (moderate) offender cost $58,116 by the time they turned 20 years of age. Four-fifths (84.2%) of the cohort were adolescent peaking (low) or adult onset (low) members, and 30.4 percent of total costs were accrued by members of these groups. On average, each offender in these low offending groups cost $9,535 or $9,971 respectively by the time they turned 21 years old. Differences between the actual costs of the offender trajectories in the current study and previous research may be explained by the length of criminal careers captured by studies, the offences included and costed in the studies and the overall costing method that is applied (Allard et al. under review).

While information about the trajectory groups and their costs provides useful information about the small group of offenders who account for a large proportion of offences, it does not enable crime prevention interventions to be targeted towards chronic and costly offenders. When the moderate and chronic groups were combined as chronic offenders, they represented 3.8 percent of the population and 15.8 percent of offenders. However, they accounted for 67 percent of offences and 68.6 percent of the costs. Because the residential location of chronic offenders may prove useful for targeting interventions, the proportion of the population in each POA who were chronic offenders was explored. The POA where chronic offenders resided when they first had contact with the criminal justice system was used to assign costs because of the emphasis placed on the early years of life by developmental crime prevention and ABS census statistics were used to determine the populations of POAs.

It was evident that chronic offenders were not randomly distributed, with two-thirds (n=224; 68.1%) of POAs having none or less than five percent of the 16 year old population identified as chronic offenders. One-tenth (n=33, 10.0%) of POAs had over nine percent of the population that were chronic offenders and 20.5 percent of chronic offenders came from these POAs. Most of these locations had a high proportion of Indigenous peoples in the population, were outer regional, remote or very remote locations and many faced extreme disadvantage.

Given that the residential POA when offenders first had contact with the criminal justice system was used to assign location, it was considered important to investigate offender residential mobility. About one-third (31.7%) of chronic offenders only had one POA, while one-third (32.1%) had three or more POA changes. While a significant proportion of chronic offenders were not residentially mobile, overall chronic offenders were substantially mobile in terms of the number of times they changed residential postal code after their initial contact with the criminal justice system.

Finally, the project identified communities that carried the burden of costly chronic offenders. The top 10 percent of POAs were identified based on total cost of chronic offenders and these postcodes were found to account for 40.4 percent of chronic offenders and 50.5 percent of the total cost of chronic offenders. Many of these POAs were located in regional Queensland, with each POA costing between $2.4 and $14.0m.

Implications for policy

The findings from this project indicate that chronic offenders represented a small proportion of offenders (15.9%) but accounted for a large proportion of offences (67.0%) and costs (68.6%). Three-quarters (77.5%) of chronic offenders were male, while one-third (33.8%) were Indigenous. On average, they committed 21.4 offences and had 6.7 finalised criminal justice system events. Chronic offenders were not found to be randomly distributed geographically and there was a substantial cost for some communities. The top 10 percent of POAs, where over nine percent of the population were chronic offenders, accounted for 20.5 percent of chronic offenders. The 10 percent most costly locations accounted for 50.5 percent of the total cost of chronic offenders.

These findings highlight the need for urgent action. Many of the communities where a high proportion of chronic offenders first had contact with the criminal justice system had extreme social and economic disadvantage. As such, these locations may benefit from community-wide programs that target the risk factors for offending by reducing substance abuse and unemployment and improving educational levels and housing conditions (Allard 2010). International evidence indicates that addressing community-wide risk factors through Vocational and Education Training, or community economic development, may reduce offending (Burghardt et al. 2001; McCord, Widom & Crowell 2001; Sherman et al. 1997). While the extent that these programs would reduce offending by Indigenous peoples remains unknown, the Closing the Gap strategy aims to improve educational and employment outcomes for Indigenous peoples (COAG 2009). As such, the current project may provide additional information to assist decisions about which locations should be targeted to most efficiently improve outcomes and reduce Indigenous overrepresentation as offenders in the criminal justice system.

A range of programs targeting the individual and their ecological environment may also be useful for reducing Indigenous overrepresentation. Early/developmental programs include parental training, home visiting, daycare/preschool programs and home/community programs (Farrington & Welsh 2003). Multi-modular programs such as Multi-Systemic Therapy typically focus on the family and the individual’s ecological environment to address risk factors occurring in multiple domains simultaneously. These programs could be implemented in locations that were either identified as having high proportions of the population who were chronic offenders or that were identified as costly locations. They could be made available to the entire communities or could be further targeted to individuals using risk assessment tools.

Evidence from meta-analyses indicates that programs targeting the family may reduce offending by between 13.3 percent and 52 percent (Aos et al. 2001; Drake, Aos & Miller 2009; Latimer 2001; Lipsey & Wilson 1998; Woolfenden, Williams & Peat 2002). Programs that adopt a Multi-Systemic Therapy framework reduce offending by between 7.7 percent and 46 percent (Aos et al. 2001; Curtis, Ronan & Borduin 2004; Lipsey & Wilson 1998; Littell, Popa & Forsythe 2005). Whether these programs would result in similar reductions in offending for Indigenous peoples remains unknown. Nevertheless, the POAs identified in the current study might be ideal locations where these interventions could be implemented and rigorously evaluated to determine whether they can have an impact on Indigenous offending.

The identified locations may also be prime sites where situational crime prevention interventions could be implemented. Situational crime prevention focuses on highly specific problems and the opportunities in specific environments that facilitate problem behaviour (Clarke & Felson 1993). Intervention aims to alter the immediate environment in which crime occurs or address factors within the context that may be causing or maintaining offending. As such, further spatial analysis of offending and crime data should be undertaken for the POAs where a high proportion of the population was found to offend or for the POAs that were found to have a high cost. Additional spatial analysis would enable determination of whether specific forms of offending occur in specific locations. The identification of micro locations would enable interventions to be planned and implemented that manipulate the environment to reduce the opportunities for offending. These interventions can result in real reductions in offending and have proven effective for reducing risk factors related to offending, such as substance abuse within Indigenous communities (Clarke 1997; Eck 2006; Richards, Rosevear & Gilbert 2011).

It must be emphasised that developing and implementing crime prevention interventions in many of the communities would be challenging. Many were disadvantaged and in regional, remote or very remote locations. The crime prevention literature suggests that successful programs have several core features, including:

  • highly skilled leaders and staff;
  • adequate funding;
  • effective coordination and collaboration mechanisms across government and non-government agencies; and
  • a high level of community involvement to ensure community acceptance and participation, that local community needs are met and to ensure that interventions are culturally appropriate (AIC 2012; Calma 2008; Cherney & Sutton 2007; Doone 2000; Gillbert 2012; Stacey and Associates 2004).

Even if interventions have these core features, they may still need to overcome the challenges resulting from poor access to services and infrastructure (Schwartz 2010).

Given the apparent usefulness of understanding geographic location for targeting crime prevention resources, other jurisdictions should consider using this approach to target interventions to reduce offending, crime, victimisation in Indigenous overrepresentation in the criminal justice system. A similar place-based approach for targeting resources that is gathering traction internationally and in Australia is justice reinvestment (Allen 2011; Clear 2011; Guthrie, Adcock & Dance 2011; House of Commons 2009; Queensland Government 2011; Schwartz 2010; Young & Solonec 2011). This approach involves using ‘justice mapping’ or ‘prisoner geographies’ to redirect a proportion of corrections budgets to the communities that generated the most costly prisoners. Mapping conducted overseas has enabled million dollar blocks to be identified and evidence is emerging that the approach is an effective way of reducing crime and expenditure on imprisonment (Schwartz 2010).The findings of the current study lend support to this approach, as costly chronic offenders were not found to be randomly distributed. Moreover, the methodology developed in the study may assist jurisdictions to assess the cost of offenders using a justice reinvestment framework.

Limitations of the project

Despite the potential importance of the findings, they should be interpreted in light of seven main limitations. First, the study was based on administrative data which is of variable quality and does not include offending that is not reported to justice agencies or attributed to an offender. Second, the study was not able to take into account cohort attrition (through death or population mobility) or migration into the cohort in assessing the offender trajectories. Taking migration and attrition into account may result in some variation in the final trajectory models identified (Eggleston, Laub & Sampson 2004). Third, the study did not control for the effects of exposure time when assessing the number of offender trajectories. Individuals in the cohort were in detention/incarceration for 62,870 days. When the number of days available for individuals in each offender trajectory group to offend is considered, the two low offender trajectory groups had the most time available to offend (<.01% of the time). Members of the moderate group were detained/incarcerated for 0.9 percent of the time, while members of the early onset and adolescent onset chronic groups were detained/incarcerated for 4.0 and 4.6 percent of the time respectively.

Fourth, criminal justice system costs were assessed based on the average cost of finalised events, taking into account how individuals flowed through the system, while the wider social and economic costs were assessed based on an update of Rollings’ (2008) assessment. In assessing criminal justice system costs, average costs were used although costs would vary based on factors such as whether the offender pleaded guilty, the offence type and the location of the offence. The cost of responding to offending in rural and remote areas is likely to be significantly higher for each event and individual than in cities. In assessing wider economic and social costs, six offence types were not assigned a cost. While these offence types could be considered less expensive, there were a large number of offences (32.7%) that were not assigned a cost. Inclusion of these costs would increase the wider economic and social costs of the trajectory groups, but particularly the adult onset (low) and adolescent peaking (low) groups. Members of these two groups had the highest proportions of the six offence categories that were not able to be included in the assessment of cost.

Fifth, the project was reliant on the POA recorded for each chronic offender when they first had contact with the criminal justice system and ABS census population data. Postcode 4000, which includes Brisbane City, was identified as being a location where a high proportion of chronic offenders first resided when they had contact with the system and as a high-cost location. Given the small residential population of this location, it is believed that this location may have been recorded when a residential address was not provided, for offenders who were homeless or for offenders who were detained. Additionally, there was considerable mobility among offenders, with two-thirds of high-risk offenders changing POAs at least once based on their contacts with criminal justice agencies and about one-tenth (9.5%) changing six or more times. However, there was no way of determining how frequently the chronic offenders moved residential address in the years prior to having contact with the criminal justice system or whether changes in POA location were not captured by criminal justice system data.

Sixth, POAs are a very crude approximation for communities. Some POAs are geographically very large with very small populations. Furthermore, while population data were available based on POAs, it should be noted that these are only approximations of POAs and that these data were subject to random allocation processes used by the ABS to prevent individual identification (ABS 2006a). Finally, there were also numerous challenges using the census data. The number of 16 year olds in 2006 was assumed to be an approximation for the cohort population. While the offender cohort would have been 16 years old in 2006, there was no way of determining the attrition from or migration into the cohort.

Directions for future research

Additional research focused on the cost of offender trajectories and considering their geographic distribution is clearly needed to promote the use of this evidence within policymaking environments. The need for this research is apparent given that jurisdictional differences in criminal justice practices, economic conditions, monetary values and geographic locations makes it difficult to generalise findings from one context to another. Moreover, there is considerable difficulty assigning market values to intangible costs and surprisingly little research has adopted a top-down costing approach based on methods such as willingness to pay. Additional research that assesses the costs of crime and assesses intangible costs will enable researchers to develop more valid and reliable cost estimates. The need for research that predicts future offending and differentiates offender trajectories based on risk factors and locations is also essential to further assist targeting of crime prevention programs.