Effective community-based supervision of young offenders

Foreword | An increasing body of research suggests that some interventions with offenders can reduce reoffending. While little of this research has focused on the impact of routine supervision of offenders on probation, parole or other community-based orders, a few studies have found that when supervisors make use of a number of specific practice skills, there is a reduced rate of recidivism for those under their supervision. Having first described the effective practice skills, the extent to which these are applied to a population of young offenders is assessed, along with the resultant effect on recidivism. The study involved the direct observation of 117 worker/client interviews conducted by juvenile justice workers in New South Wales. It was found, as with earlier studies generally done with adults, that when workers used particular practice skills, the young people under their supervision had lower reoffending rates. It also found that workers who provided a counselling role made more use of the effective practice skills than workers who did not.

Adam Tomison

A lot has been written about what works in interventions with offenders (eg Andrews & Bonta 2008; McIvor & Raynor 2008; McNeill et al. 2005). In recent years, there have also been a small number of studies focusing on what works in the routine supervision of offenders on probation, parole or other community-based orders. These studies have found that certain supervision skills offered by supervisors can lead to reduced levels of reoffending (Bonta et al. 2011; Dowden & Andrews 2004; Robinson et al. 2011; Trotter 1996). These studies have been predominantly undertaken with adult offenders.

This paper describes a study that examined the relationship between the use of these practice skills by supervisors in Juvenile Justice in New South Wales and reoffending rates by their clients (those under their supervision). It was hypothesised that it would be possible to identify the extent to which supervisors used particular practice skills through the direct observation of interviews by trained research officers. It was also hypothesised that the more the effective practice skills highlighted in earlier research were used, the less frequently the offenders under supervision would reoffend.

Literature review

A review of research (Trotter 2006) suggested that the following worker skills are related to reduced offending and increased compliance by those under supervision:

  • The worker is clear about their role. This includes helping the client to understand the dual social control and helping aims of supervision, and other issues such as confidentiality and what is expected of the client.
  • The worker models and reinforces pro-social values and actions, and makes appropriate use of challenging or confrontation. So, the worker is reliable and fair, and questions rationalisations for offending.
  • The worker helps offenders with problems that are identified by the client rather than the worker.
  • The worker encourages clients to focus on problems or issues that are related to the person’s offending (eg drugs, peers, employment, family issues).
  • The worker helps clients to develop strategies to address these issues. These strategies focus on practical issues and may include cognitive behavioural and/or relapse prevention strategies.
  • The worker takes a holistic approach to client issues rather than focusing on just one or two specific problems or symptoms.
  • The worker develops a therapeutic alliance with the client. In other words, they work in a collaborative, friendly, optimistic way so that the client develops trust in the worker as someone who can genuinely help them with their problems.

These principles are consistent with the best practice principles enunciated and researched by others in the field (eg Dowden & Andrews 2004; Raynor, Ugwudike & Vanstone 2010; Robinson et al. 2011). This is not to say that there is universal agreement on best practice principles in offender supervision. Bonta and Andrews (2010, 2008) have spent many years developing the ‘Risk/Needs/Responsivity’ model, which incorporates most of the principles outlined above but places particular emphasis on focusing on high-risk offenders, on using actuarial tools to assess levels of offender risk and needs, and on the use of predominantly cognitive behavioural approaches to assist offenders to address their risk-related needs.

By contrast, Ward, Meltzer and Yates (2007) have argued that the focus on risk and risk assessment has been at the expense of opportunities for offenders to develop ‘good lives’ on their own terms. He and others (eg McNeill et al. 2005) have emphasised the importance of therapeutic alliance and the worker/client relationship, which they argue may be compromised by a focus on risk factors that may be defined by workers rather than clients.

Several studies have specifically examined the relationship between worker skills and reoffending rates of clients. Some of these studies have also examined the extent to which training has influenced the use of the skills by workers. One study also looked at the relationship between use of the skills, education in social work and welfare, and further offending. These studies are briefly reviewed below.

Andrews et al. (1979) published a seminal study conducted on probation in Ontario, Canada, which focused on the use of skills in the routine community supervision of offenders. They found that probation officers who made use of pro-social modelling and reinforcement, problem solving and who had high levels of empathy, had clients with significantly lower recidivism.

A further study undertaken in Victoria (Trotter 1996) offered training to 12 community corrections officers (community supervisors of adult offenders on court and parole board orders) in role clarification, problem solving, pro-social modelling and relationship skills. Those who received training were significantly more likely to make use of skills (as determined by examination of file notes) and had significantly lower re-offence rates in their client groups at 18 months and four year follow-up by comparison with a control group of clients of officers who had not received training. There were a total of 261 clients in the study and the differences in recidivism between the two groups were between 20 and 45 percent, depending on the recidivism measure used. In a further study using the same population (Trotter 2000), it was found that supervisors who had academic qualifications in social work or welfare made more use of the effective practice skills and their clients had lower recidivism rates at statistically significant levels compared with the clients of workers without training or trained in other disciplines.

A study undertaken in adult probation in Canada (Bonta et al. 2011) followed up 143 clients, supervised either by one of 33 supervisors who had undertaken training in effective practice skills, or by one of 19 supervisors who had not undertaken such training. The training course followed collaboration with staff involved in the earlier Australian study (Trotter 1996) and included similar content, although the Canadian training placed greater emphasis on cognitive behavioural interventions. The workers who volunteered for the project then provided tape recordings of worker/client interviews, which were analysed by researchers using a detailed coding manual to assess the skills of the workers. The researchers reported high levels of inter-rater reliability and found that those who had received training made more use of each of the practice skills of structuring, relationship, behavioural techniques and cognitive techniques. They also found that those under the supervision of the trained officers had significantly lower rates of recidivism after controlling for client risk levels. Thirty-three percent committed a further offence within two years in the non-trained control group compared with 24 percent for clients of those who were trained. The clients of those officers who made maximum use of ongoing clinical supervision had a recidivism rate of only 15 percent. It was found that the most influential skills in terms of reduced offending were those relating to cognitive behavioural principles by contrast with relationship and other skills which were not significantly related to recidivism.

A larger study with similar aims was conducted by Robinson et al. (2011), with US courts and the Middle District of North Carolina. They used a sample of 41 officers in the trained experimental group and 26 in the control group, and a total of more than 1,000 pre-trial and post-conviction adult clients. They found that those in the experimental group who undertook three and a half days training in effective supervision skills were more likely to use the effective practice skills when tapes of interviews were analysed. The researchers used a coding manual that was developed following consultation with researchers in the earlier studies. The training and the training manual focused on ‘Active Listening, Role Clarification, Effective Use of Authority, Effective Disapproval, Effective Reinforcement, Effective Punishment, Problem Solving and Teaching, Applying and Reviewing the Cognitive Model’ (Robinson et al. 2011: 59).

The offenders supervised by officers in the experimental group had a re-offence rate of 26 percent compared with 34 percent in the control group for moderate to high-risk clients. This study, like the earlier ones, was conducted with volunteers and experienced some methodological problems in terms of dropouts, however, the experimental and control groups showed no significant differences in terms of demographics or risk levels.

The evidence is mounting therefore that the skills offered by supervisors to offenders sentenced to community supervision make a difference to the reoffending rates of the offenders. The studies undertaken to date have focused on adult offenders in Australia, the United States and Canada. No studies of this type have focused specifically on young offenders in Australia or elsewhere. Further, the studies conducted to date have used analysis of file notes or coding of audiotapes of interviews to determine the nature of skills used by workers. None have observed interviews in person—a method that might provide information about non-verbal interactions. The various studies have also used different definitions of skills. In addition, none of the studies have considered the relationship between levels of education of staff in counselling skills or their specific designated role as a counsellor versus worker. There is also some uncertainty as to which aspects of the effective practice principles have the most impact, with one earlier Australian study (Trotter 1996) pointing to pro-social modelling and reinforcement as most strongly related to positive outcomes, by contrast with the Canadian study (Bonta et al. 2011), which found the use of cognitive techniques to be the most influential. Some of these issues are addressed in this study.



  • To examine the relationship between the use of effective practice skills by supervisors in Juvenile Justice New South Wales and re-offence rates by clients.
  • To consider which of the skills, if any, have the most impact on reoffending rates.


This study was undertaken in the Department of Juvenile Justice in New South Wales. Community-based supervision represents the primary form of intervention with young offenders in New South Wales, as well as throughout Australia (AIHW 2011). In Australia, in 2008–09, around 7,200 young people were under juvenile justice supervision on any given day. Most (90%) were under community-based supervision, with the reminder in detention. In New South Wales during 2009–10, 4,521 young offenders were under the supervision of the NSW Department of Juvenile Justice on community-based orders (NSW Government 2010). Aboriginal and Torres Strait Islander young people continue to be overrepresented in community supervision, as well as in detention (AIHW 2011).

After receiving university and NSW Department of Juvenile Justice ethics approval, juvenile justice staff with responsibility for direct supervision of young offenders were invited to be involved in the project. Forty-eight staff members initially volunteered. For each worker, the next five clients allocated to them became eligible for the study. Interviews between the staff members and young people were then observed within three months of the young person receiving their court order. In total, 117 interviews were observed, conducted by 46 workers over a period of four years; 39 in a pilot project and the remainder as part of a project funded through a Criminology Research Council (now known as the Criminology Research Grants program) grant. In some cases, five interviews were observed for each worker, however, in other cases, only one or two interviews could be observed due to staff and client turnover. All interviews were with different clients.

Fifteen Juvenile Justice Counsellors were involved in 32 of the interviews and 33 Juvenile Justice Officers conducted the remaining 84 interviews. Juvenile Justice Counsellors and Juvenile Justice Officers each provide direct supervision to young offenders on probation, parole or other community-based supervision orders. Juvenile Justice Counsellors have relevant tertiary qualifications and have a counselling or problem-solving role, whereas Juvenile Justice Officers are not required to have tertiary qualifications and are generally expected to focus more on compliance and practical issues. The officers were most often aged between 35 and 44 years, and the average years of experience working in juvenile justice was 10. Sixty-four of the interviews were conducted by female officers and 55 by male officers.

The organisation provides regular training in effective practice skills. Training in the four key skills of role clarification, pro-social modelling, problem solving and relationships has been conducted throughout the agency over the past five years, along with the introduction of and training in a cognitive behavioural method known as TARGETS or CHARTS, which involves structured methods of addressing client issues through the use of worksheets.

The 117 young people had an average age of 15.82 years, with the youngest being 12 years old and the oldest being 18 years old. Eighty-four percent (n=98/117) were male, 43 were on supervised bonds, 32 on probation, 19 on suspended sentences, 15 on parole and the remainder on other forms of supervision (eg bail supervision). The mean length of their orders was 11 months. They had an average of 1.46 prior convictions. The most common offences for which they received the current order included assaults (n=33), break and enter (n=23), robbery (n=12), theft (n=11) and property damage (n=11). Twenty-three percent (n=27/117) identified as Aboriginal.

The interviews took place at a number of venues. Thirty-one percent were undertaken in juvenile justice offices, 28 percent in clients’ homes and 40 percent in another community setting. The average time for an interview was 30.48 minutes; however, there was considerable variation in the length of interviews, with the shortest being five minutes and the longest 102 minutes.


A coding manual was developed following collaboration with other researchers on similar projects (Bonta et al. 2011; Raynor, Ugwudike & Vanstone 2010), although it was different in some respects to the coding manuals used on those projects. The manual reflected the effective practice skills referred to in the literature review and outlined in Trotter (2006). The manual was framed to define the skills and assist in the accuracy and reliability of the estimates of the extent to which the skills were used in interviews. It was divided into 15 sections including set up of the interview, structure of the interview, role clarification, needs analysis, problem solving, developing strategies, relapse prevention/cognitive behavioural techniques, pro-social modelling and reinforcement, nature of the relationship, empathy, confrontation, termination, use of referral/community resources, non-verbal cues and incidental conversations. Each of the 15 sections contained a number of items that could be rated on a five-point scale. For example, the problem solving section included problem survey, problem ranking, problem exploration, setting goals, timeframe, review, developing a contract, developing strategies, ongoing monitoring and time spent conducting problem solving.

For the skill to be rated high, it needed to be implemented in a way that was consistent with the research about good practice referred to in the literature review. For example, problem solving would be rated high if the worker frequently helped clients to identify their own problems and goals, and assisted them to identify strategies themselves in order to achieve the goals. It would be rated low if the worker identified problems with minimal input from the client and then set goals and strategies for the client.

Each of the interviews was coded either by the researcher who observed the interview or subsequently by another researcher from an audiotape of the interview. When the coding was undertaken from an audiotape, the coder also had access to a non-verbal checklist, which was completed by the observer following each interview. The non-verbal checklist provided information about the body language and non-verbal interactions between the workers/counsellors and the clients.

Inter-rater reliability

Three research officers conducted field observations. Ninety-seven observations were completed by the first research officer, who was employed continuously on the project for a period of four years. Sixteen observations were completed by an Indigenous research officer and four were completed by another research officer. The project purposefully employed an Indigenous research officer at the request of the department because of the large number of Indigenous young offenders in NSW Juvenile Justice and because it was anticipated that an Indigenous research officer might help to identify skills that were of particular relevance to Indigenous clients.

The coding was undertaken by three coders. The second and third coders did not observe the interviews but coded from the tapes and the non-verbal cues checklist. Each of the coders was trained in using the coding manual and cross-coded a number of interviews using the audiotapes of the interviews prior to doing the final coding of the interviews. Detailed discussions were undertaken to ensure that each of the coders had consistent interpretations of the wording in the coding manual.

Twenty of the interviews were coded by the research officer who observed the interview and subsequently cross-coded by another research officer using the audiotapes and the non-verbal checklist. There was a high degree of consistency in the ratings. For example, the correlation on the overall global skill score between first and second coders was .741 (p<.000), on time spent discussing role clarification it was .548 (p<.006), on time spent on problem solving it was .626 (p<.002) and on pro-social modelling .561 (p<.005).

The global score is a score out of 10 relating to the interview as a whole. If, for example, the worker did not use any of the effective practice principles in the interview, then a score of one would be allocated. If the worker showed some use of the effective practice principles, they would be given a score of five. If the worker used all of the principles in the interview, then they could be scored at 10.

For the overall global score to be rated high, the worker would display high-level skills in role clarification, pro-social modelling, problem solving and relationship, and each of the skills would be used in a collaborative way. It would be rated low if the worker did not reinforce the young person’s pro-social comments, identified problems with minimal input from the client, made no attempt to help the young person understand the purpose of supervision and if the worker appeared to be punitive or blaming.

The coders attempted to score the interviews independently of the offender’s response. In other words, if the worker was using high-level skills but the client was not engaged, the skills scores would still be high. This approach acknowledged that clients might be disengaged for reasons other than workers’ skills, such as poor language skills or earlier experiences with services.


The recidivism measure reported in this paper is any further offence, for which the young person was found guilty by a court, within two years of receiving their order. The data was taken from police records. The two year period was used based on the evidence that most young people who reoffend do so within a period of two years (Carcach & Leverett 1999). While there are obvious limitations to using only one measure, there was a significant correlation between this measure and other recidivism measures used in the study; for example, placement in detention or prison within two years (r .351 p<.000).


Use of skills by workers and client recidivism

There was a positive relationship between the linear global score on use of the skills and reduced reoffending using a comparison of means in SPSS (Global Score 6.07 for those supervisors of young people who did not reoffend and 5.56 for those who did reoffend). This was, however, outside statistical levels of significance.

Similarly, when a dichotomous score was used, it was found that workers with high global skill scores (6 or above) had clients with lower rates of reoffending after two years compared with those with low global scores (see Table 1). However, the differences were outside the .05 level of statistical significance.

Table 1: Overall use of skills by workers (scored 6 or more and 5 or less) by any further offence in 24 months by clients
  Reoffended in two years
Skills score 5 or less 74% (39/53)
Skills score 6 or more 62.5% (40/64)

There was, however, a statistically significant association between the recidivism rates of clients of supervisors who had particularly low global scores when compared with the other clients. As shown in Table 2, if the probation officers were allocated a score of less than five, their clients offended significantly more often than the clients of other probation officers. That is, those supervised by workers with more skills survived two years without reoffending at almost twice the rate of those whose workers who displayed fewer skills.

Table 2: Overall use of skills by workers (scored 5 or more and 4 or less) by any further offence in 24 months by young people
  Reoffended in two years
Skills score 4 or less 81% (26/32)
Skills score 5 or more 62% (52/85)
One tailed Fischer’s exact test p= <.04

This significant difference was also evident when a regression analysis was undertaken in SPSS, taking account of the assessments derived from the YLSI (the standard risk assessment measure used in NSW Juvenile Justice). Therefore, the differences cannot be explained by the risk levels of the clients as shown in Table 3. When Indigenous status of the client was taken into account in the regression analysis, the differences remained within significant levels, although when gender was taken into account, the p value was .07, which was slightly outside the conventional significance level of .05.

Table 3: Logistic regression analysis of overall use of skills by workers (scored 5 or more and 4 or less) by any further offence in 24 months by clients including client risk levels
    B S.E Wald df Sig. Exp(B)
Step 1a Skill score -1.088 .553 3.869 1 .049 .337
  YLSI .116 .029 15.690 1 .000 1.123
  Constant .642 1.056 .370 1 .543 1.901

a: Variable(s) entered on step 1: gs4, FR.YLSI_score

The use of other skills and client recidivism

Most of the skills coded during the observations were related to lower reoffending by clients. In most cases, however, this did not reach statistically significant levels. For example, the more time workers were involved with clients undertaking problem solving, the lower the recidivism rates of the clients. When strategies were developed in the sessions to address problems, this was related to reduced offending, particularly where the young people developed the strategies themselves, rather than the worker. In each instance, when the workers used cognitive behavioural skills and relapse prevention skills, the clients had lower rates of recidivism. Workers who used pro-social modelling and relationship skills also had clients with lower reoffending rates. Workers who were rated as open and honest, non-blaming, optimistic, enthusiastic, used appropriate self-disclosure and who were friendly all had clients with lower reoffending.

None of these measures, however, reached conventional levels of statistical significance. The two worker skills that were most strongly related to reduced further offending were the use of rewards by the worker and a non-blaming approach. Fifty-five percent (n=27/49) of the young people reoffended when the use of rewards by the worker was scored three or more on the five point scale, compared with 76 percent (n=52/68) when the use of rewards was scored two or less. This was significant at the p<.05 level on a chi square analysis, although it was not statistically significant when the YLSI was included in a regression analysis. High scores were given on this item if the worker identified the client’s pro-social actions and comments (victim empathy for example) and praised the client or provided concrete rewards (reduced frequency of appointments for example) in response to the pro-social actions and comments of the client.

Those who scored above two on the five point scale for the non-blaming measure had a recidivism rate of 61 percent (n=40/66) compared with 76.5 percent (n=39/51) for those who scored two or below. This was just outside the conventional significance level (p=.056) after taking risk into account in the regression analysis. The non-blaming item was scored high if the worker explored the young person’s circumstances (work, family or drug use for example) and did not inadvertently or directly attribute blame to the young person for their situation.

Staff role and qualifications

As mentioned earlier, Juvenile Justice Workers may be employed as Juvenile Justice Counsellors or as Juvenile Justice Officers. The counsellors made more use of the effective practice skills, being twice as likely to be rated above five on the global score (45%, n=38/84 for workers and 80%, n=26/33 for counsellors; p<.01).

They also had clients with lower recidivism (54.5%, n=18/33 for counsellors, compared with 73%, n=61/84 for officers), yet the counsellors supervised clients who were higher risk (ie scored higher on the LSIR). The differences in recidivism between the groups was statistically significant (p<.05) after taking risk levels into account through the regression analysis as shown in Table 4. The statistically significant differences between the staff groups remained if other variables such as client gender or Indigenous status were included in the analysis.

Table 4: Regression analysis of staff position (JJ Officer or JJ Counsellor) client risk level and any further offence in two years
    B S.E Wald df Sig. Exp(B)
Step 1a FR.YLSI_score .126 .030 17.354 1 .000 1.134
  Staff_position -1.269 .501 6.403 1 .011 .281
  Constant .205 .726 .080 1 .788 1.227

a: Variable(s) entered on step 1: FR.YLSI_score, Staff_position


A growing number of studies have indicated that correctional interventions characterised by certain features lead to reduced levels of recidivism. More specifically, a small group of studies has found that the nature of the skills used by supervisors in the routine supervision of offenders on community-based orders relates to the recidivism of offenders under supervision (Bonta et al. 2011; Robinson et al. 2011; Trotter 1996). These studies have also shown that supervisors are more likely to use these skills if they have received specific training and supervision based on the skills. One study (Trotter 2000) also found that supervisors who had completed social work and welfare qualifications (courses in which these skills are commonly taught) were more likely to use the skills and more likely to have clients with lower recidivism. There is some doubt, however, about the precise nature of the effective practice skills, with some studies supporting the importance of cognitive behavioural skills and others placing more emphasis on relationship skills (Bonta et al. 2011; Trotter 1996; Ward 2010). The studies, to date, have also focused on adult offenders rather than young offenders and they have used various methods to assess the use of skills.

This study aimed to examine the relationship between the use of skills in youth justice and client recidivism using observation of interviews and a detailed coding manual. It was found that when the observers rated the workers as having good skills, the clients had fewer further offences in the two year follow-up period, although this only reached statistically significant levels when the clients of workers who were rated as making minimal use of the skills were compared with the clients of other workers. The workers with few skills seemed to be particularly ineffective.

The study found the skills with the strongest associations with reoffending were the use of rewards and a non-blaming approach. The relationship between the use of rewards by supervisors and low reoffending is consistent with earlier research on pro-social modelling and reinforcement (Trotter 1996), and with the various ‘what works’ meta-analyses (Bonta & Andrews 2010). It seems clear that it is a core skill in work with offenders.

Similarly, the importance of the client–worker relationship and of collaboration between worker and client (arguably the antithesis of a blaming approach) has been consistently emphasised in the research on effective supervision (Andrews & Bonta 2008; McIvor & Raynor 2008; McNeill et al. 2005). Again, the ability to be non-blaming seems to be a core skill in effective offender supervision.

It is interesting that when client gender was included in the regression analysis, the relationship between the global skills score and reoffending was slightly weaker. The numbers are too small to conclude that young women did not benefit from the skills in the same way as young men, even though previous studies have suggested that women may benefit from different skills to men (Trotter, McIvor & Sheehan 2012). More work is needed with larger samples to tease out the impact of different skills with different offender populations.

This study has a number of limitations. In particular, the sample was dependent on volunteers. It may be that those workers who volunteered were more skilled than other workers. Also, workers may have behaved under observation in a different way to the way they would usually behave. They might have used more skills under observation than they would usually use. The requirement to undertake research under ethical guidelines makes it difficult to control for these factors.

Nevertheless, when the results of this study are considered alongside the earlier research, it does support the hypothesis that workers with certain skills will have clients with lower recidivism and that the extent to which those skills are used by individual workers can be determined through a process of observation. It also provides support for the view that workers who have qualifications in social work or psychology and who are given a counselling role are likely to make more use of effective practice skills and have clients with lower recidivism.

These findings have implications for selection, training and the roles of youth justice staff. There is potential for widespread reductions in recidivism if juvenile justice organisations prescribe a counselling role to supervisors and employ staff with relevant qualifications. Ongoing training and supervision focused on effective practice skills may provide for further reductions in reoffending. The benefits are likely to be further increased if training and supervision is accompanied by regular observation and analysis of interviews between juvenile justice workers and their supervisors, and with feedback, discussion and coaching in order to provide for ongoing skill development.


The assistance of Phillipa Evans and Tamara Saunders who undertook the data collection is acknowledged, along with the support provided by Juvenile Justice NSW and the Australian Criminology Research Council.


All URLs correct at November 2012

  • Australian Institute of Health and Welfare (AIHW) 2011. Australia’s welfare 2011. Canberra: AIHW. http://www.aihw.gov.au/publication-detail/?id=10737420537
  • Andrews DA & Bonta J 2008. The psychology of criminal conduct. Cincinnati: Anderson Publishing
  • Andrews DA, Keissling JJ, Russell RJ & Grant BA 1979. Volunteers and the one-to-one supervision of adult probationers. Toronto: Ontario Ministry of Correctional Services
  • Bonta J & Andrews D 2010. Viewing offender assessment and rehabilitation through the lens of the risk-needs-responsivity model, in McNeil F, Raynor P & Trotter C (eds), Offender supervision: New directions in theory, research and practice. Devon: Willan Publishers
  • Bonta J et al. 2011. An experimental demonstration of training probation officers in evidence-based community supervision. Criminal Justice and Behavior 38(11): 1127–1148
  • Carcach C & Leverett S 1999. Recidivism among juvenile offenders: An analysis of times to reappearance in court. Research and Public Policy series no. 17. Canberra: Australian Institute of Criminology. https://www.aic.gov.au/publications/rpp/rpp17
  • Dowden C & Andrews DA 2004. The importance of staff practice in delivering effective correctional treatment: A meta-analytic review of the literature. International Journal of Offender Therapy And Comparative Criminology 48(2): 203–214
  • McIvor G &Raynor P (eds) 2008. Developments in social work with offenders. London: Jessica Kingsley
  • NSW Government. Juvenile Justice 2010. Annual report 2009–10. http://www.djj.nsw.gov.au/publications.htm
  • Raynor P, Ugwudike P & Vanstone M 2010. Skills and strategies in probation supervision: The Jersey study, in McNeil F, Raynor P & Trotter C (eds), Offender supervision: New directions in theory, research and practice. Devon: Willan Publishers
  • Robinson C, Vanbenschoten S, Alexander M & Lowenkamp C 2011. A random (almost) study of staff training aimed at reducing re-arrest (STARR): Reducing recidivism through intentional design. Federal Probation 75(2): 57–63
  • Trotter C 2006. Working with involuntary clients. Sydney: Allen & Unwin
  • Trotter C 2000. Social work education, pro-social orientation and effective probation practice. Probation Journal 47: 256–261
  • Trotter C 1996. The impact of different supervision practices in community corrections. Australian and New Zealand Journal of Criminology 29(1): 29–46
  • Trotter C McIvor G & Sheehan R 2012. The effectiveness of support and rehabilitation services for women. Australian Social Work 65(1): 6–20
  • Ward T 2010. The good lives model of offender rehabilitation, in McNeil F, Raynor P & Trotter C (eds), Offender supervision: New directions in theory, research and practice. Devon: Willan
  • Ward T, Melser J & Yates PM 2007. Reconstructing the risk-need-responsivity model: A theoretical elaboration and evaluation. Aggression and Violent Behavior 12: 208–228

About the Authors

Christopher Trotter is Professor in the Social Work Department and Director of the Criminal Justice Research Consortium, Monash University.