What works in young offender treatment: A meta-analysis

Tags: responsivity, young offenders, effect size, meta-analysis, human service, correctional treatment, D. A. Andrews1 Department of Psychology, the principles, young offender treatment, human service programs, Inappropriate Service, Andrews, correlation coefficient, recidivism rates, BESD, recidivism rate, Carleton University, Phi coefficient, Craig Dowden, rehabilitation programs, Reducing Reoffending, D. A. Andrews, Criminal Justice and Behavior, P. Gendreau, juvenile delinquents, R. L. Izzo, J. Bonta, J. McGuire, Chichester, Psychology Department, American Society of Criminology, C. Dowden, R. R. Ross, John Wiley & Sons
Content: What works in young offender treatment: A meta-analysis by Craig Dowden and D. A. Andrews1 Department of Psychology, Carleton University Several meta-analytic reviews strongly support the Sclinically relevant and psychologically informed principles of human service, risk, need and general responsivity. More recently, meta-analyses have demonstrated that these principles are applicable to female offenders2 and are effective in reducing both general3 and violent4 recidivism. The current investigation provides an in-depth examination of the principles of human service, risk, need and general responsivity for Young Offenders (younger than 18 years). Further analyses are conducted on the "more promising" and "less promising" treatment targets outlined by Andrews and Bonta.5 The results demonstrate that the mean effect size under conditions of adherence to each of the principles is significantly higher than for conditions of non-adherence. These results have important implications for both correctional administrators and front-line staff involved in delivering correctional treatment programs to young offenders. Introduction Several meta-analyses have revealed that correctional treatment programs have been effective for young offenders.6 Andrews, Zinger, Hoge, Bonta, Gendreau and Cullen7 conducted one of the most influential meta-analyses that presented the characteristics of the most effective correctional programs for both adult and juvenile offenders. They presented evidence that programs that adhere to the principles of risk, need and responsivity yield the largest reductions in reoffending. However, their paper did not have separate tests for the principles of risk and need for the entire sample of studies. Therefore, the purpose of this paper is to conduct a meta-analysis on an expanded sample of studies using updated and more systematic coding procedures to explore the importance of the principles of risk, need and responsivity in delivering effective correctional treatment for young offenders. Methodology Sample of studies: This study used the two samples of studies reported by Andrews, Dowden and Gendreau.8 The first sample (k = 131) contained the juvenile offender studies used in the Andrews, Zinger, Hoge, Bonta, Gendreau and Cullen meta-analysis. The second sample (k = 98) included additional studies collected by Andrews and his colleagues after the publication of their 1990 paper, as well as studies gathered by Dowden.9 Procedure: The coding manual used for the present study incorporated items taken directly from Andrews and colleagues, several items introduced by Lipsey,10 as well as new variables introduced by Dowden. The measure of interrater reliability was determined by dividing the total number of correct classifications by the total number of coding classifications. The rates of agreement for the four main variables introduced in this meta-analysis were 100% (Any Treatment, r = 1.00) and 90% for each of
the remaining variables (Risk, Need and Responsivity, r = .79). The interrater agreement was 76% (r = .88) for the four-level Type of Treatment variable.
The measure of effect size used for this report was the Pearson Product Moment correlation coefficient and, more specifically, the Phi coefficient. The Phi coefficient was used because it can be readily translated into the binomial effect size display (BESD).11 The BESD converts the Phi coefficient into a value that reflects the simple difference between the recidivism rates of the treatment and control groups. A correlation coefficient of .30, for example, translates into a recidivism rate of 35% for the treatment group and a recidivism rate of 65% for the control group (i.e., .30 becomes a 30 percentage point difference).
Overall results
The meta-analysis yielded 229 tests of the effectiveness of correctional treatment from 134 primary studies. Approximately 84% of the studies were composed predominantly or entirely of male offenders.
The overall mean effect size for the sample was +0.09 with a 95% confidence interval of +0.07 to +0.12. These results suggested that the effects of correctional interventions were mildly positive. Using the BESD, this value represented a recidivism rate of 45.5% for the intervention group and a 54.5% recidivism rate in the control group.
Further exploration of the data revealed that considerable variability existed within the effect sizes (from -.43 to +.83, SD = .21). Not surprisingly, the type of correctional intervention accounted for some of this variability. For example, the mean effect size for interventions based solely on criminal sanctions was -.02 (n = 54) compared with a significantly different mean effect size of +0.13 (n = 175) for human service programs, F = 23.47 (n = 1,227), p < .001, measure of association Eta = .31.
Clearly, the introduction of human service within a justice context is associated with strong reductions in the reoffending levels of young offenders. However, separate analyses were conducted on the principles of risk, need and responsivity to determine their relationship with reduced recidivism.
Table 1
Mean Effect Sizes and Number of Contributing Tests of Treatment for the Principles of Human Service, Risk, Need and Responsivity
Variable label
Adheres to principle
No
Yes
Eta
Human service
-0.02 (54) 0.13 (175) 0.31***
Risk
0.03 (61) 0.12 (168) 0.20**
Criminogenic need
-0.01 (126) 0.22 (103) 0.55***
General responsivity: Behavioural
0.04 (169) 0.24 (60) 0.42***
**p < 0.05; ***p < 0.001 Risk, need and responsivity Both the within-sample and aggregate-sample approaches to coding risk were used. Note that the aggregate approach was used only when a primary study failed to differentiate the risk level of their clients. In the aggregate approach, a study was coded as high risk if the majority of its offenders had formally penetrated the judicial system at the time of the study and/or had a prior criminal record. The meta-analysis supported the risk principle of case classification because correctional interventions were associated with a significantly higher mean effect size when delivered to higher-risk (+.12) versus lowerrisk (+.03) offenders, F = 9.04 (n = 1,227), p < .01 (see Table 1). General responsivity was coded, in the same way used by Andrews and colleagues (1990), as being met if the program was behavioural or used several treatment methods such as modelling, graduated practice, role-playing and several other skill-building techniques. The results revealed that for young offenders, the mean effect size for behavioural programs (+.24, k = 60) was significantly larger than the mean effect size for non-behavioural programs (+.04, k = 169), F = 47.73 (n = 1,227), p < .0001 (see Table 1). Programs were coded as appropriately adhering to the need principle if the majority of the treatment targets within the program were criminogenic needs. Programs that targeted an equal or greater number of noncriminogenic needs were coded as inappropriately adhering to the need principle. Programs that had appropriately addressed the need principle yielded a significantly larger mean effect size (.22; k = 103) than programs that did not (-.01; k = 126), F = 98.52 (n =1,227), p < .0001. Type of treatment The new approach to coding the Type of Treatment variable introduced by Andrews, Dowden and Gendreau was used. A simple count was conducted on the number of the principles of risk, need and responsivity that were appropriately addressed within the program and the coding was assigned based on this score. Criminal sanctioning approaches, however, were automatically placed in the Inappropriate Service category. An Analysis of Variance revealed significant differences between the different levels of this variable, F = 41.56 (n = 3,225), p < .001, Eta = .60. Follow-up contrasts using the Scheffe correction demonstrated that Most Promising Service (.28; k = 44) yielded a significantly larger mean effect size than each of the remaining categories (p < .05). In addition, the Promising Service category (.21; k =44) was associated with a significantly higher mean effect size than either the Weak Service (.08; k = 111) or Inappropriate Service (-.04; k = 30) categories, (p < .05). The Weak and Inappropriate Service categories were statistically indistinguishable. These findings demonstrate that the clinically relevant and psychologically informed principles of human service, risk, need and responsivity are key determinants
of the therapeutic potential of a treatment program. Criminogenic versus noncriminogenic needs Table 2 lists the Percentage Distributions for the most frequently targeted criminogenic needs, as well as the mean effect size for each need when it was and was not targeted in a particular program and its corresponding relationship with effect size; Table 3 lists these items for noncriminogenic needs. Inspection of Table 2 reveals that each of the criminogenic needs targeted in treatment was associated with a positive mean effect size. Clearly, criminogenic needs are the key when developing effective correctional treatment programs.
Table 2
Criminogenic Needs Targeted: Rank Ordered by
Frequency and Their Correlation with Effect Size
Targeted need
Frequency r
Academic
51 0.23***
Other criminogenic needs
47 0.36***
Anger/antisocial feelings
41 0.28***
Self-control
40 0.29***
Family: affection
24 0.33***
Pro-Social model
19 0.19**
Antisocial attitudes
17 0.13*
Family: Supervision
17 0.35***
Vocational skills
17
0.09
Barriers to treatment
12 0.30***
substance abuse treatment: Any 11
0.04
Vocational skills + job
9 0.26***
Reduce antisocial peers
8
0.11
Relapse prevention
7
0.07
* p <0 .05; **p < 0.01; ***p < 0.001
Inspection of Table 3 reveals that each of the noncriminogenic needs were negatively associated with effect size. In other words, targeting these needs in correctional treatment programs was associated with increased recidivism in the intervention group. Programs that used a "fear of official punishment" approach (i.e., shock incarceration), in particular, yielded a significant negative relationship with effect size. Table 3
Noncriminogenic Needs Targeted: Frequency and Correlation with Effect Size
Targeted need
Frequency r
Vague emotional/personal problems 59 -0.06
physical activity
36 -0.03
Family: Other interventions
22 -0.11
Fear of official punishment
15 -0.18**
Increase cohesive antisocial peers
15 -0.12
Target self-esteem
14 -0.09
Increase conventional ambition
12 -0.00
Respect antisocial thinking
7
-0.05
*p < 0.05; **p <0.01; ***p < 0.001
Conclusion
This meta-analysis provides strong empirical support for the applicability of the principles of human service, risk, need and responsivity for young offenders. In addition, increased adherence to these principles is associated with increased reductions in reoffending. These findings suggest that the clinically relevant and psychologically informed approaches to reducing recidivism, outlined by many of the scholars of the rehabilitation literature, are indeed effective for young offender populations.
1. 1125 Colonel By Drive, Ottawa, Ontario K1S 5B6. 2. C. Dowden and D. A. Andrews, "What works for female offenders: A meta-analytic review," Crime and Delinquency (in press). 3. D. A. Andrews, C. Dowden and P. Gendreau, "Clinically relevant and psychologically informed approaches to reduced reoffending: A meta-analytic study of human service, risk, need, responsivity and other concerns in justice contexts," Criminology (under review). 4. C. Dowden and D. A. Andrews, "Effective correctional treatment and violent reoffending: What works!" Canadian Journal of Criminology (under review). 5. D. A. Andrews and J. Bonta, The Psychology of Criminal Conduct (Cincinnati, OH: Anderson Publishing Co., 1998). 6. C. J. Garrett, "Effects of residential treatment of adjudicated delinquents: A meta-analysis," Journal of Research in Crime and Delinquency, 22 (1985): 287­308. See also M. W. Lipsey "What do we learn from 400 research studies on the effectiveness of treatment with juvenile delinquents?" What Works:
Reducing Reoffending, J. McGuire, Ed. (Chichester, UK: John Wiley & Sons, 1995): 63­78 and R. L. Izzo and R. R. Ross, "A meta-analysis of rehabilitation programs for juvenile delinquents: A brief report," criminal justice and Behavior, 17 (1990): 134­142. 7. D. A. Andrews, I. Zinger, R. D. Hoge, J. Bonta, P. Gendreau and F. T. Cullen, "Does correctional treatment work? A clinically relevant and psychologically informed meta-analysis," Criminology, 28 (1990): 369­404. 8. D. A. Andrews, C. Dowden and P. Gendreau, "Clinically relevant and psychologically informed approaches to reduced reoffending: A meta-analytic study of human service, risk, need, responsivity and other concerns in justice contexts." 9. C. Dowden, A Meta-Analytic Examination of the Risk, Need and Responsivity Principles and their Importance Within the Rehabilitation Debate, unpublished M.A. thesis (Ottawa, ON: Psychology Department, Carleton University, 1998). 10. M. W. Lipsey, "The efficacy of intervention for juvenile delinquency: Results from 400 studies," paper presented at the 41s tannual meeting of the American society of Criminology (Reno, NV: 1989). 11. R. Rosenthal, Meta-analytic Procedures for Social Research (Newbury Park, CA: Sage Publications, 1991).

File: what-works-in-young-offender-treatment-a-meta-analysis.pdf
Title: CSC Forum - May 1999, Volume 11, Number 2
Published: Mon Apr 5 08:56:00 2004
Pages: 6
File size: 0.03 Mb

Copyright © 2018 doc.uments.com