Only 10.3% of the 6,488 juveniles who appeared in the NSW Children’s Court in 2007 were given a control order, yet 48% of the budget of the NSW Department of Juvenile Justice is spent keeping juvenile offenders in custody. To date, however, only two Australian studies have examined the effect of custodial sentences on juvenile reoffending.

Kraus (1974) matched each of 350 juveniles given a non-custodial sanction against a comparable offender given a custodial sanction. Juveniles were matched on year of birth, category of offence, age at time of first offence, number of previous (proven) offences, type of previous proven offence and number of previous custodial sanctions. He found lower rates of reoffending among vehicle thieves who received a custodial penalty, but higher rates of offending for those receiving custodial penalties in each other category of offence.

Cain (1996) examined reconviction rates among a sample of 52,935 juveniles convicted in the NSW Children’s Court between 1986 and 1994. He found that juveniles given custodial sentences were more likely to reoffend than juveniles given non-custodial sentences, but the study included no controls for prior criminal record or Indigenous status.

The Kraus (1974) and Cain (1996) studies both have limitations. Kraus (1974) was not able to control for a wide range of other factors potentially relevant to penalty choice and risk of reoffending (eg school performance, level of parental supervision, race, socioeconomic status). His methods of analysis were also relatively unsophisticated by modern standards. Cain (1996) used more sophisticated analytical methods and a much larger sample than Kraus (1974) but was similarly restricted in the range of controls he was able to use.

This study seeks to build on the work carried out by Kraus (1974) and Cain (1996) by using more sophisticated methods of analysis than Kraus (1974) and a much wider range of controls than Cain (1996). The question addressed in the study is whether, other things being equal, juveniles who receive a custodial penalty are less likely to reoffend than juveniles who receive a non-custodial custodial penalty.

The data for the current study were obtained from a longitudinal cohort study of juvenile offenders. Two groups of offenders (152 given a detention sentence, 243 given a non-custodial sentence) were interviewed at length about their family life, school performance, association with delinquent peers and substance abuse. They were then followed up to determine what proportion in each group was reconvicted of a further offence. Cox regression was used to model time to reconviction.

The study found no significant difference between juveniles given a custodial penalty and those given a non-custodial penalty in the likelihood of reconviction.

On an average day in 2006–07, 941 young people were held in detention across Australia (AIHW 2008: 51). The costs associated with juvenile detention are very high. For example, although only 10.3% of the 6,488 juveniles who appeared in the NSW Children’s Court in 2007 were given a control order, 48% of the budget of the NSW Department of Juvenile Justice is spent keeping juvenile offenders in custody (NSW Department of Juvenile Justice, personal communication 2009).

Given the high cost of juvenile detention, one would expect to find a large body of Australian research examining its potential benefits. To date, however, little research has been conducted on the effect of custodial sentences on juvenile recidivism. It is known that more than two-thirds of the young people who receive a control order from the NSW Children’s Court are convicted of a further offence within two years of their custodial order. It is not known what their reconviction rate would have been had they not received a custodial penalty. This study addresses this issue.

Conventional economic theories of crime (eg Becker 1968) contend that offenders allocate their time to legitimate and illegitimate activities according to the expected returns (ie costs and benefits) from each. A number of sociologists, however, have argued that imprisonment actually increases the risk of reoffending. There are three main variants of this argument. The first contends that prison is criminogenic because it is an environment which reinforces deviant values and which is conducive to the acquisition of new criminal skills (Clemmer 1940; Sykes 1958). The second variant contends that prison is criminogenic because it stigmatises offenders (Becker 1963; Braithwaite 1988; Lemert 1951). The third contends that prison increases the risk of reoffending because it reduces the offender’s capacity (on release) to obtain income by legitimate means (Fagan & Freeman 1999).

The Kraus (1974) and Cain (1996) studies both have limitations. Kraus (1974) made a commendable effort to match juveniles receiving custodial and non-custodial sanctions, but was not able to control for a wide range of other factors potentially relevant to penalty choice and risk of reoffending (eg school performance, level of parental supervision, race, socioeconomic status). His methods of analysis were also relatively unsophisticated by modern standards. Cain (1996) used more sophisticated analytical methods and a much larger sample than Kraus (1974) but was similarly restricted in the range of controls he was able to use.

This study seeks to build on the work carried out by Kraus (1974) and Cain (1996) by using more sophisticated methods of analysis than Kraus (1974) and a much wider range of controls than Cain (1996). The question we seek to address is whether, other things being equal, juveniles who receive a custodial penalty are less likely to reoffend than juveniles who receive a non-custodial custodial penalty. The data for the current study were obtained from a longitudinal cohort study of juvenile offenders. A sample of juvenile offenders who received custodial and non-custodial sanctions were surveyed and then followed up to determine whether, after controlling for other factors likely to influence recidivism, juvenile offenders who received control (custody) orders reoffended more quickly than juvenile offenders who received non-custodial sentences.

Fifty-two percent of the sample had a proven offence subsequent to their index sentence during the follow-up period. The mean time to reconviction (for those who were reconvicted) was 163 days (median=110 days), with a standard deviation of 178 days. Tables 2 and 3 contain descriptive statistics for variables found to have a statistically significant relationship with time to reconviction at p<0.25.

Table 4 shows the results of the Cox regression analysis. Two models are shown. Model A gives the unadjusted effect of penalty type on time to reconviction. Model B gives the adjusted effect of penalty type on time to reconviction, after controlling for number of prior court appearances. Surprisingly, this was the only factor among those listed in Table 1 that remained significant when included in the multivariate analysis with a variable measuring type of penalty imposed.

Note: The column labeled ß shows the regression coefficient associated with each variable in each model. The column labeled “SE” shows the standard error associated with the regression coefficient. The column labeled “p-value” shows the probability of obtaining the observed value of ß by chance. p-values less than .05 indicate that the variable in question is exerting a significant effect on time to reoffend. The column labeled “HR” shows the hazard ratio associated with the variable. A hazard ratio of more than one indicates that the variable in question increases the instantaneous risk of reoffending. A hazard ratio of less than one indicates that the variable in question reduces the instantaneous risk of reoffending. The final columns show the 95 percent confidence interval around the estimated hazard ratio.

The first point to note is that the hazard ratio associated with the custody variable in Model A is 1.74, which indicates that, prior to the introduction of controls, juvenile offenders given a custodial sentence are 74% more likely to be reconvicted at any given time than those who receive a non-custodial penalty. When prior criminal record is introduced into the model (see Model B), juveniles given a custodial sanction remain more likely to be reconvicted, but the hazard ratio associated with the custody variable falls from 1.74 to 1.33 and is no longer statistically significant.

Figures 1 and 2 illustrate this effect. The X axis in each figure shows free time since the index court appearance. The Y axis shows the proportion of offenders in each group who have not yet been reconvicted of a further offence. Figure 1 shows the unadjusted difference in time to reconviction between the custody and non-custody groups. Figure 2 shows the adjusted difference. It can be seen from Figure 1 that, prior to controlling for previous court appearances, the survival (non-reconviction) rate in the custodial group is substantially lower than the survival rate in the non-custodial group throughout the follow up period. The same pattern appears in Figure 2, but the differences between the groups are obviously much smaller.

Figure 4. Figure 1: Proportion not reconvicted by free time (days) since index court appearance (Model A) Custodial sentence?

Figure 5. Figure 2: Proportion not reconvicted by free time (days) since index court appearance (Model B) Custodial sentence?

The results of this study suggest that, other things being equal, juveniles given custodial orders are no less likely to reoffend than juveniles given non-custodial orders. These results are inconsistent with the two previous Australian studies on specific deterrence, both of which found evidence that juveniles given custodial penalties are more likely to be reconvicted. The difference in findings is probably due to the fact that the present study more effectively controlled for prior criminal record.

The finding that prison exerts no specific deterrent effect is consistent with overseas evidence on the specific deterrent effect of custodial penalties reviewed earlier in this article. It is important to consider, however, that the long-term effects of custodial penalties might be quite different to their short-term effects. Fagan and Freeman (1999), for example, using data from a national panel study of 5,332 randomly selected youths, found that incarceration produced a significant negative effect on future employment prospects, even after adjusting for the simultaneous effects of race, human capital and intelligence. There have been no studies on the effect of juvenile detention on juvenile employment prospects in Australia, but Hunter and Borland (1999) examined the effect of an arrest record on Indigenous employment prospects using data from the 1994 National Aboriginal and Torres Strait Islander Survey. Controlling for age, years completed at high school, post-school qualifications, whether the respondent had difficulty speaking English, alcohol consumption and whether the respondent was a member of the “stolen generation”, they found that an arrest record reduced Indigenous employment for males and females by 18.3 and 13.1 percentage points, respectively (Hunter & Borland 1999). On this basis, Hunter and Borland (1999) estimated that differences in arrest rates for Indigenous and non-Indigenous Australians might explain about 15% of the difference in levels of employment between these two groups.

These adverse effects of imprisonment on employment outcomes and the absence of strong evidence that custodial penalties act as a specific deterrent for juvenile offending suggest that custodial penalties ought to be used very sparingly with juvenile offenders. Fortunately, a range of non-custodial programs now exist which, in the United States at least, have been shown to be very effective in reducing juvenile recidivism. In the United States, they have been found to be considerably less expensive than a custodial sentence (Aos, Miller & Drake 2006). Western Australia and New South Wales are currently trialing an intensive supervision program (ISP) known in the United States as multi-systemic therapy. The NSW Bureau of Crime Statistics and Research is currently evaluating the ISP. It will be interesting to see whether it proves as effective here as it has been in the United States (MacKenzie 2002).

Aos S, Miller M & Drake E, Evidence-based public policy options to reduce future prison construction, criminal justice costs, and crime rates, Olympia, WA: Washington State Institute of Public Policy, Opympia, Washington State, 2006.

Australian Institute of Health and Welfare (AIHW) Juvenile justice in Australia 2006/7 AIHW, Canberra, 2008.

Becker GS, “Crime and punishment: an economic approach” Journal of Political Economy (1968) 76 169–217.

Becker HS, Outsiders: studies in the sociology of deviance, New York, NY: Free Press of Glencoe, New York, NY, 1963.

Braithwaite J, Crime, shame and reintegration, Cambridge University Press, Cambridge, UK, 1988.

Cain M, Recidivism of juvenile offenders in New South Wales, NSW Department of Juvenile Justice, Sydney, NSW, 1996.

Clemmer D, The prison community, Christopher, New York, NY, 1940.

Cottle CC, Lee RJ & Heilbrun K, “The prediction of criminal recidivism in juveniles: a meta-analysis” Criminal Justice and Behaviour (2001) 28(3) 367–394.

Doob A & Webster C, “Sentence severity and crime: accepting the null hypothesis” in Tonry M (ed) Crime and justice: an annual review of research: vol 30, The University of Chicago Press, Chicago, IL, 2003.

Fagan J & Freeman RB, “Crime and work” in Tonry M (ed) Crime and justice: a review of research: vol 25, University of Chicago Press, Chicago, IL, 1999, pp 225–290.

Hua J & Fitzgerald J, “Matching court records to measure re-offending” Crime and Justice Bulletin no 95, NSW Bureau of Crime Statistics and Research, Sydney, 2006.

Hunter B & Borland J, “Does crime affect employment status? The case of Indigenous Australians” Economica (1999) 67(1) 123–144.

Kraus J, “A comparison of corrective effects of probation and detention on male juvenile offenders” British Journal of Criminology (1974) 49 49–62.

Lemert EM, Social pathology McGraw-Hill, New York, NY, 1951.

MacKenzie DL, “Reducing the criminal activities of known offenders and delinquents” in Sherman LW et al (eds) Evidence-based crime prevention, Routledge, London, UK, 2002 at pp 330–404.

McGrath A, “Offenders’ perceptions of the sentencing process: a study of deterrence and stigmatization in the New South Wales Local Court” Australian and New Zealand Journal of Criminology (2009) 42(1) 24–46

Nagin D, Cullen F & Jonson C (forthcoming), “Imprisonment and reoffending” in Tonry M (ed) Crime and justice: an annual review of research: vol 38, University of Chicago Press, Chicago, IL.

Sherman L et al, Preventing crime: what works, what doesn’t, what’s promising, National Institute of Justice, Washington, DC, 1997.

Sykes G, Society of captives: a study of a maximum security prison, Princeton University Press, Princeton, NJ, 1958.

Villettaz P, Killias M & Zoder I, The effects of custodial vs non-custodial sentences on re-offending, report to the Campbell Collaboration Crime and Justice Group, Institute of Criminology and Criminal Law, University of Lausanne, Switzerland, 2006.