Can criminal justice decision-makers be relied on to get it right every time?

The recent sentencing in Queensland of a convicted paedophile to an indefinite sentence under the Penalties and Sentences Act (1992) (pdf, 743 kb), and the present attempt by the Queensland Government to keep a convicted rapist in prison past his release date under the Dangerous Prisoners (Sexual Offenders) Act (2003) (pdf, 520 kb), will no doubt be welcomed by many in the community. Certainly, these proceedings have found broad support in the media, or were at least reported without much in the way of serious analysis or criticism. Both individuals concerned were convicted of despicable crimes and there will be little sympathy for their personal fates. However, the disgust that we may feel towards these two offenders should not prevent us from questioning the rational and ethical bases of their sentences.

The principal rationale for imposing an indefinite sentence - as opposed to giving a fixed-term sentence - is protection of the community. It is believed that there are some individuals who present such a danger to the community that they must be kept in prison until their risk of reoffending no longer exists. It is argued that just when the individual will be safe to release - if ever - cannot be properly determined at the time of sentencing. So, the release date is left open, to be decided by experts at a later time. The offender's progress in custody can be monitored and they will be released only when they are declared to no longer present a serious danger to the community. The principle of indefinite sentencing is consistent with the criminal justice goal of incapacitation. However, there are two problems.

First, an indefinite sentence violates the just-deserts principle that a person should receive a punishment that is proportional to the harm they have done. This principle is in turn based on the fundamental premise that a person should only be sentenced for the crimes that they have committed, not for the crimes they may commit. Indefinite sentences, on the other hand, involve a punishment that varies according to a person's post-offence behaviour. Thus two offenders committing similar crimes may serve vastly disparate sentences. Keeping a person past their release date is in effect punishing them, not for their crimes, but for their failure to rehabilitate. Moreover, their continued incarceration represents a punishment for crimes they have not yet committed.

The second problem with indefinite sentences is a practical one, though with serious ethical implications. Indefinite sentences rest on the assumption that it is possible to distinguish reliably between those who will and those who will not be a danger to the community in the future. In fact, the ability to accurately predict recidivism is a skill that mental health professionals conspicuously lack. The research is clear: psychologists and psychiatrists - particularly when they are relying solely on their clinical judgement - get it wrong almost as often as they get it right. The task becomes even more difficult when predictions about post-release behaviour are made on the basis of prison behaviour - how a person behaves in the highly structured and artificial prison environment may say very little about how they will behave in the "real" world.

Criminal justice decision-makers are particularly prone to false positive predictions, that is, predicting that someone will reoffend when they do not. This is due in part to an understandable conservatism in decision-making - the desire to err on the side of caution. No-one wants to be seen to be responsible for releasing a prisoner who goes on commit further serious crimes. More generally it reflects an overestimation by many practitioners of the recidivism rates of serious offenders. In the case of sex offenders, in particular, there is a widespread assumption that reoffending is inevitable. However, there is now a great deal of evidence to show that, once caught, most sex offenders do not go on to be reconvicted of new sex offences.

There have been two important findings that consistently emerge from studies of sex offender recidivism. First, the majority of sex offenders are found to be in the lower risk categories - relatively few imprisoned sex offenders (about 10 per cent) can be categorised as a high recidivism risk. Second, of these high-risk offenders, just over half are found to be reconvicted of new sexual offences within 15 years of their release from prison. Thus for every 100 incarcerated sex offenders that we could justify categorising as high risk, about 48 will not be reconvicted for new sexual offences after 15 years.

The well-documented inability of mental health professions to predict an offender's potential danger to the community does not stop them from making confident assertions in court and to parole boards, nor does it stop courts and parole boards from continuing to seek expert opinion on the matter. Eliminating prediction error cannot be achieved simply through better training. A large part of the problem lies in the nature of offending behaviour itself. People offend not just because of who they are (dispositional factors), but their offending also depends upon the post-release circumstances (situational factors). The use of standardised prediction instruments (so-called actuarial methods) to supplement clinical judgement can improve predictions. But even if accuracy rates were increased to, say, a respectable 75 per cent, one in four offenders would still misclassified.

Balancing the rights of individual offenders on the one hand and community safety on the other is a tricky business. Indefinite prison sentences for dangerous offenders, certain to reoffend, violates the principle of just-deserts, but arguably can be justified in terms of the criminal justice goal of incapacitation. However, the issue becomes even more problematic if there is no accurate basis for distinguishing those who represent a serious danger to the community, and those who do not.

Both the indefinite sentencing provisions of the Penalties and Sentences Act (1992) and the post-sentence detention provisions of the new Dangerous Prisoners (Sexual Offenders) Act (2003) require a court to be satisfied to a "high degree of probability" that an offender represents a "serious danger to the community". Setting such a high threshold of probability is no doubt a reflection of the government's aim to protect against the arbitrary imposition of indefinite detention. However, in setting the threshold so high, the legislation has perhaps inadvertently relied on the ability of professionals to provide advice that is currently beyond their level of knowledge and expertise.