1. The last couple of decades have seen numerous studies reporting offender reconviction rates. They provide much empirical data on which to base rational decisions about delivery of psychological treatment resources.

  2. Definition of terms

    Three specific terms used in this report are defined as follows:

    General base rates refers to the number convicted in any one year of a particular offence as a proportion of the total number convicted for the year (excluding traffic offences) eg. the proportion of people convicted for property offences is higher than for violent offences. Even within major categories such as property, violent and sexual offences, general base rates vary by sub-categories eg. within the category 'property offences,' the proportion of people convicted of theft is much higher than for arson.

    Reconviction base rates refers to the probability of a person being reconvicted for the same type of offence. The rates vary for different offences eg. reconviction base rates are higher for property offences than for violent offences. Within major categories such as property, violence and sexual offences, reconviction base rates vary by subcategories of those offences.

    Reconviction probability refers to the probability an offender will be reconvicted for any type of offence.

    The greatest reduction in reconvictions will be achieved by targeting psychological treatment to those committing offences with high general base-rates and high reconviction base rates. But since this would direct treatment almost exclusively to property offenders, other factors must over-ride such a strategy. Some crimes are so serious and costly in social, personal and economic terms, they must take precedence over other offences. This will be dealt with under the heading Special Units.

  3. Probability of reconviction

    Numerous articles in the international literature report reconviction rates following a community-based sentence or release from prison. They vary widely: some USA studies cover felony convictions only, while others include felony and misdemeanour; follow-up periods range from a few months to several years; some studies of community-based sentences confine follow-up to the term of the sentence; some report separately on those participating in 'rehabilitation' programmes and those not, while others make no such distinction; some report on frequency and seriousness of reconviction, others do not; some report reconvictions for specific offence categories, and for offender characteristics such as previous offending, age and race, and others omit this information. Not surprisingly, there is considerable variation in reported reconviction rates (Maltz 1984).

    Despite these variations, it is possible to conclude that in the USA, UK and Australia, around 60 percent of those given a community-based sentence and 70 percent of those released from prison will be reconvicted within two years, counting reconvictions for any offence (Oldfield 1997, Morgan 1993, Crosland 1995, Burgoyne 1979 a,b,c, Mair 1993, Feder 1991, Broadhurst & Maller 1990, 1992, Farrington 1986, Blumstein & Cohen 1987, Freeman 1996, Gendreau et al 1996; see also references for treatment effectiveness, as such studies include recidivism rates for non-treated control groups).

    New Zealand figures are fairly consistent with the above. Community Corrections has a reconviction rate of about 64 percent, and Public Prisons about 67 percent (Mulligan 1991, Forsythe & Love 1983, Oxley 1979, Asher 1985, Rush & Love 1982, Killalea & Piesse 1967, Bakker & Riley 1994, Lovell & Norris 1990, Bakker et al 1995 a,b).

    Psychological treatment resources clearly cannot cope with 60 to 70 percent of the total corrections population. But reconviction alone is a crude measure and should be qualified by frequency and seriousness of reconviction. Unfortunately most reconviction rate studies lack this information. A conservative conclusion can be drawn from those that do provide it: about 40 percent of those reconvicted are reconvicted once only and often for a minor offence (Rush & Love 1982, Forsythe & Love 1981, Broadhurst & Maller 1990, 92, Burgoyne 1979, Oldfield 1997).

    Taken overall, this information suggests about a quarter to a third of the corrections population at any time comprises offenders who repeatedly reoffend, often seriously. There is some evidence these offenders commit most crime. Changing their behaviour will bring about the greatest reduction in reoffending.

  4. Predicting reconvictions

    Serious repetitive offenders must be identified before resources can be directed to them, but predicting reconvictions is an inexact science. No technique can identify precisely who, in the total offender population, will be reconvicted. Nevertheless, the Psychological Service Prediction Rating Scale gives an accurate probability of reconviction for individual offenders. Fully implementing this scale in the Department of Corrections, as part of the Integrated Offender Management System, along with the Serious Young Offenders project, the Maori Reference Group, and other departmental initiatives, should increase prediction power in the medium-term.

    A sensible short-term strategy would be to take stock of total available psychological treatment resources and calculate how many people they can cope with. The Prediction Rating Scale will indicate the probability level above which this number falls, allowing treatment resources to be directed to all above this cut-off point. So, if psychological treatment can be offered to 2,000 offenders and the Prediction Rating Scale shows 2,000 people score 60 or higher, then those scoring 60 or higher become the target group for treatment, unless exceptional reasons dictate otherwise. Possible reasons those scoring below the cut-off point might still receive treatment, and those scoring above it might not, are discussed later in this paper.

  5. Reconviction and setting

    Setting here refers to community-based, whether residential or not, and prison.

    On the face of it, people who offend frequently and seriously and are highly likely to be reconvicted, are more likely to be in prison than on a community-based sentence. There are little data on the number or percentage of high-risk offenders serving community-based sentences, compared to those serving prison sentences, but this distinction is important in deciding if one setting should take precedence over the other for treatment. The issue has not been addressed in the international literature.

    New Zealand data suggest a significant proportion of offenders serving community-based sentences offend as frequently and seriously as those serving prison sentences. The average offence seriousness rating for people sentenced to prison in 1995 was 250, compared to average ratings of 26 for community service, 42 for periodic detention, 57 for supervision and 63 for community programmes. This suggests prisons do hold more serious offenders. (The seriousness rating scale rates offences according to how serious judges have deemed each offence to be, by way of severity of sentence over a five-year period 1990 to 1995.)

    These figures are averages, however, and closer examination shows that in 1995 the seriousness rating scale scored 37 percent of those given a community-based sentence in the range of 100 to 500 plus, and 18 percent at more than 500. Given the prison average of 250, it is clear many serving community-based sentences have offended as seriously as many in prison, or more so.

    A total of 22,450 were given a community-based sentence in 1995. Eighteen percent ie. the percentage scoring 500 plus, is 4,041. This compares to 6,234 prison sentences in the same year, a significant percentage of which must be for crimes well below a seriousness rating of 500 plus, given the 250 average prison rating (Spier 1996). Seventyeight percent of those released from prison were re-convicted within five years, with half re-incarcerated. Community-based sentence figures were: periodic detention, 84 percent reconvicted and 36 percent imprisoned; community service, 64 percent and 13 percent; supervision, 76 percent and 28 percent (Bakker & Riley 1997).

    These figures clearly contradict any suggestion that psychological treatment services be directed to a specific setting. The only conclusion to draw is that between a quarter and a third of people serving community-based sentences are serious repetitive offenders — at any time about the same number as the whole prison population.