REPEAT IMPRISONMENT

This article is also available as a PDF - Download the PDF here

Have your say below

INTRODUCTION

Imprisonment aims to prevent crime and enhance community safety by removing offenders from the public arena and acting as a deterrent to potential offenders, as well as meeting society's need for reparation or retribution for crimes committed. (Endnote 1) However, while a period of imprisonment may deter some people from re-offending, in others it may foster further criminal behaviour. (Endnote 2)

Measuring repeat imprisonment is one way of gauging recidivism (repeated or habitual participation in crime). Studying the characteristics of people who have been imprisoned more than once, and understanding trends in criminal career development can provide valuable evidence for designing crime prevention strategies.

At 30 June 2009, there were 29,300 prisoners in Australia. This is equivalent to an imprisonment rate of 175 prisoners per 100,000 adults in Australia. Since 1989, the imprisonment rate has increased by around two-thirds.

DATA SOURCE AND DEFINITIONS This article is based on a forthcoming research paper An Analysis of Repeat Imprisonment Trends in Australia using Prisoner Census Data from 1994 to 2007. The analysis is based on a longitudinal dataset built from the annual National Prisoner Census collected by the ABS. The analytical model used is explained later in this article (see the box: Analysis of the repeat imprisonment dataset). The Prisoner Census provides a snapshot of the adult prisoner population at 30 June each year, and is collected from administrative data sources maintained by corrective services agencies in each state and territory. Within each state or territory, every prisoner is assigned a unique prisoner identification number. This number allows the presence or absence of individual prisoners to be followed over time at each Prisoner Census, and enabled the construction of a 14 year (1994-2007) longitudinal dataset. This dataset is used to identify prisoners with multiple imprisonment episodes. Prisoner refers to a person held in custody. The imprisoned population includes remandees, some of whom may be found not guilty by the court. In all states and territories people remanded or sentenced to adult custody are aged 18 years and over, except Queensland where 'adult' refers to people aged 17 years and over. Release refers to a proxy measure derived from the absence of a prisoner's record in a subsequent Prisoner Census. Information on the release of prisoners is not collected by the census, so people's 'disappearance' from the Prisoner Census is used as a proxy of their release from prison, and their 'reappearance' as a proxy of their reimprisonment. The interval between the year of release and the subsequent reimprisonment is approximated by the number of years between the release year and the first census year of the reimprisonment. The Prisoner Census does not capture all prison episodes, that is it does not capture the entire inflow and outflow of prisoners during the year. Short prison episodes can be missed if they do not span 30 June, which may result in both an underestimate of the number of prisoners serving multiple spells of imprisonment and an underestimate of the number of prison spells for prisoners identified as serving multiple terms. It is also not possible to link prison episodes that occurred in different states and territories.

HOW COMMON IS REIMPRISONMENT?

Over half (56%) of the people in prison in 2009 had been imprisoned before. However, this does not necessarily indicate the rate of reimprisonment, as it does not account for the people who are released from prison, but not reimprisoned. It is also influenced by the number of first time prisoners entering the system, and the length of sentences.

A more valid measure of reimprisonment can be made by following over time a group of people who have been released from prison, and taking the proportion of that group who re-enter the prison system at a later date. Over the four years from 1994 to 1997, 28,600 prisoners were released from Australian prisons (the 1994-1997 release cohort). The analysis in this article is based on this group of people.

Within 10 years of their release, two in five people in the 1994-1997 release cohort had been reimprisoned. The rate of reimprisonment increased relatively rapidly in the early years following release, then levelled out over time.

WHO GETS REIMPRISONED?

As the characteristics associated with reimprisonment are often aligned with other characteristics, a regression model was used to isolate the most important factors (see the box: Analysis of the repeat imprisonment dataset). Results of the modelling showed that reimprisonment was strongly associated with already being a recidivist prisoner, as opposed to being in prison for the first time. Also strongly associated with reimprisonment were the characteristics of being young, of Aboriginal or Torres Strait Islander descent, or, to a lesser extent, being male.

During the 10 years after being released, men were more likely than women to return to prison. Although this gap was quite small at the beginning it increased with the passage of time. By the tenth year, 40% of released men had been reimprisoned at least once, compared with 31% of released women.

Younger prisoners were more likely than older prisoners to be reimprisoned following release. Within 10 years of being released, the reimprisonment rate for the teenager group (those aged 17-19 years when released) was 61%, compared with 23% for those aged 35 years and over.


The reimprisonment rate of Aboriginal and Torres Strait Islander people within 10 years of release was around 1.7 times that of non-Indigenous people.

ANALYSIS OF THE REPEAT IMPRISONMENT DATASET The 1994-1997 release cohort is the group of prisoners who were released at least once between 1 July 1994 and 30 June 1997. In the release cohort and the total prisoner population, around nine in ten prisoners were male, and about one-fifth were Indigenous. Among first-time prisoners, the median age when imprisoned was 28 years. A small proportion (less than 10%) of prisoners were aged 17-19 years when released from prison, while around one-third were aged 35 years and over. Around 40% of the prisoners were in New South Wales prisons. Logistic regression is a type of multivariate analysis. This technique is used to show the effect that each individual factor has on the likelihood of a person being reimprisoned, when all other factors are held constant. For example, characteristics like state or territory of imprisonment and age may be associated with each other so that the association between jurisdiction and reimprisonment is a reflection of the age profiles of different states and territories, rather than a result of that jurisdiction itself. The multivariate analysis disentangles the effects of the following characteristics: sex; age at release; Indigenous status; prior imprisonment; state and territory of imprisonment; previous offence; and the length of the previous prison episode. This statistical technique is applied to the 1994-1997 group 10 years after release.

States and territories

Within 10 years of their release, 48% of prisoners in the Northern Territory had been reimprisoned, compared with the national average of 39%. However, this high reimprisonment rate reflects the demographic characteristics of its prisoner population (such as Indigenous status and age) which are associated with high rates of reimprisonment. After adjusting for these and other factors using logistic regression, Northern Territory prisoners showed an average level of reimprisonment propensity (that is, it was not significantly different from the average across all jurisdictions).


CRIMINAL CAREER DEVELOPMENT

Developing an understanding of the frequency of offending, and the types of crimes committed by chronic offenders, may assist in crime prevention.

In the Prisoner Census, information is collected on only the most serious offence of sentenced prisoners, and the most serious charge for unsentenced prisoners. Analysis of criminal career development is based on the most serious offence/charge, referred to as the 'offence'.

People in the 1994-1997 release cohort were most likely to have been in prison for assault and acts intended to cause injury, and burglary. Illicit drug offences and theft were also common offence types for which prisoners were originally imprisoned.



How frequently were people reimprisoned?

Almost one-fifth (19%) of the 1994-1997 release cohort had been reimprisoned only once by June 30, 2007. One in ten were reimprisoned twice, 6% were reimprisoned three times, and a further 6% were reimprisoned four or more times.

Reimprisonment by offence type

The following analysis looks at patterns of specialisation in offence types and movements from one type of offence to another using descriptive methods.

The reimprisonment rate varied according to the offence type for which the prisoner was originally imprisoned. Members of the 1994-1997 release cohort who had been in prison for burglary or theft had the highest reimprisonment rates (58% and 53% respectively). At the other end of the spectrum, people whose previous offence was illicit drugs or sexual assault and related offences had the lowest reimprisonment rates (24% and 21% respectively).


Specialisation

Offence specialisation was measured by taking the proportion of repeat prisoners whose reimprisonment was for the same offence as that for which they were originally imprisoned. Burglary had the highest rate of specialisation. Just over half (54%) of those previously imprisoned for burglary were reimprisoned for this same offence by June 30, 2007.

Other offence types with a high degree of specialisation (around 50%) included acts intended to cause injury, road traffic offences, illicit drugs and sexual assault and related offences.


Offence type changes by repeat offenders

The probability of reimprisonment chart shows previous offence types along the vertical axis and subsequent offence types along the horizontal axis. The size of the circles are proportional to the probability that a person previously imprisoned for an offence was, at some later date, reimprisoned for the same or a different type of offence. In other words, the pattern along the diagonal indicates the probability of repeat imprisonment for the same offence, while vertical patterns in the matrix indicate progression into particular offence types.

There was a relatively high probability that people previously imprisoned for acts intended to cause injury, robbery, burglary or theft would later be reimprisoned for those same offences. In addition, these offences also attracted a high proportion of prisoners who were previously imprisoned for other offences.

Many offenders also tend to be reimprisoned for offences against justice at some stage. This can be reasonably assumed as attributable to breaches of justice orders. For example, prisoners may be paroled, seriously breach the parole conditions, and then are returned to prison.

Unless offenders started their criminal careers with sexual assault and related offences, deception, or illicit drug offences, they did not tend to commit this sort of crime later.



ENDNOTES

1. Roche, D., 1999, 'Mandatory Sentencing' in Trends & issues in crime and criminal justice Paper No. 138, Australian Institute of Criminology, <www.aic.gov.au>.2. Rawnsley, T., 2003, 'Working Paper No. 2003/02: Dynamics in Repeat Imprisonment: Utilising Prison Census data', cat. no. 1351.0, ABS, Canberra, <www.abs.gov.au>.
HAVE YOUR SAY

Please direct all statistical enquiries to the National Information and Referral Services (NIRS) by emailing client.services@abs.gov.au

Articles in Australian Social Trends are designed to provide an overview of a current social issue. We aim to present an interesting and easy-to-read story, balanced with appropriate statistics. The articles are written as a starting point or summary of the issues, for a wide audience including policy makers, researchers, journalists and people who just want to have a better understanding of a topic. For people who need further information, we provide references to other useful and more detailed sources. Tell us if we are achieving this aim by emailing social.reporting@abs.gov.au