Reconviction patterns of released prisoners: A 60 months follow-up analysis (pdf: 185KB)
Arul Nadesu
Principal Strategic Adviser
Policy, Strategy and Research
Department of Corrections
2009
Reducing re-offending is an important objective for most correctional services. As such, measures of recidivism, particularly reconviction and re-imprisonment rates, are key indicators of organisational performance. The current study is intended to assist the Department of Corrections in meeting its strategic objectives. In addition to providing straightforward data on recidivism, it is hoped also that the study will be a valuable reference tool for those interested in correctional trends and issues, and will inform discussions on improving New Zealand’s correctional system.
The data presented here are based on the “recidivism index” (RI) methodology used in the Department of Corrections’ annual reporting of reconviction. This method quantifies the rate of reconviction and re-imprisonment for specified sub-groups of offenders, over follow-up periods of defined length, after release from a custodial sentence, or following commencement of a community sentence or order. Conviction and sentencing data is obtained from the Ministry of Justice’s Case Management System (CMS) database.
This report summarises patterns of reconviction and re-imprisonment amongst almost 5000 offenders who were released from prison during the 12 months period 1 April 2002 to 31 March 2003. The same cohort of offenders was the subject of previous reports in this series, which provided reconviction and re-imprisonment rates within 36-months and 48-months follow-up periods[1]. The current report provides similar data, but now with a 60-months follow-up period: that is, figures represent reconvictions for offences that occurred within 60 months of each individual offender’s release date (up to 31 March 2008)[2].
Recidivism figures are produced for two potential outcomes: reconvictions leading to any sentence administered by the Department of Corrections (community-based or prison), and reconviction leading solely to a term of imprisonment. Imprisonment figures are generally considered the more critical of the two measures, as this outcome is associated with more serious offences and higher costs of sentence administration.
Recidivism figures exclude reconvictions which do not result in sentences administered by the Department (fines, convicted and discharged, etc). These figures also exclude re-sentences for breaches of community sentences or recalls to prison for breaches of parole conditions. In summary all reconvictions data presented here should be interpreted as restricted to convictions for a new offence resulting in imprisonment or a community-based sentence.
It is also important to note that most of the figures here represent any reconviction or
re-imprisonment during the 60-months follow-up period: some of the individuals re-imprisoned may have had multiple, successive terms of imprisonment during that period. However, in places the report also provides information on multiple re-imprisonments of categories of offenders during the 60-months follow-up period.
Many offenders are sentenced to a prison term for more than one offence. Therefore, when offenders are grouped by offence type, the convention is to identify the most serious offence type[3] (MSO) for which they were imprisoned for their original sentence. However, another way of identifying the offence type is according to an individual offender’s ‘lead offence’. This refers to the offence which incurred the greatest number of days of imprisonment as ordered by the judge for that particular offender. It should be noted that in most cases the lead offence and MSO are one and the same; however, those scenarios where the two do not coincide are of interest, and a comparison of the two measures’ utility is explored in this report. Grouping by the lead offence rather than the MSO is used in only one section of the study.
The average number of days it takes for an offender to commit their first proved re-offence is examined by offence type. This calculation is based on about 2600 prisoners, who were convicted of a new offence and were returned to prison during the 60-months follow-up period. Although many offenders had multiple offences during the 60-month follow-up period, the first proved re-offence is considered for this study.
A subsequent report in the series is planned (expected early in 2010) to give comparative figures for a five-year follow-up from a different cohort of offenders, those released from prison during the 12 months period 1 April 2003 to 31 March 2004. The study of two cohorts of offenders released from prisons in successive years will indicate possible trends in base rates of offending. The study will also examine some other factors such as offenders’ education, family relationship and association with gangs and whether these factors are significantly related to reconviction and re-imprisonment rates.
Countries differ markedly in how criminal justice data are handled: reconviction and
re-imprisonment rates are influenced by legislation, sentencing practices, resource levels of criminal justice sector agencies, as well as volumes of crimes committed and rates of detection and resolution. Consequently, comparisons of reconviction or re-imprisonment rates between countries are usually a fraught exercise. Nevertheless, the study provides a benchmark for five-year recidivism rates for offenders released from New Zealand prisons.
Across the entire sample, 52 percent of offenders released from New Zealand prisons in 2002/03 were convicted of a new offence and were returned to prison at least once during the 60-months follow-up period.
To clarify the above graph, the shape of the two curves is illustrative of the relationship between the volume of offenders re-imprisoned and elapsed time over the 60-months follow-up period. The lower (blue) line is the cumulative proportion re-imprisoned, and indicates that numbers rise steadily early in the follow-up phase, such that by the one-year mark, 26 percent of the sample had already re-offended and been reconvicted and returned to prison. By the two-year mark, 37 percent had re-offended and subsequently reconvicted and returned to prison, by the three-year mark 44 percent of the sample, by the four-year mark 49 percent of the sample, with the 52 percent figure attained by 60 months.
The upper (red) curve in the above graph shows the proportion of all those re-imprisoned (in percentages) returning to prison by successive three-monthly intervals. This shows, for example, that of those who were re-imprisoned over 60 months, about half (51 percent) were re-imprisoned within the first twelve months. From that point the rate of “relapse” slows, with a further 21 percent returned to prison by the 24-month point, 14 percent returned to prison by 36 months, eight percent returned to prison by 48 months, and the remaining six percent returned to prison over the final twelve months.
The relationship between age and reconviction rates has been the subject of considerable criminological analysis in the past. A great many studies have established a very strong (inverse) correlation between age and reconviction rate.
The graph below gives rates of re-imprisonment and reconviction for offenders of different age bands (note that prisoners’ ages here are as at the time of release). Seventy one percent of those aged under 20 were re-imprisoned within 60 months, while just 35 percent of those aged over 40 were re-imprisoned. In other words, prisoners aged under 20 have double the likelihood of returning to prison than do those aged over 40.
In the previous section, re-imprisonment rates of prisoners by age group were examined, and it was found that the risk of re-imprisonment for young offenders was very high. In this section, young offenders are examined more closely by way of a statistical procedure known as survival analysis. Survival analysis presents data in terms of time elapsed until an event of interest occurs. Time here is reported in terms of months from the beginning of the follow-up period for each individual offender, while the event of interest (obviously) is re-offending.
Four hundred and sixty-three offenders (see Appendix 1) were under the age of 20 at the time of release. The graph below indicates that, of the 463, 107 re-offended within three months and subsequently commenced prison sentences; in other words, 356 offenders “survived” the first three months from their individual release dates. The survival curve below shows similar information by successive three-monthly intervals: initially, there is a rapid decline in “survival”, with almost half of the cohort of young offenders returning to prison within the first twelve months.
Of the 463 offenders, 136 offenders survived without a re-imprisonment for a new offence during the 60-months follow-up period. However, of the 136 “surviving” offenders, 80 were in fact re-convicted and commenced a community sentence over the same period. This left only 56 offenders of the original 463 who were not convicted for a new offence.
In this section, we explore re-imprisonment rates on the basis of the offence committed by prisoners for their original sentence. For this discussion, it is important to distinguish between offence group and offence type. Offence groups are broad categories of crimes encompassing many different offence types. For example, Dishonesty (an offence group) comprises a variety of offence types, the main ones being: burglary, theft, fraud and car conversion. The table below provides more detail on these groupings.
Graph 5 illustrates re-imprisonment rates by offence groups. When examined in this context, the re-imprisonment rate is by far the highest among dishonesty offenders, with 68 percent returning to prison within five years. By contrast, the re-imprisonment rate of sex offenders is just 27 percent. About half of offenders convicted of violence return to prison within five years. However further refinement can be gained from dividing these broad offence groups into their component offence types, as graph 6 shows.
A number of interesting observations can be made when offenders are examined in this way. When considered as a unified offence group (as in graph 5), 27 percent of sex offenders were re-imprisoned within 5 years. However, this offence group consists of offenders with either child victims or adults (i.e., rapists). Re-imprisonment rates for those convicted of child-sex offences are much lower (21%) than those convicted of rape (35%). This is likely to reflect the fact that rape offenders tend to be more generally orientated towards criminal offending (of all types), which is less common amongst child-sex offenders, a proportion of whom have no convictions for other types of offences. It is also possible that reporting of sexual abuse, particularly involving child victims, is less likely to be reported.
Each year a considerable number of offenders are sentenced to prison in New Zealand for either disqualified driving or drunk driving (these two offences make up 95% of all traffic offences resulting in imprisonment). While the overall rate of re-imprisonment for traffic offenders is 46 percent, a significant difference emerges between drink-drivers (36%) and disqualified drivers (57%). Many disqualified drivers are in fact persistent offenders with extensive criminal histories, of which disqualified driving is simply one aspect.
Once again, it is important to recall that for each term of imprisonment, there may exist multiple convictions in different offence groups, but offenders are categorized here by the most serious offence for which they were originally sentenced.
| Violence |
Sexual |
Drugs & Antisocial |
Dishonesty |
Property |
Traffic |
Admin |
|---|---|---|---|---|---|---|
| Homicide |
Child sex |
Drugs (Cannabis) |
Fraud |
Property Damage |
Drink-Driving |
Breaches |
| Robbery |
Rape |
Drugs (Not cannabis) |
Theft |
Property Abuses (Trespassing) |
Disqualified Driving |
Immigration |
| Assaults |
Family Offences |
Burglary |
Against Justice |
|||
| Intimidation & Threats |
Car Conversion |
The following section examines more closely the reconviction dataset by disaggregating offenders according to the number of previous prison terms. Those for whom their release in 2002/03 was from their first-ever prison term are designated in the following as “first-timers”. The remainder, who had served two or more previous prison terms, are designated as “recidivists”.
As noted above, across the entire sample of released offenders, about one third had been sent to prison for the first time. The re-imprisonment rate of these first-timers is 32 percent; in contrast, the re-imprisonment rate of the remainder – the recidivists - is 63 percent.
In general, reconviction and re-imprisonment rates for sex offenders[5] are low relative to other offence types. However, when disaggregated by previous sentences, this finding does not hold true for the recidivist sex offenders, whose rate of re-imprisonment is 49 percent; for the first-timers the rate of re-imprisonment is only ten percent over 60 months. Recidivist sex offenders are thus five times more likely to return to prison than first-timer sex offenders. It is intended that re-imprisonment rates for recidivist child sex offenders and recidivist rapists will be provided separately in subsequent reports.
The above graph clearly shows that re-imprisonment rates do not vary significantly between recidivist offenders and first-timers for dishonesty offences, burglary, car conversion and theft. These dishonesty first-timers are likely to have had many prior convictions and, by the time they receive a prison term, to have exhausted all community sentence options available to the judge. Similarly, re-imprisonment rates do not vary markedly between recidivist disqualified driving offenders and their first-time counterparts. Many disqualified drivers are in reality persistent offenders, for whom disqualified driving is just one aspect of their extensive criminal histories. When such offenders are imprisoned for disqualified driving, they are typically also (concurrently) convicted of other offences such as drink-driving, burglary, car conversion and theft.
Assaults, intimidation and threats, family offences and drink-driving indicate considerable differences in re-imprisonment rates between recidivist offenders and
first-timers. Such recidivist offenders are in fact two times more likely to return to prison than first-timers.
The following section examines only offenders who were released on parole[6]. The sample of 1588 parolees was divided into the same two groupings of first-timers and recidivists. About a third of the parolees were first-timers. It is important to note that offenders kept to their sentence expiry dates are also included in this study. It is intended that these offenders will be excluded and further analysis will be provided in subsequent reports.
Many offence types are excluded for this study due to low numbers in certain categories. Violence, sex, drugs and burglary are the most common offence types, which tend to incur longer sentences.
Most notable in the graph is that the re-imprisonment rate for recidivist sex offenders is 45 percent and for the first timers only seven percent over 60 months. Recidivist sex offenders are thus six or seven times more likely to return to prison than first-timer sex offenders.
Similarly, as is apparent from the graph below, the rate of re-imprisonment over 60 months for recidivist Pacific offenders (57%) is not much lower than recidivist NZ Europeans (61%) and recidivist Maori offenders (66%). The rate of reconviction for recidivist Pacific offenders (78%) is only slightly lower than recidivist NZ Europeans (79%), while the rate of reconviction for recidivists Maori offenders is 83 percent over 60 months.
The graph below shows the difference in the rate of re-imprisonment between recidivist offenders and first-timers for each age group at release. Although there is a strong inverse correlation between age and re-imprisonment, recidivist offenders tend to have a higher rate of re-imprisonment compared to first-timers regardless of their current age. Recidivist offenders who are over 40 at release are 4.2 times more likely to return to prison than first-timers who are over 40 at release.
In general, the overall re-imprisonment rate of female offenders (36%) is lower than that
of male offenders (53%). The graph above contains some interesting findings in relation to female offenders: the rate of re-imprisonment for recidivist female offenders (54%) is not greatly dissimilar to that of recidivist male offenders (64%).
Recidivist female offenders turn out to be 2.5 times more likely to return to prison than first-timer females, while recidivist male offenders are just 1.9 times more likely to return to prison than first-timer males. However, the rate of reconviction for recidivist female offenders (80%) is almost same as recidivist male offenders (81%); clearly there is a “hard core” of female offenders who are as criminally inclined as their male counterparts.
In summary the analysis confirms simply that, the more time in the past someone has been in prison, the more likely they are to return to prison following any given release.
The graph below relates to those offenders who were re-imprisoned within 60 months, and indicates the average number of days elapsed before an offender committed their first proved re-offence. Data is presented by offence type relating to the original conviction, and the number of days relates to the date of the offence (not the date of conviction). Once again, dishonesty offenders proved to be the most prolific, in that offenders convicted of burglary, theft, car conversion, fraud and property abuses
re-offended within relatively short periods after release. Offenders convicted of drink driving and drugs took longer to re-offend. This latter finding may be a consequence of the fact that drink-driving and drug offences typically are not crimes reported by the public, rather they tend to be offences that are detected directly by Police. Offenders convicted of homicide and sex against children also took longer time to re-offend.
A proportion of all offenders recorded as having been re-imprisoned during the follow-up period will have experienced more than one term of imprisonment during that 60-months period. The figures reported below count distinct “aggregate sentences” for each released prisoner during the 60-month follow-up period. An aggregate sentence of imprisonment reflects the fact that many offenders are sent to prison after being convicted for multiple charges, each of which attracted separate sentences of imprisonment (often of varying lengths). Counting aggregate sentences however may underestimate the frequency of imprisonments, as some prisoners are convicted and sentenced to a further term of imprisonment while already a sentenced prisoner (the second sentence may be simply subsumed within the earlier aggregate sentence).
As noted above, across the entire sample of offenders (n = 4945), 52 percent (n = 2563) were convicted of a new offence and were returned to prison at least once during the 60-months follow-up period. These re-imprisoned offenders accumulated a total of 5235 sentences of imprisonments for new offences over the 60 months period. At the extremes, one offender released in 2002/03 was returned to prison 19 different times during the 60-months follow-up period. Around 145 offenders released in the period returned to prison at least five times in five years.
Many recidivists appear almost endlessly to cycle through a sequence of offending, conviction, imprisonment, release, and rapid return to further offending. This situation creates a huge cost to the criminal justice sector as well as to society. This situation underlines the need across the justice sector to find effective options to target recidivism, and to reduce the costs imposed by such behaviour on society.
The following sections highlight some interesting trends of re-imprisonment frequency among selected demographic groups.
About 52 percent of offenders who were re-imprisoned over 60 months for a new offence returned to prison at least twice. About 40 percent of female offenders, who were re-imprisoned over 60 months for a new offence returned to prison at least twice.
About 52 percent of Maori offenders who were re-imprisoned over 60 months for a new offence returned to prison at least twice. About 64 percent of youth offenders (under 20 at release), who were re-imprisoned over 60 months for a new offence returned to prison at least twice.
An important offender characteristic featured in this report is the “offence type”. As a great number of offenders are sent to prison for multiple offences, offenders are typically grouped according to their most serious offence (MSO) for which they were imprisoned on the original sentence. MSO rankings are determined from the Ministry of Justice “Seriousness of Offence Scale”, which determines offence seriousness in accordance with the average number of days imprisonment ordered by judges, for that specific offence type, over the past five years.
Another way of identifying “offence type” when an offender is convicted of multiple and diverse offences is by identifying the offender’s “lead offence”. A lead offence is the offence which, on the date of that particular offender’s imprisonment, incurred the greatest number of days of imprisonment as ordered by the sentencing judge.
Table 2 presents a comparison of the total number of offenders and the re-imprisonment rate by the MSO and lead offence. Overall, the two classifications produce key differences for some offence types, but minimal differences for others.
Unsurprisingly, very serious offences such as homicide and sexual offences recorded either no differences or little differences when MSO and lead offence methods were used. On the other hand, some offence types recorded considerable differences when derived as the MSO and the lead offence. In particular, the number of offenders whose lead offence was drink driving was significantly higher than the number of offenders whose MSO was drink driving (587 compared with 447 respectively). Similarly, the number of offenders whose lead offence was disqualified driving was significantly lower than the number of offenders whose MSO was disqualified driving (317 compared with 475 respectively).
This discrepancy appears to be due to a number of offenders being convicted for both driving while disqualified and drink driving. Under the MSO classification, driving while disqualified is considered to be a more serious offence; however, in practice, the sentence which judges apply to the separate offences indicates that they typically regard the drink-driving offence(s) as more serious.
Similarly, the number of offenders whose lead offence was burglary was significantly lower than the number of offenders whose MSO was burglary (728 compared with 850 respectively). These results indicate that under the MSO classification some offenders were grouped into the burglary offence type despite receiving longer sentences for other dishonesty offences committed such as car conversion and theft.
In most cases, the differences between the two classifications had little impact on the re-imprisonment rate over five years. For instance, under Lead Offence classification the re-imprisonment rate for burglary is 73.6 percent. Under MSO classification the
re-imprisonment rate for burglary is 72.2 percent. However, re-imprisonments for offences such as minor assaults, intimidation and threats, receiving and fraud have significant differences under both classifications.
The central finding of this study is that, amongst offenders released from New Zealand prisons in 2002/03, 52 percent were convicted of a new offence and received a further prison sentence within 60 months of the first release. Of those who were re-imprisoned over 60 months, about half were re-imprisoned within the first twelve months. Clearly, the first year after release is the highest-risk period for relapsing into old patterns of behaviour.
The re-imprisonment rate of young offenders was particularly high. Offenders who commence their offending careers during their teenage years are considerably more likely to become persistent offenders, particularly if their initial crimes are ones that result in a prison sentence. Previous study in this series showed that many prisoners were first convicted and imprisoned when they were young. Interventions with youth offenders in prisons are thus an important priority for the Department, and if effective would have significant “down-stream” benefits.
The re-imprisonment rate of dishonesty offenders - those convicted of theft, car conversion and burglary - was very high. These are of course crimes with very high base-rates (that is, the offences are more common), which means that those who tend to engage in such crimes probably do so, relatively speaking, with high frequency.
An interesting finding emerged with sex offenders: these individuals are typically known to have relatively low rates of reconviction. However, when separated into first-timers and recidivists, it became apparent that there is a sub-set of sex offenders who are very active criminals, for whom the sexual offending is but one expression of their general antisociality.
This finding was also apparent in general figures showing relationships between original offence type and subsequent re-offence type. This indicated that, amongst the recidivistic offenders, there is only a modest tendency towards offence specialisation. The latter was apparent only amongst violence and driving offenders; in the main there was a distinct pattern of versatile offending, spanning all categories of offence types.
Overall, the most striking finding of the study is that, the more often in the past someone has been in prison, the more likely they are to return to prison following any given release. Conversely, the great majority of those individuals, who come to prison on their first-ever sentence of imprisonment, do not in fact return to prison. This is perhaps the most heartening finding in what is otherwise a somewhat bleak picture of recidivistic offending.
Reconviction Patterns of Released Prisoners (March 2008),
Policy Strategy and Research, Department of Corrections.
Reconviction Patterns of Released Prisoners (March 2007),
Policy Strategy and Research, Department of Corrections.
Annual Report 2005/06 (2006), Department of Corrections.
[1] 'Reconviction Patterns of Released Prisoners' ( A 36-months and A 48-months follow-up Analysis) available at http://www.corrections.govt.nz/public/research/reimprisonment-report/.
[2] The data set also included reconvictions on dates up to November 2008 when the offence date was prior to 31 March 2008
[3] MSO rankings are determined from the Ministry of Justice Seriousness of Offence Scale, which orders offences in accordance with the average number of days imprisonment ordered by judges, for that specific offence type, over the past five years
[4] The high base rates of dishonesty offences can be appreciated from the fact that in the year to June 30 2008, over 220,000 dishonesty offences were recorded by Police; on the other hand, just on 3500 sexual offences were recorded.
[5] Child sex offenders and rapists are combined and reported as sex offenders
[6] Offenders subject to a determinate of more than two years can apply to the NZ Parole Board to be released on parole at any time from the completion of a third of the sentence