Findings

Leave to apply

As noted above, Leave to Apply for Home Detention currently is granted by the sentencing judge. Of the total sample of 7606 offenders, 2557 (34% of the total sample) were granted Leave to Apply.

The following table sets out the proportions of Maori and European offenders who were granted Leave to Apply, out of all Maori (4023) and Europeans (2751) in the sample. This indicates a statistically significant difference of approximately ten percentage points between Maori and European males, and Maori and European females.

Table 1: Number and percentages who were granted leave to apply, by ethnicity

The following sections present data on the relationship between ethnicity of offenders and a range of factors which ordinarily would be expected to influence such judicial decision-making. In general, the disparities in Table 1 above are replicated in the following tables of data: that is, a clear trend emerges consistently whereby Maori offenders, when sub-grouped variously via certain categories of interest, have lower rates of success in terms of Leave to Apply for Home Detention.

It should be noted however that simple associations between ethnicity and individual factors can be misleading: when multiple related variables potentially play a role in influencing outcomes across a large sample of individuals, statistical analysis (of the type reported below) is essential to precisely quantify the relative contribution of relevant factors to outcomes. Such analysis is the only adequate means of clarifying whether the relevant variables partly or fully account for the outcomes of interest.

Potential explanatory factors

Seriousness and type of offence

Legislation requires that, in deciding whether to grant Leave to Apply, judges take into account the seriousness of the offence, and associated considerations of potential risk to the community. Offence seriousness can be quantified in terms of the average sentence type and length that particular offences attract.

The following figure indicates the spread of offence seriousness rankings, by ethnicity. To explain the graph, the first data points to the left indicate that around 70% of those in the sample were sentenced for offences which typically attract imprisonment sentences of 0-12 months; data points on the right-hand side indicate that around 5% of the sample were sentenced for offences which typically attract imprisonment sentences of four years or longer.

Figure 2: Seriousness of sample offenders’ offences, by ethnicity

The above table indicates a small tendency for Maori offenders in the sample to have been convicted of offences of a higher level of seriousness than those of New Zealand European offenders.

Offenders seeking Leave to Apply may also face different outcomes depending on the category of offence. Some types of offences are associated with lower rates of obtaining Leave to Apply. The following table gives the numbers of Maori and New Zealand European offenders in each offence category, along with the percentage of each given Leave to Apply.

These figures indicate a tendency for Maori to be slightly over-represented in violence offending, as well as driving offences, but to be less likely to appear for sexual offences.

Table 2: Sample offenders’ offence class, by ethnicity and % granted Leave to Apply

Risk of re-offending

Each offender managed by the Department of Corrections has a calculated risk score which indicates the likelihood of future reconviction or imprisonment. Judges are not necessarily aware of the offender’s risk score; however, risk scores can function as a kind of “shorthand” for the length, “density” and seriousness of an offending history, factors which judges are aware of, and which are considerations in decision-making around sentencing.

Risk scores of Maori and New Zealand European offenders in the full sample are revealed on the following table. In each risk band, Maori were less likely to obtain Leave to Apply. However, across the entire sample, Maori also tended to be concentrated in the higher risk-score bands: thus a large proportion of the Maori offenders whose risk scores were recorded - 76% - were in risk bands over 0.50. Notably, as risk increases, rates of obtaining Leave to Apply tend to decrease sharply. In contrast, 64% of New Zealand Europeans were in these higher-risk bands ⁶.

Of those granted Leave to Apply, the average risk score for Europeans was lower, at 0.34, while the average score for Maori was 0.41 - in other words, Leave to Apply is granted to Maori offenders who are rated, on average, higher-risk than Europeans.

Table 3: Sample offenders’ risk scores by decile band, ethnicity, and rates granted Leave to Apply

Previous history of serious violent or sexual offending, and sentence non-compliance

According to informant interviews, a history of serious violent or sexual offending, and/or of breaches of previous sentences and orders, influence judges’ decisions on whether to grant Leave to Apply. Figures in the following table indicate that Maori in the sample had proportionally more previous convictions for serious violent or sexual offending. They also indicate that Maori offenders were more likely to have previous convictions related to non-compliance with earlier sentences or orders.

Table 4: Sample offenders’ prior convictions for violence and breaches, by ethnicity, and rates granted Leave to Apply.

Gang membership

Similarly as with sentence non-compliance, informant interviews confirmed gang membership as a potential obstacle to obtaining Leave to Apply: gang membership (which is often noted in pre-sentence reports) is likely to result in judges forming the view that release on Home Detention will be a higher-risk proposition.

Current or previous gang membership was noted for 866 Maori offenders in the full sample (21.5%), while this feature was noted for only 127 Europeans (5%).

Table 5: Sample offenders’ gang membership, by ethnicity, and rates granted Leave to Apply.

In summary, the above analysis indicates a tendency for Maori offenders to present with characteristics which, singly or in combination, may reduce the likelihood of obtaining Leave to Apply when sentenced in court. As noted above however, the extent to which such factors adequately account for the low rates of obtaining Leave to Apply amongst Maori requires the kind of statistical analysis presented below.

Offender applications for approval to the Parole Board

As noted above, data was also extracted to indicate whether those offenders with Leave to Apply did in fact pursue an application to the Parole Board to serve the sentence under Home Detention conditions. It appears that it is not uncommon for a proportion of offenders to fail to initiate applications; others do so, but then withdraw the application before a decision is made. These current figures were analysed by ethnicity: these indicate that, once granted Leave to Apply, 87% of European offenders proceeded with an application, while just 78% of Maori did so.

Table 6: Sample offenders who failed to pursue application for Home Detention, by ethnicity

It is not entirely clear why significant numbers of offenders with Leave to Apply do not take the necessary steps to apply for Home Detention, or initiate an application but then fail to follow through with it. Informants suggested that such decisions can occur when an offender assesses their chances of success with the Parole Board to be low, usually because of a lack of a suitable home address to put forward. It was also suggested that the Home Detention application process itself is quite challenging, involving multiple steps and stages; a proportion of offenders seem to decide that it is not worth the effort.

A Probation informant also considered that reluctance to have a Probation Officer visiting and interviewing family members could be a factor in some cases; offenders occasionally express dislike for the idea that members of the household will be told about the offender’s offending history, or be asked to consent to a criminal history check themselves. However, in the absence of readily accessible data that sheds light on why such decisions are made, this issue of a differential between Maori and European in Home Detention applications could not be further investigated.

Parole board decision-making

An offender imprisoned but with Leave to Apply for Home Detention must initiate (and follow through on) an application to the Parole Board, in order to be granted permission to serve the sentence under Home Detention conditions. Below are the numbers and percentages of those actually granted Home Detention, relative to the numbers who followed through an application to a Parole Board hearing.

Table 7: Approval rate of applications for Home Detention, by ethnicity

As was observed with the Leave to Apply figures noted above in Table 1, a clear disparity emerges between Maori and New Zealand European offenders in rates of approval, with significantly fewer Maori having their applications for Home Detention being granted. The scale of the difference at this point is very similar to that at the Leave to Apply stage, with an approximately ten percentage-points difference.

This finding is examined with reference to Parole Board decision-making, and the range of considerations that apply at that stage. As was done with the Leave to Apply data, the following table contrasts the percentages of Maori and New Zealand European in each offence category that were approved for Home Detention. This indicates that, irrespective of the type of offence, Maori offenders were approximately one-third less likely to be approved for Home Detention as were New Zealand Europeans.

Figure 3: Approval rate of applications for Home Detention, by offence type and ethnicity

Data were also analysed with respect to the length of the prison sentence which offenders received at time of sentencing. Interestingly, almost no offenders with sentences of under four months were approved for Home Detention; because the processes required to facilitate release on Home Detention are quite time and resource-intensive, this effort can not be justified when as a result the offender spends no more than two or three weeks on Home Detention. Leave to Apply is commonly withheld in the majority of such cases; further, in law Home Detention cannot be approved (Section 35(3) Parole Act 2002) when two weeks or less remain to serve of the imprisonment sentence. In relation to sentence length per se, no clear differences emerged between Maori and New Zealand Europeans with respect to average sentence length, or in respect to rates of approval of Home Detention applications. These data are attached in Appendix 1.

Categorisation of Parole Board reasons for declining applications

As noted above, database records were obtained which included a brief narrative summary of the Board’s reasons for its decisions in either declining or approving an application. All records were then categorised, and then analysed with respect to offender ethnicity.

The following table lists the main categories of reasons for Parole Board non-approval of Home Detention applications, with the numbers and percentages of offenders in each ethnicity.

Table 8: Reasons for non-approval of HD applications by ethnicity

As is apparent from the figures, a range of reasons are given by the Board in support of decisions to decline Home Detention applications. The figures in the above table indicate that New Zealand Parole Board decisions not to approve release on Home Detention most commonly was the result of a perception that such a release would be inconsistent with public safety. When individual cases were categorised according to reasons given, a greater proportion of Maori (22%) offenders was declined Home Detention than of Europeans (16%) for reasons relating to risk ⁷. On all other factors, differences in rate of approval between Maori and non-Maori were small. With respect to unsuitability of proposed address, Maori were no more or less likely to be declined than were European for this particular reason.

“Procedural issues” in the above table often relate to instances where an offender appears before the Board but the necessary reports, or information for particular parts of the report, are not available or missing. Such difficulties accounted for 39 instances of refusal - around 2.5% of the denials in the sample.

A small difference emerges with respect to refusal on the grounds of non-availability of rehabilitation programmes. Parole Board member informants stated that a viable rehabilitation option was often critical to approvals of Home Detention applications. Slightly more Maori were refused permission for this reason. Given that uptake of programmes in prison by Maori is not problematic, the finding may reflect the possibility that a greater number of Maori offenders sought release to locations - perhaps rural - where suitable programmes were unavailable.

As a group, the average risk score of Maori offenders making applications to the Parole Board was higher than that of European offenders. Of those refused at the Board hearing stage, the gap between average RoC*RoI scores was as follows: 0.40 for New Zealand European males, and 0.48 for Maori males. Similarly as at the Leave to Apply stage, the Parole Board was also less inclined to approve Home Detention applications as risk score increased. However, average scores for those granted Home Detention were 0.26 for Europeans, and 0.30 for Maori males.

Statistical analysis of main data set

As a means of quantifying the relationship between Home Detention outcomes and potentially relevant offender characteristics, the main data set was subject to sophisticated statistical analysis. Using a technique known as logistic regression ⁸, the analysis sought to quantify the relative contribution, to each outcome, made by the various factors included in the data set (for details of the analysis, see attached report “Ethnic bias in rates of access to Home Detention”, by John Horwood, Christchurch School of Medicine).

Collectively the results of the statistical analysis point to certain conclusions. Much of the apparent disadvantage experienced by Maori offenders (in terms of rates of access to Home Detention at both the sentencing and parole hearing stages) appears largely explicable by ethnic differences on two broad factors: (a) offence seriousness (a combination of indicators including most serious offence type, relative offence seriousness, and sentence length) and (b) offending history (a second combination of indicators involving number of previous custodial sentences, a history of violent and/or sexual offending, previous breaches of sentences, and gang membership); these factors were, in and of themselves, highly predictive of access to Home Detention.

When the influence of these factors was taken into account, the ethnicity-related ten percentage point differences in Leave to Apply and Parole Board approval rates shrank: just 3.6% fewer Maori were given Leave to Apply for Home Detention, and 2% fewer Maori offenders were approved for Home Detention, than would have been expected for a corresponding group of New Zealand European prisoners with similar presenting characteristics and backgrounds. These adjusted differences are of the same order as those observed between Pacific and New Zealand European offenders.

Statistical analysis of the type conducted here is not without its limitations however. There may be residual sources of confounding that were not well represented in the model and which could explain remaining ethnic differences in access. For example, the analysis could not account for ethnic differences in factors such as prisoner demeanour, behaviour or presentation at either the sentencing or parole hearing stages. Nor was it possible to incorporate variables such as content of relevant official reports presented at time of decision-making. Such factors may well also influence Home Detention outcomes.

In summary however, the logistic regression analysis indicates that disparities in rates of access for Maori were largely (though not fully) explained by critical differences in this sub-population of offenders.

Breaches of Home Detention

Breach rates amongst those approved for Home Detention were also examined. Analysis of data indicated that the number of offenders in the sample subsequently breached for violating the conditions of the Home Detention order was relatively small - just 3.6% of all such offenders, although Maori offenders were somewhat more likely to feature in such cases.

Table 9: Rates of breaches in Home Detention, by ethnicity

Maori were more likely to be breached, although the actual number of cases (21) was low - less than 5% of the Maori sub-group of Home Detainees. Interestingly, eight of the 21 cases of Home Detention breach amongst Maori occurred with females.

In order to understand what might be driving the higher number of breach instances amongst Maori, Probation Officer case notes for all individual cases (around the time of breach action) were reviewed, with the circumstances resulting in breach action noted. However, this exercise proved inconclusive: a very disparate range of events and behaviours were recorded as leading up to the initiation of breach action, but no particular trends or themes were apparent.

6. Analysis of the offending histories of the sample confirmed that almost three-quarters (73%) of Maori in the sample had histories of sentenced offending that commenced prior to age 20, while fewer Europeans did (59%). Age at first conviction is an important factor in the risk score calculation.

7. Section 35(2)(a) of the Parole Act states that Home Detention “can be granted if the offender will not pose an undue risk to the safety of the community …if he or she is detained on Home Detention.” This is thus the paramount test, and concern, for the Parole Board. This must be satisfied on reasonable grounds and s.35(2)(b) sets out other factors, such as seriousness of offence and rehabilitation, for consideration.

8. Logistic regression is a statistical procedure designed to quantify the relative contribution made to certain binary (yes/no-type) outcomes when a wide range of variables are potentially relevant to an outcome.