Sentencing in New Zealand: a
statistical analysis
Foreword | Acknowledgements | Executive Summary | Introduction | Chapter 2 | Chapter 3 | Chapter 4 | Chapter 5 | Chapter 6 | Chapter 7 | Chapter 8 | Chapter 9 | Chapter 10 | Chapter 11 | References | Appendix I | Appendix II | Tables | Figures
5.1 Background
The suspended prison sentence was introduced as a new sentencing option by the Criminal Justice Amendment Act 1993. Prison sentences of not less than six months and not more than two years can be suspended for a period not exceeding two years. If the offender is convicted of another imprisonable offence within the suspension period the suspended sentence can be activated.
The use of suspended sentences has been analysed in some detail in other reports due to concern over its widespread use (Spier 1995-1998, Triggs 1998). Around 3000 cases a year result in a suspended sentence, but the expected equivalent drop in the number of offenders sentenced to prison has not occurred. Therefore, it appears that a significant proportion of suspended sentences are replacing (or more often being added to) community-based sentences, rather than replacing prison sentences as intended.
In this chapter, the factors influencing the use of suspended sentences are examined. The implications of the findings are discussed in more detail in section 11.4.2. Because suspended sentences were only introduced in 1993, no analysis of changes over time is presented. This analysis includes all cases resulting in a suspended sentence, regardless of whether the suspended sentence was imposed alone or in combination with a community-based sentence or monetary penalty.
5.2 Single variable analysis of current factors influencing sentencing
Overall, 3.6% of offenders who had a proved case in 1995 received a suspended sentence, either alone or in combination with another sentence.
The single factor most strongly predictive of a suspended sentence was the seriousness of the offence (Figure 5.1). Over 20% of offences in the seriousness range 150 to 300 resulted in a suspended sentence (that is, offences where the average length of imprisonment is 150-300 days).
Suspended sentences also tend to be more often used for offenders convicted on several charges and offenders with a considerable past history of offending, although no single value of any criminal history characteristic on its own resulted in more than an 8% probability of a suspended sentence.
Figure 5.1: Percentage of offenders receiving a suspended prison sentence in 1995 for selected variables
A higher percentage of men (3.8%) than women (2.6%) received a suspended sentence, but this difference is much less than for prison sentences (the imprisonment rate of 10% for men compares to a rate of 3% for women). Slightly more Māori (4.8%) and Pacific peoples (4.2%) than Pakeha/Other offenders (3.4%) received a suspended sentence. People aged 25-29 (4.0%) or 30-34 (3.9%) were more likely to receive a suspended sentence than younger age groups.
By offence group, serious offences against the person had the highest percentage of proved cases resulting in a suspended sentence (15.0%), although suspended sentences were used mainly for the less serious offences in this group, such as serious assault. For other offence groups, 8.4% of proved cases resulted in a suspended sentence for domestic violence, 5.3% for drug offences, 4.3% for property offences, 2.1% for traffic offences, 1.6% for minor offences against the person, 1.4% for offences against justice (excluding breaches of periodic detention at 0.2%), and 0.9% for disorder offences.
5.3 Multivariate analysis of current factors influencing sentencing
5.3.1 The fit and accuracy of the 1995 model
The logistic regression models for suspended sentences achieved a significant overall fit to both the full and half 1995 data, as indicated by a log likelihood ratios significant at the 0.0001 level of probability. The results of the test phase (Figure 5.2) show a similar fit for the model developed using one half of the 1995 data (the 'fitted data') compared to the other, unseen half of the 1995 data (the 'test data').
However, neither set of data achieved a close fit to the ideal line and the residual (unexplained variation) term was significant. At a probability level of 10-15% the actual percentage of people receiving a suspended sentence was above the predicted probability, while the opposite occurred at probabilities over 25%. This indicates that there is some systematic bias in the error term of the model. Therefore the following results must be interpreted with caution.
Also, the suspended sentence model differs from the prison model in that it does not predict over the full range of probabilities. Very few offenders scored a predicted probability of more than 0.3 (a 30% probability of a suspended sentence). Thus, there were no factors or combination of statistical factors that predicted a high probability of receiving this sentence.
Figure 5.2: Plot of the predicted probability versus the actual proportion of offenders receiving a suspended sentence, 1995 fitted and test data
5.3.2 Results of the 1995 model
When all statistical factors are considered together, the most important factors in determining the use of suspended prison sentences appear to be the seriousness and type of offence (Table 5.1). The model suggests that offences of moderate to high seriousness (seriousness scores of 180-365) have the highest relative risk of resulting in a suspended sentence, although offences in the low to moderate seriousness range (>20-180) and high seriousness range (>365) also have a high risk of receiving a suspended sentence relative to the low seriousness reference group.
All types of violent offences and drug offences also have odds ratios of more than 1.0 and therefore increase the probability of receiving a suspended sentence, once the effect of other factors are accounted for. Relative to the reference group (property offences), breach of periodic detention, other offences against justice, and traffic offences are less likely to result in a suspended sentence.
Offenders who have already accumulated a history of convictions of moderate seriousness and those who have already served the more serious community-based sentences or a prison term are also relatively more likely to receive a suspended sentence than offenders without these characteristics.
Table 5.1: Logistic regression model for suspended prison sentences, 1995
Note: An odds ratio of >1.0 indicates a high relative risk (i.e. more likely to receive this sentence than the reference group). The most significant variable (highest Wald Chi-square, lowest probability P), is rank '1'.
Offenders aged under 20, and especially those under 17, are less likely to receive a suspended sentence. Unlike prison sentences, females are as likely to receive a suspended sentence as males. As for imprisonment, ethnicity was not a significant factor in determining the use of suspended sentences. Not was the type of plea significant.
In summary, these are similar characteristics to those that indicate a moderate level of risk of imprisonment, with the exception of the lack of a gender effect. The implications of these results are discussed in section 11.4.2, especially in relation to the known net-widening effects associated with the use of suspended sentences.
