Comparison of socio-economic and reconviction outcomes for offenders sentenced to home detention or a short sentence of imprisonment respectively
Dr Wayne Goodall
Principal Strategic Analyst, New Zealand Department of Corrections
Dr Goodall is a member of the Department’s research and analysis team. He has been in Corrections for four years and has over 20 years’ experience in the justice sector. Wayne has a PhD in criminology; his thesis addressed regional variation in sentencing in New Zealand’s District Court.
Two recent studies of socio-economic and reconviction outcomes for similar people sentenced to home detention and short terms of imprisonment found positive outcomes favouring the use of home detention.
Both studies provide evidence that rates of employment are modestly higher and rates of benefit uptake lower for those who served home detention, relative to outcomes for those released after serving short sentences of imprisonment.
The findings in relation to reconviction are mixed, with only small differences in re-offending rates found, and thus do not support a conclusion that either sentence type is more or less likely to result in subsequent reconviction.
Although both studies have limitations that mean the results and conclusions should be treated with caution, the positive socio-economic outcomes, much lower cost, and absence of negative reconviction effects, support the Department’s general practice of recommending home detention whenever practicable when a short sentence of imprisonment would otherwise be imposed.
The sentence of home detention, introduced in 2007, was explicitly intended in legislation to replace – where appropriate – short sentences of imprisonment. Around 3,300 sentences of home detention are handed down by sentencing courts each year, and around 1,600 people are serving sentences of home detention at any given point in time. The costs of administration for home detention are significantly lower than imprisonment: the all-inclusive cost of the sentence to Corrections is $63 per day per offender, vs $330 for sentenced prisoners. Further, the sentence is associated with relatively low rates of re-offending: for example, in 2018, the rate of reconviction was 18.9% within 12 months from sentence commencement, and the imprisonment rate was 6.6%; these figures compare to 43.3% and 29.7% for those released from prison after serving sentences of between 12 and 24 months. However, as is made clear below, simple comparisons of this nature can be misleading.
In the last few years, two unpublished studies (Business & Economic Research Limited (BERL), 2018 and Dixon & Morris, 2015) of socio-economic and reconviction outcomes for similar people sentenced to home detention and short terms of imprisonment have been produced. Both studies attempt to control for systematic differences between those sentenced to home detention and those sentenced to imprisonment. Both studies employed propensity score matching to ensure as far as possible that any differences arising from characteristics of the offenders or their offending were eliminated. It is generally accepted to be the best approach for studies of this type.
Both studies addressed the same question:
To what extent do offenders who serve a short term of imprisonment have different reconviction, employment and benefit receipt rates during and after their sentence than similar offenders who serve a sentence of home detention?
The studies differ in scope in two ways. Dixon & Morris (2015) limited the comparison to offenders serving their first prison or home detention sentence, BERL (2018) included all those serving a relevant sentence and accounted for prior imprisonment as a control variable. The time periods studied varied – Dixon & Morris (2015) covered cases sentenced in 2008 and 2009 whereas BERL (2018) covered those sentenced over six years from 2008 to 2013. Samples in both studies included men and women.
It is important to stress that propensity score matching can never fully overcome biases. There will almost always be what are termed “omitted variables”, variables that are likely to be important but that are not available. These variables are more important the more they are likely to be associated with outcomes and the more they are likely to be prevalent for either home detention or imprisonment.
Both studies recognise certain systematic differences between those serving either sentence. Those on home detention necessarily have suitable accommodation for the sentence, and this is likely (to an unknown degree) to be available post sentence. Those in prison are less likely to have the same level of accommodation. Accommodation is a protective factor against re-offending. Second, those on home detention who are living with others are more likely to have prosocial support; the other residents have had to agree to the person serving the sentence in their residence. Not all home detainees will have prosocial support but it is reasonable to think it will be more prevalent than for those serving a prison sentence. It is also a protective factor. The net result is that any difference between the two sentence types is likely to be overstated; a strong positive difference in favour of home detention is required before any firm conclusions can be drawn.
There may be other unmeasured factors that bias the comparison. Judges may be adept at systematically choosing people who will be successful on home detention. One such factor that is not controlled for in either study is the level of educational attainment/acquired skills that could both increase the likelihood of future employment and indirectly reduce the likelihood of re-offending.
Interpretation of the BERL (2018) study is made more difficult because it does not include all the supporting information and the reporting of the results varies between the models.
The choice of the measurement period or point in time
The validity of the comparison between the sentence types is dependent on making a fair choice of when to measure from (the start or end of a sentence) and whether to contrast the position at a point in time (for example, the proportion on a benefit at a fixed point) or across a period (for example, the cumulative proportion reconvicted within 12 months). A number of the contrasts in both studies suffer from methodological shortcomings.
The most significant shortcoming is the contrast in the BERL (2018) study that compares the position of home detainees measured from sentence commencement to short sentenced prisoners from release. The opportunity to offend is different for both groups; the restrictive nature of home detention suppresses offending during the course of the sentence. Conversely, comparison on this basis creates a mismatch in terms of the opportunity to be employed, with released prisoners having greater opportunity than those on home detention.
The contrast in both studies of reconviction rates measured from sentence commencement has a similar flaw. In both studies the effective sentence length for home detention was greater; those serving short sentences of imprisonment had more time free within the 12 month period in which to offend. In some cases prisoners are released on the commencement date because they have already served the time in custodial remand.
The preferred contrast for reconviction is “from release” for prisoners and “sentence end” for home detainees. Assuming relative equality in release and post detention conditions, the opportunity to offend is similar.
For employment and benefit receipt both studies measure from sentence commencement and BERL (2018) also measure from the end of home detention and release from prison. Both report the averages across time periods and graph the difference across time at fixed points (monthly) for measurement from sentence commencement, unfortunately BERL (2018) did not graph the results when measuring from completion/release. Using the average proportion across a period tends to be biased by home detainees being able to access the welfare system and to be employed during the course of the sentence. The bias tends to fall away as time passes, and the second and subsequent years’ results are a fairer indication of difference. To simplify the consolidation of the two studies, the difference at 12 months is highlighted and thereafter the averages for the second and third year after commencement are discussed.
Both studies show that after matching there is no or minimal difference between the groups in the rates of employment and benefit receipt before sentence commencement.
As expected, in the 12 months after sentence commencement, home detainees are much more likely than prisoners to be on a benefit. However, both studies indicate that about 12 months after sentence commencement the rates are about the same.
Dixon & Morris (2015) found that beyond 12 months from sentence commencement the rates of benefit uptake diverged. In the second year the average benefit rate for prisoners was about 50%, 5 percentage points higher than home detainees. In the third year, it was about 47%, 7 percentage points higher. The results for both years were statistically significant.
The BERL (2018) study did not fully replicate Dixon & Morris’s (2015) results. BERL (2018) found there was no difference 12 months from sentence commencement. It found the 1.3 percentage point difference in the second year in the average rate was not statistically significant, but the 2.7 percentage point difference in the third year (55.7% for prisoners versus 53% for home detainees) was statistically significant.
Consistent with the results for benefit receipt, both studies found that home detainees were more likely to be in employment. At the 12 month point Dixon & Morris (2015) found home detainees were about 3 percentage points more likely to be employed; BERL (2018) found a 5 percentage point difference – in both cases the result was statistically significant.
Dixon & Morris (2015) found that beyond 12 months from sentence commencement the difference in rates of employment increased. In the second year the average employment rate for home detainees was 38%, 6 percentage points higher than prisoners. In the third year the rate was about 40% and 8 percentage points higher than for prisoners. The results for both years were statistically significant.
The BERL (2018) study did not replicate Dixon & Morris’s (2015) results. BERL (2018) found the difference in rates narrowed but were still statistically significant. In the second year the average employment rate for home detainees was 32%, 3.7 percentage points higher than prisoners. In the third year the rate remained at 32% and was 2.8 percentage points higher than for prisoners.
Both studies report results for reconviction for new offending that occurred within one and two years of sentence completion/release. There is limited evidence of differences between the two sentence types.
Dixon & Morris (2015) found released prisoners were 4 and 5 percentage points more likely to be reconvicted after one and two years respectively.
BERL (2018) found that released prisoners were a statistically significant 3.4 percentage points more likely to be reconvicted within one year, but after two years the rates were almost identical (0.4 percentage points different and not statistically significant).
Discussion and conclusion
The studies provide support for the proposition that home detention offers integrative or reintegrative benefits relative to imprisonment. Home detainees are less likely to be on a benefit and more likely to be in employment in the medium term (12 months or more after sentence commencement).
Except for the BERL (2018) finding of a lower likelihood of reconviction for home detainees after one year, there was no statistically significant difference in the likelihood of reconviction. If we assume the omitted variables around availability of accommodation and higher levels of prosocial support would, if known, be more prevalent for home detainees, the actual results may be overstated. Based on the available evidence it is concluded that there is no difference in real reconviction rates between home detainees and those released from a short term of imprisonment.
This conclusion is consistent with the limited range of external research cited in the two studies. Therefore, the belief that home detention necessarily results in lower rates of reconviction is not supported. The raw or unadjusted rates differ due to differences in characteristics of the offenders serving the two types of sentence not the structure of the sentence types. However, what is apparent from these studies is that home detention sentences, which essentially allow offenders who would otherwise be imprisoned to remain in the community, certainly do not increase the likelihood of offending. Therefore, given the socio-economic benefits, and the fact that home detention has the advantage of being significantly lower-cost, it can be argued that home detention is a safe and cost-effective sanction of the courts.
What does this mean for Corrections? The Department’s general philosophy of recommending home detention whenever practicable when a short sentence of imprisonment would otherwise be imposed is sound. In those instances where home detention is feasible (there is a residence available) and there are no clear reasons for preferring imprisonment, home detention should be recommended.
BERL (2018). The case for home detention. Unpublished paper.
Dixon, S. & Morris, M. (2015). The impact of alternative sentences on adult offenders’ subsequent outcomes: Prison versus home detention. Unpublished paper.