Scottish Government

4. Home Detention CurfewStatistical Analysis of HDC

4.1 The statistical analysis considers three types of HDC decision: the decision to release a prisoner, the decision to refuse release, and the decision to recall a prisoner who is currently on an HDC licence. Table 4.1 provides a broad overview of these decisions and other relevant information for the period from July 2006 through March 2010, the time frame on which most of the statistical analysis is based. The base numbers used for the bulk of the analysis are presented in Annex A for reference.

Table 4.1 Summary of Key Aspects of HDC Use in Scotland 2006-2010

HDC Decisions*	16,084
Releases	7,292
% released	45%
% licences successfully completed	77%
Recalled	1,567
Recalls as % of releases	21%
Average daily population on HDC	331
% HDC releases who are women	11%
% Total prison receptions of women (2008/09)	7%
Main Offence Types on HDC (vs. % in prison for these offences on 30 June 2008)
Drugs	23% (14%)
Violence	19% (37%)
Dishonesty	18% (12%)
Misc. (incl. common assault, breach of the peace, etc.)	16% (9%)
Motor Vehicle	12% (3%)

*'Decisions' includes all 'in-process' refusals plus releases. By excluding some types of refusals, 'Decisions' is not exactly the same as all applications made to HDC. See paragraphs 4.10-4.13 for definitions of different types of refusals.

HDC Releases

4.2 The number of HDC decisions to refuse or release, in other words the volume of HDC workload, has generally increased over time, reflecting the larger population of prisoners involved with the scheme. Figure 4.1 shows the number of prisoners released on HDC licences between July 2006 and the end of March 2010. In total, there were 7,292 separate releases during this time, with the vast majority of prisoners (93.7%) experiencing only one period on HDC. At the other extreme, one prisoner is recorded as having five separate episodes on HDC. Following initial growth following implementation of the scheme, releases on HDC were highest in the period between Quarter 2 of 2007 and the end of 2008. More recently, and perhaps reflecting the stabilisation of the prison population more generally, the number of decisions to release prisoners has fallen slightly, to a level comparable with the first half of 2007.

Figure 4.1: Number of Prisoners Released on HDC 2006 to 2010

4.3 The profile of those released on HDC to a large extent reflects the profile of the Scottish prison population more generally, which is mainly male and predominated by younger people (around 40% are in their 20s). Table 4.2 provides summary data on HDC releases. Male prisoners make up approximately 90% of those released on HDC, and approximately 60% of those granted HDC were age 30 or under, while less than one in five were aged 41 or over.

4.4 The offence profile of those on HDC tends, on average, to be less serious than that of the overall prison population (see Table 4.1). In addition, those released on HDC do not appear to be exclusively drawn from a small range of offences. Around one third of those released were originally convicted of offences which are recorded under the heading 'Other Crimes'. This group is dominated by those convicted of Drug-related Offences (who account for just under one quarter of all prisoners released on HDC).

Table 4.2: HDC Releases 2006 to 2010, by Prisoner Gender, Age, Main Offence and Sentence Length

Prisoner Characteristic	Percentage of Prisoners Released on HDC
Gender
Female	11.1
Male	88.8
Age Group
20 and Under	15.3
21-30	43.6
31-40	25.1
41-50	12.0
51-60	3.5
61 and over	0.6
Main Offence of Conviction
Crimes of Dishonesty	17.6
Crimes of Indecency	<0.1
Fire-raising, Vandalism etc	2.6
Motor Vehicle Offences	11.9
Non-Sexual Crimes of Violence	18.5
Other Crimes	33.5
Miscellaneous Offences	15.8
Other Jurisdiction Charge	<0.1
Sentence Group on Release
< 6 Months	15.9
6 Months <12 Months	33.2
12 Months <18 Months	19.3
18 Months <24 Months	10.5
24 Months <36 Months	16.2
36 Months <48 Months	4.8
48 Months and Over (including lifers)	0.2

4.5 Prisoners released on HDC are generally serving short sentences, only 5% of those given an HDC licence were serving sentences of three years or longer. However, while those on sentences of less than six months make up around 15% of those accepted for HDC, only 46 prisoners on sentences of three months or less have ever participated in the scheme. Interview data suggest that the amount of time needed to process HDC applications will both discourage applicants serving this sentence length from applying, and make it difficult to complete timeously the applications of those that do (see section 4.3 for more discussion).

4.6 Figure 4.2 provides a snapshot of the total number of HDC releases by prison while Figure 4.3 shows the release rate by prison (releases in proportion to a prison's average daily population). The figures show first that, unsurprisingly, the largest establishments, which also have the largest pool of prisoners eligible for HDC (e.g. Barlinnie), make the most releases. However, Figure 4.3 shows that the number of HDC releases at a given prison establishment are not simply a function of how many people are eligible, and some of the smallest prisons (e.g. Inverness) are releasing, proportionately, the most prisoners.

Figure 4.2: Total Number of HDC Releases 2006 to 2010, by Establishment ^*

Figure 4.3: Ratio of HDC Releases to Average Daily Population 2006 to 2010, by Establishment ^*

^*There are no prisoner releases to HDC from Peterhead or Shotts Prison; figures for Open Estate include both Castle Huntly and Noranside where appropriate. Low Moss closed in 2007 and Addiewell Prison has only been in operation since late 2008.

4.7 The number of HDC releases from a particular prison is likely to be affected by a range of factors, such as the characteristics of its prisoners. For instance, it is not surprising that Glenochil's release rate is relatively low since it holds long term prisoners, and by far the greatest users of HDC are short term prisoners. [7] The higher release rates of Inverness and Aberdeen may be partly explained by the fact that, as suggested in interviews (see the qualitative analysis in section 4.2, and also Audit Scotland, 2008) these prisons have experienced higher than average levels of crowding. Finally, Cornton Vale, the women's prison, and the Open Estate both have higher than average levels of HDC releases. In the former case, this comports with usage patterns elsewhere (e.g. England and Wales) where women have higher rates of HDC release than men; in the latter case, it might be expected that the prisons which focus on supporting the transition of prisoners back to their communities would make greater use of all schemes of conditional liberty.

4.8 Prisoners granted HDC might be expected to place demands on the different agencies involved in CJA activities. Table 4.3 shows the number of releases to each Community Justice Authority (CJA) in Scotland, and creates a ratio of releases relative to the population size of each CJA. This shows that most CJAs are taking in similar proportions of prisoners (typically between 100 and 150 releases per 100,000 people). However, two CJAs, Glasgow and Southwest, are receiving considerably higher rates of releases (the analysis of releases by SIMD identifies some implications of this). In general, areas with relatively high rates of HDC releases are also the ones with high imprisonment rates.

Table 4.3: HDC Releases 2006 to 2010 by Community Justice Authority Area

Community Justice Authority	Total HDC Releases	Rate per 100,000 people*
Glasgow	1,293	219.7
Southwest	1,122	217.6
Tayside	569	142.4
Lanarkshire	845	132.6
Northern Strathclyde	779	124.6
Fife and Forth Valley	713	108.8
Lothian and Borders	998	106.3
Northern	855	102.5
Outwith Scotland/Address Unknown/No Fixed Abode	118	n/a

*Based on mid-year population estimates for 2009

Decisions to Refuse HDC

4.9 The data presented in this section provide a profile of decisions to refuse HDC release during the study period (i.e. the third quarter of 2006 to the first quarter of 2010). As this analysis is based on formally recorded decisions, it makes no reference to any informal influences which might affect whether a prisoner applies for an HDC licence.

4.10 Decisions to refuse HDC licences can be made for several reasons and at different stages. When attempting to contextualise patterns of HDC release, certain types of refusals may be seen as forming a more appropriate basis for analysis. In particular, those refusals which are due to statutory rules about who is eligible for HDC or where a prisoner chooses to opt out of the HDC decision making process - in other words, those cases which are removed from the system before an assessment of individual suitability has begun - are excluded from analysis. To this end, for the analysis presented in this section, we have classified refusals into two broad groups, 'Pre-process' Refusals and 'In Process' Refusals.

4.11 Pre-process Refusals cover those due to:

• Prisoner Withdrawal/Application Form Not Returned
• Statutory Exclusions [8]
• Not Having a Scottish Address [9]
• Previous HDC Recall [10]
• Cases with High/Medium Supervision which Result in Refusal with Review

4.12 In contrast, In Process Refusals include:

• Cases with High/Medium Supervision which Result in Refusal Without Review
• Adverse Community Assessment
• Outstanding Warrants
• Failed Risk assessment
• Other Reason for Refusal

4.13 For the most part, this distinction appears relatively clear-cut with refusal reasons such as Statutory Exclusions, being ineligible due to a previous HDC recall, and the prisoner withdrawing from the system being clearly distinct from reasons which can be seen as part of the assessment of a prisoner's application for HDC. The classification of refusals due to concerns about prisoners requiring High or Medium Supervision is more complicated, however. Prisoners rejected due to concerns around supervision level can be refused HDC outright, or be offered a review at a subsequent date. It seems appropriate that any measure of refusals which is to be used to contextualise release decisions should be concerned with final decisions resulting in refusal, and as such it was decided to exclude from the In Process Refusal category refusals on the grounds of supervision level where a review was granted.

4.14 In total there were 6,043 Pre-process Refusals and 8,792 In Process Refusals recorded during the period under consideration (see Annex A for additional breakdowns). Figure 4.4 shows that Pre-process and In Process Refusals rise and fall largely in tandem, and that there was a notable jump for all refusals in the middle of 2008, after which a higher plateau of refusals has been established.

Figure 4.4: HDC Refusal Decisions 2006 to 2010

4.15 Table 4.4 summarises reasons for refusals between 2006 and 2010. Within the category of In Process Refusals, no single reason dominates. In contrast, the Pre-process Refusal category shows a strong bias towards prisoners who are denied due to Statutory Exclusions and to a slightly lesser extent, prisoners refused on the grounds of their supervision level.

Table 4.4: Reasons for Refusal of HDC 2006 to 2010

Reason for Refusal	Percentage of Refusals
Pre-process Refusals
Application Form Not Returned	2.3
High / Medium Supervision (Review Granted)	35.8
No Scottish Address	1.8
Previous HDC Recall	8.6
Prisoner Withdrawal	5.0
Statutory Exclusions	46.3
In Process Refusals
Adverse Community Assessment	21.3
High / Medium Supervision (HDC Refused)	29.9
Outstanding Untried Warrant	5.5
Risk Assessment Failure	23.1
Other Reason	20.1

4.16 Figure 4.5 illustrates the changing mix of reasons for a Pre-process Refusal, while Figure 4.6 provides the same information for In Process refusals. Leaving aside for a moment the striking growth in refusals on the grounds of a statutory exclusion or a prisoner's supervision level, it is notable that the number of prisoners excluded from the system due to being recalled when on a previous period of HDC release has increased over time. [11]

4.17 In both Figures 4.5 and 4.6, rapid growth in refusals on prisoner supervision level grounds from the middle of 2008 can be clearly observed. It would appear that it is this growth in recorded refusals due to prisoner supervision level, which is mostly responsible for the overall growth in refusals identified in Figure 4.4. The data themselves do not tell us why this sudden growth in recorded refusals occurred. One potentially relevant change at this point was the extension of the maximum period a person could be on HDC from 4.5 to six months. The effect of this would be to bring forward the HDC qualification date for some prisoners, which might mean that some prisoners who would eventually be downgraded to a low risk level under the Prisoner Supervision System would still be at medium or high risk levels at the point when they could first be considered for HDC.

Figure 4.5: Reasons for Pre-process HDC Refusals 2006 to 2010

Figure 4.6: Reasons for In Process Refusals 2006 to 2010

4.18 It should be noted that it is possible for prisoners to be refused HDC on multiple occasions. Reflecting this, 13.3% of prisoners are recorded as having been refused HDC on more than one occasion. However, less than 2% received three or more refusals suggesting that prisoners are not generally constantly applying for, and being rejected from HDC.

4.19 The remainder of the analysis of refusals will focus on In Process Refusals, the most appropriate base for comparing to releases, with information broken down by the two periods either side of the middle of 2008 (i.e. second quarter), when the large spike in refusals occurred.

4.20 Table 4.5 provides a breakdown of refusals by prisoner characteristic. As with the data on HDC releases, this breakdown largely reflects the make-up of the wider prison population, for instance, around nine out of ten prisoners refused HDC are male. Similarly, those refused HDC are generally young, and on shorter sentences. However, reflecting the patterns identified with regards to HDC releases, prisoners on particularly short sentences are again notable for their absence in the data, with those on sentences of three months or less accounting for only 4.4% of prisoners refused HDC.

4.21 The overall ratio of releases to refusals was 1.2 before mid 2008 but only 0.6 from this point onwards. This means that prior to the middle of 2008, just over half of prisoners who made it to a final decision were released on HDC (for every prisoner receiving a refusal, 1.2 were given a release). In contrast, from the middle of 2008 onwards, only around 37% of decisions resulted in a release.

4.22 The last two columns of Table 4.5 present release ratios for each prisoner characteristic given in Table 4.2. [12] All but one of the post-mid 2008 release ratios in Table 4.5 are lower than the equivalent figures for the earlier period. The one exception to this is those prisoners serving sentences of four years or greater, though this amounts to less than 30 decisions, too small a number to draw any conclusions.

Table 4.5: Comparison of HDC Refusals and Release Rates before and after mid 2008

	Percentage of Refusals*		Ratio of Releases to Refusals*
Prisoner Characteristic	2006 (Q3) - 2008 (Q2)	2008 (Q3) - 2010 (Q1)	2006 (Q3) - 2008 (Q2)	2008 (Q3) - 2010 (Q1)
Gender
Female	11.4	7.6	1.2	0.9
Male	88.6	92.4	1.3	0.6
Age Group
20 and under	12.5	16.9	1.8	0.4
21-30	50.5	44.2	1.0	0.6
31-40	25.5	25.1	1.2	0.6
41-50	9.5	10.9	1.4	0.7
51-60	1.8	2.5	2.1	0.9
61 and over	0.3	0.4	2.1	0.9
Main Offence of Conviction
Crimes of Dishonesty	27.4	24.2	0.8	0.4
Crimes of Indecency	0.1	0.1	1.5	0.0
Fire-raising, Vandalism etc	2.5	2.8	1.3	0.6
Motor Vehicle Offences	9.6	6.0	1.6	1.1
Non-Sexual Crimes of Violence	14.2	14.1	1.7	0.8
Other Crimes	22.1	24.8	1.8	0.8
Miscellaneous Offences	23.8	27.6	0.7	0.4
Other Jurisdiction Charge	0.2	0.1	0.3	0.0
Sentence Group
< 6 Months	36.8	31.1	0.6	0.2
6 Months <12 Months	31.4	35.5	1.2	0.6
12 Months <18 Months	13.7	14.7	1.8	0.7
18 Months <24 Months	6.4	7.1	1.9	0.9
24 Months <36 Months	8.3	9.2	2.3	1.1
36 Months <48 Months	2.9	2.1	2.0	1.4
48 Months and over (including lifers)	0.2	0.2	0.2	1.0

*Refusals mean specifically In Process Refusals, as explained in para.4.19.

4.23 However, looking across Table 4.5 as a whole, it is possible to make several observations. Firstly, there are proportionately more releases among older prisoners. Likewise, this release ratio is generally higher for prisoners on longer sentences. There is no clear evidence of a gender difference in granting HDC, with no consistent pattern appearing across the two different time periods. With regards to the offence for which a prisoner was originally convicted, it is hard to make any detailed statement because each offence category combines a wide range of different offences, and as shown in Tables 4.1 and 4.4, some categories contain very few cases. However, the release rates associated with Crimes of Dishonesty and Miscellaneous Offences (both categories which contain several hundred decisions) do appear notably lower than those for other types of convictions. [13]

4.24 Overall, what can be concluded from this data is that after the middle of 2008, a lower proportion of those applying for HDC are being released. This suggests that the expansion of the scheme to allow for earlier consideration of release, which occurred in 2008, did not result in a significant increase in releases.

4.25 Figure 4.7 shows the pattern of In Process refusals across the two periods either side of the middle of 2008, while Figure 4.8 presents establishment data of the release to refusals ratio for the two periods. Reported numbers of releases at some establishments does raise some issues about the robustness of the data. [14]

4.26 The majority of establishments have similar total numbers of refusals across both time periods. However, a handful of prisons (Polmont, Perth and most notably Barlinnie) have recorded substantially more refusals in the second period compared to the first. The overall increase in refusals recorded either side of the middle of 2008 would appear to be a result of the differences associated with these establishments.

4.27 Figure 4.8 shows interesting differences within individual prisons, which are likely linked to changes in decision making practices from 2008 onwards. Release ratios for Barlinnie, Polmont and Perth are much lower after the middle of 2008. The apparent outlying nature of Polmont in the period prior to the middle of 2008 again raises questions about data integrity. During this time, Polmont has in excess of 2.5 recorded HDC releases for each recorded refusal. Although this figure may, to some extent, be explained by the unique nature of Polmont as a Young Offenders Institution, this figure is in marked contrast to the figures for other establishments, and the figure for Polmont itself from the middle of 2008 onwards.

Figure 4.7: HDC In Process Refusals pre and post mid 2008, by Establishment

HMP Addiewell opened in December 2008 and therefore provides no data for the earlier of the two periods; as noted, the Open Estate and Glenochil recorded no refusals.

Figure 4.8: HDC Ratio of Releases to In Process Refusals pre and post mid 2008, by Establishment

HMP Addiewell opened in December 2008 and therefore provides no data for the earlier of the two periods; as noted, the Open Estate and Glenochil recorded no refusals .

Decisions to Recall Prisoners on HDC

4.28 Of the 7,292 releases on HDC, 1,796 resulted in a prisoner returning to custody before their period on HDC was completed, giving an overall return rate of 23%. Prisoner returns are recorded under two headings, Readmissions (typically voluntary return to custody) which account for 13.4% of prisoner returns, and Recalls (return to custody due to breaches of conditions) which account for 86.6% of prisoner returns.

4.29 The rate of recalls between 2006 and 2010 was 21%. This rate is high in relation to the rate for England and Wales which was around 12% in 2009 (Ministry of Justice 2010). [15]

4.30 Table 4.6 provides a breakdown of the reasons why prisoners on HDC were recalled. Around two thirds of recalls relate to prisoners breaching licence conditions or not observing the terms of their curfew. Just over 16% of recalls are due to interfering with, or preventing the installation of, the monitoring equipment required at a prisoner's home address. Just 101 recalls (7% of all recalls) were due to the serving of a new warrant [16]. Another 6% were recalled when the householder of their home address withdrew their consent for them to live there.

Table 4.6: Reasons for Recall among Prisoners on HDC 2006 to 2010

Reason for Recall	Percentage of Recalls
Breach Of Licence Conditions	23.5
Serious Breach Of Curfew (>6 Hrs)	37.6
Minor Breach Of Curfew (>2<6 Hrs)	1.4
Cumulative Breach Of Curfew	6.0
Assault (Or Threat) On Contractor	0.3
Intentional Damage To Monitoring Equipment	12.4
Preventing Installation Of Monitoring Equipment	3.9
Preventing Maintenance Of Monitoring Equipment	0.2
New Warrant Served	6.5
Presence At A Restricted Location	0.1
False Breach (SPS/Serco Error)	0.4
Withdrawal Of Householder Consent	6.1
	1.5

4.31 Table 4.7 provides a summary and recall rates for prisoners with particular characteristics. Rates of recall for male and female prisoners are similar. However, recall rates amongst younger age groups appear to be substantially higher than for older ones, a finding consistent with research on HDC use elsewhere. The recall rates associated with Crimes of Dishonesty, Miscellaneous Offences (just over 70% of these are petty assault and 20% are breach of the peace) and Fire-raising and Vandalism appear notably higher than those for Non-Sexual Crimes of Violence and Other Crimes (primarily made up of drugs offences).

Table 4.7: Characteristics of Individuals Recalled from HDC and Associated Recall Rates

Prisoner Characteristic	Percentage of Recalls	Recall Rate
Gender
Female	10.7	20.8
Male	89.3	21.6
Age Group
20 and Under	18.8	26.4
21-30	48.2	23.8
31-40	22.6	19.3
41-50	8.7	15.5
51-60	1.5	9.6
61 and over	0.2	7.3
Main Offence of Conviction
Crimes of Dishonesty	22.6	27.5
Crimes of Indecency	0.0	n/a
Fire-raising, Vandalism etc	3.4	28.3
Motor Vehicle Offences	9.0	16.4
Non-Sexual Crimes of Violence	18.6	21.8
Other Crimes	24.8	15.9
Miscellaneous Offences	21.5	29.0
Other Jurisdiction Charge	0.1	50.0
Sentence Group on Release
< 6 Months	10.7	14.5
6 Months <12 Months	35.7	23.2
12 Months <18 Months	21.3	23.8
18 Months <24 Months	12.4	25.5
24 Months <36 Months	15.4	20.3
36 Months <48 Months	4.4	19.7
48 Months and Over (including lifers)	0.0	n/a
OVERALL RECALL RATE		21.5%

4.32 There is no clear linear pattern between the sentence length of an individual, and their likelihood of recall, with the lowest recall rates associated with prisoners who have served sentences of less than six months, while the next lowest recall rate is associated with those who have served between three and four years.

4.33 The most notable point about differences between recall rates by the establishment they were released from (Figure 4.9) is that the lowest recall rate is associated with those individuals released from the Open Estate. Such a finding is of little surprise given one of the major functions of the Open Estate is to prepare prisoners for a return to community living, and how prisoners at the Open Estate can be generally expected to have lower risks of re-offending. Still, this data may provide some indirect evidence of the 'success' of open prison, though it must be remembered that the age and offence profile of the Open Estate means its population is likely to have a lower rate of recall regardless of how well the regime is working.

4.34 At the opposite end of the distribution, the relatively high recall rate associated with Polmont is likely an aggregate level reflection of the higher risk of recall for younger prisoners and again serves to illustrate the point of how differences between establishments may reflect differences in their respective populations.

Figure 4.9: HDC Recall Rate 2006 to 2010, by Establishment

Success Rates by Length of Time on HDC

4.35 In this section we present the analysis of a sample of the main dataset, concerning 1,058 prisoners released on HDC licences in the six months between 1 July 2008 and 31 December 2008. This dataset includes additional variables around the length of time a prisoner spent on HDC, and more detailed information about the location of their home address during their time on HDC.

4.36 Around one-fifth (21.9%) of this sample were recalled due to breaching their licence terms at any time during their period on HDC, a figure comparable with the recall rate for the overall study sample.

4.37 Using information about how long an individual spent on HDC, the length of their HDC licence, and the reason for the ending of their time on HDC (liberation or recall due to breach), it was possible to construct an indicator of what proportion of their HDC release period each individual served before recall. A prisoner who avoided recall received a proportion of 1.0, i.e. they successfully completed all their time on HDC; a hypothetical prisoner recalled on their first day of an HDC licence would receive a value of 0.0, i.e. they successfully completed none of their HDC period. In this respect, this indicator is more nuanced than the indicator used in the previous section, as it considers an element of time to recall, rather than simply indicating whether a prisoner is recalled.

4.38 Table 4.8 presents two simple regression models for linking the proportion of HDC licences successfully completed to the overall length of an individual's HDC licence. The first model analyses success rate on HDC by the length of time a person was on HDC. This model suggests that, on average, the longer a person is on HDC the more likely they are to be recalled to custody. The second model, a quadratic model, includes not only the length of an individual's HDC licence, but, also the square of this value, which appears to provide a slightly improved approximation of the pattern of recalls within the data. [17] The application of the quadratic model, illustrated in Figure 4.10, shows like the first model that the success rate of those on HDC decreases the longer the period of an HDC licence, but that this relationship begins to reverse for those on licences of 130 days or longer.

Table 4.8: Regression Models of the Proportion of HDC Licences Completed to Overall Length of an Licence Period

Variable	Licence Length Model	Quadratic Model
Constant	1.220 (0.088)*	1.507 (0.163)*
HDC Licence Length	-0.002 (<0.001)*	-0.010 (<0.001)*
HDC Licence Length Squared		0.001 (<0.001)*
Bayesian Information Criterion	-6854.527	-6852.111

*Indicates significance at the 0.05 level (standard errors given in brackets).

Figure 4.10: Predicted Proportion of Licences Successfully Completed by Length of HDC Licence

Success by Geographic Area and Level of Neighbourhood Deprivation

4.39 The dataset involving just those prisoners released on HDC licences in the second half of 2008 also included additional geographic information which allowed for analysis to be conducted at a small area level. In particular, it was possible to identify in the Data Zones of 95% of the registered home addresses within this sample. Data Zones are small area geographic identifiers, each of which contains between 100 and 500 household residents. The boundaries of each zone are intended, where possible, to reflect physical boundaries and natural communities. As far as possible, a single Data Zone will contain households with similar social characteristics.

4.40 Data Zones are the major unit of reference used in the Scottish Index of Multiple Deprivation (SIMD). Figure 4.11 suggests that the home addresses of those released on HDC licences are disproportionately concentrated in deprived areas. Taking a simple binary indicator of whether an individual was recalled to custody during their period of HDC, Table 4.9 presents T-tests to compare prisoners recalled to custody and those who successfully complete HDC, in terms of their average ranking on different SIMD domains (lower values represent more deprived areas). Although it is important to be careful in interpreting this area level analysis (as highlighted above, the risk of recall appears related to different individual level characteristics, and it is likely that individuals with similar characteristics will live in similar types of areas), Table 4.9 suggests a clear pattern, with those individuals recalled to custody appearing, on average, to have home addresses in more deprived areas. This pattern is present not only with reference to an area's overall standing on SIMD, but also for the majority of separate domains which contribute to the overall index. The exception to this pattern is that there appears to be no significant difference between those who successfully complete HDC, and those recalled, in terms of their score on the Geographic Access to Services component of the index.

Figure 4.11: HDC Releases by Scottish Index of Multiple Deprivation Ranking, July through December 2008

Table 4.9: Comparison of SIMD Domains for Those Completing versus Recalled from HDC

SIMD Domain	Mean Rank for Successful HDC Period	Mean Rank for Recall Due to Breach	T-Statistic
Overall SIMD	1800.2	1406.8	3.5*
Employment	1859.7	1437.6	3.8*
Income	1817.1	1447.5	3.3*
Health	1796.4	1458.4	3.1*
Education	1845.5	1523.3	2.9*
Housing	2258.1	1939.4	2.7*
Crime	2064.5	1753.3	2.7*
Geographic Access	3727.9	3747.4	-0.149

* Significant at the 0.05 level.

Tests based on unequal variances across groups.

Significance based on two-tail test.