Housing and Reoffending: Supporting people who serve short-term sentences to secure and sustain stable accommodation on liberation

ANNEXE 4: STATISTICAL ISSUES

1. The purpose of this Annexe is to set out the findings of the research on a number of statistical issues which, taken together, make it difficult to obtain a clear picture of the scale of homelessness among people entering and leaving custody. This Annexe will cover:

• Data on people who are homeless on admission to prison.
• Data on people who lose their accommodation while in prison.
• Data on Housing Benefit claims in custody.
• Data from the Core Screen and other in-prison processes.
• Data on homelessness prevention work in prison.
• Data on people leaving prison.

Data on people who are homeless on admission to prison

2. Data prepared for this research by the SPS (pghs A3:5-11) indicated that, on 3rd April 2015, a total of 196 individuals in custody and serving short sentences (out of 3249) had their addresses recorded on admission as “no fixed abode” (NFA).

3. However, staff (in discussions) suggested that this figure was unreliable, as individuals sometimes provided the police, Procurator Fiscal or Sheriff Clerk with a false accommodation address (when they were homeless), either so that documents could be served to the address, or to try to influence Sheriffs’ sentencing. Prison staff indicated that they tried to check whether the address provided by the court was correct, but this relied on an individual’s co-operation (which may not be forthcoming). It was suggested that, as a result of these issues, the NFA figure was likely to be an underestimate.

4. The periodic Scottish Prison Survey asks those in custody (both serving short and long term sentences) about their living arrangements before admission. While this provides a valuable insight into people’s previous housing circumstances, it cannot provide a definitive overall number of people coming into prison who are already homeless, as result of the categories used.

5. In the 2103 survey (Carnie et al, 2013), the SPS used a number of “common sense” categorisations, designed to give a clearer picture to policy makers and staff in prisons of individuals’ living circumstances than might be possible just by using the term “homeless” (which, although it has a legal definition, has a less clear meaning in general use). This survey allowed respondents to choose from any or all of the categories available, making it impossible to identify the number who could be considered homeless on the basis of a legal definition. SPS informed the research team that a number of changes had been made to the 2015 survey both to amend existing, and add new categories.

Data on people who lose their accommodation while in prison

6. Using current data, it is not possible to ascertain the number of individuals who lose their tenancy or accommodation while in custody. While the 2013 Scottish Prison Survey (Carnie et al, 2013) covered this issue, the information is not reliable as an overall indicator, as it could only cover those who knew that they had lost their accommodation by the time the survey was completed. This would not include those who lost their accommodation at a later point in their sentence (for example, as a result of abandonment or eviction proceedings).

Data on housing benefit claims in custody

7. At present, the DWP Stat-Xplore system, which otherwise provides considerable detail in relation to the breakdown of Housing Benefit claims, does not identify the number of claims which ended because someone was imprisoned, or which were started in prison. In addition, the geographical basis of published data is the area in which the property is located, not the location of the claimant at the time of the claim.

Data from the Core Screen and other in-prison processes

8. The Core Screen and other prison processes to identify housing (and other needs) were described in detail in Annexe 2 (pghs A2:55-57).

9. Data is available on the overall number of referrals logged onto PR2. The data made available by the SPS to the research team showed that, in 2013-2014, 9849 housing-related referrals were made (covering those serving both short and long sentences).

10. However, it was clear that the way the data is currently organised makes an accurate assessment of the nature of referrals very difficult. This is not a fault of the system per se. Different approaches appear to have evolved within individual prisons, probably to meet local operational needs. This has, however, led to difficulties in providing a dataset which can give a consistent national picture.

11. Data in relation to referrals was found to be recorded inconsistently across different prisons. Only around half of the referrals recorded the name of the agency to which the referral was being made, while the remainder had been given a generic label of “housing agency”.

12. Data on “outcomes” also appeared to be unreliable, as much of the information related not to outcomes, but, for example, to cases being “listed for assessment” and “on-going”. In 14% of cases, the “outcome” was listed as “interviewed” and in 32% of cases, “advice given”. Overall, 16% of cases had “missing” data.

13. SPS informed the research team that, as part of the development of the Core Screen, it was intended that changes would be made to the way referrals were recorded on PR2.

Data on homelessness prevention work in prison

14. As noted in Annexe 3 (pghs A3:50-58), some local authorities (and third sector housing staff) carried out homelessness prevention work at the point of admission to custody, or soon after. It is not possible, at present, to ascertain how many individuals received homelessness prevention support at that stage, or how many fell within the scope of Housing Options.

15. The Scottish Government’s PREVENT 1 statistical return collects information from local authorities to monitor Housing Options / homelessness prevention in Scotland. However, it does not identify where, or in what circumstances Housing Options assessments are carried out. The returns only capture work done by local authorities, and, therefore, in the context of work in prison, would not reflect work done by Sacro, Shelter or others (pghs A3:13-48).

16. Data for 2014-15 (Scottish Government, 2015h) suggest that, across Scotland, 3,495 Housing Options approaches were made during that period where the reason for the approach was discharge from prison or other institutional care (about 6% of all cases). This covered those who had served both short and long sentences. However, it was not possible to ascertain how many of these assessments were undertaken in custody, and how many following liberation.

17. Published PREVENT 1 data also indicate considerable regional variations in the recorded numbers of Housing Options assessments where prison was the last known address (and which cannot be explained by relative population size). However, in discussions, it was suggested that this may reflect the fact that different local authorities were at different stages in terms of using the system, and that, over time, a more consistent dataset would become available.

18. It was noted that the Tayside Housing Options Hub, Scottish Government and the SPS are discussing ways of using PREVENT 1 data to improve management information about those entering or leaving custody.

Data on people leaving prison

19. There was found to be little reliable evidence on the number of people who serve short sentences who are homeless on liberation. The most commonly quoted estimate was made by the Home Office in 2003/4 (Home Office, 2005), and suggested around 30%. This estimate is not strictly comparable to Scotland, as it used different definitions of “short term” and “homeless”. The Scottish Prison Survey (Carnie et al, 2013) found that about a third of those surveyed did not know where they would be living on liberation, but this was based on their expectations, not their experience on liberation.

20. The SPS gathers statistics on the number of people who leave custody and who are classified as having “no fixed abode” (referred to as “NFA”). A total of 448 liberations from short sentences (out of nearly 10,000) were NFA in 2011-2012 (Scottish Government, 2013)^[24]. This represents only 4.5% of those liberated from short sentences.

21. This cannot, however, be taken as a proxy for those who were homeless on liberation. This is as a result of the definitions used. An individual with a pre-allocated place in a hostel or other temporary accommodation would not be counted in the NFA figure, even though they had been assessed and accepted as homeless. However, an individual required to present as homeless at a housing office on the day of liberation may be recorded as NFA, even though the research found they would be virtually certain to have at least a temporary address by the end of that day.

22. At present, the main source of information on homelessness among those who have served sentences comes from local authorities’ HL1 returns. This is helpful, and indicative of the scale of the problem in general terms, although it only covers those who approach the local authority (and would not include, for example, an individual sleeping on a friend’s floor who had not asked for support).

23. There are issues in both the collection process and the analysis which mean that these published statistics are unlikely to provide a comprehensive view of homelessness as it affects those leaving custody. Three main issues were noted:

• Individuals are not compelled to disclose the reason for becoming homeless, and may choose not to cite leaving prison. Individuals could, for their own reasons, cite other reasons, such as the breakdown of a relationship, or eviction.
• There are likely to be issues with the timing of some individuals’ applications, which make it likely that other reasons (rather than leaving prison) may be cited. Housing staff (and reintegration staff) reported that some of those leaving custody did not present as homeless until days, or even weeks afterwards, even though they had not been living in stable accommodation in the intervening period.
• The Scottish Government’s published statistics derived from HL1 returns do not differentiate between those leaving custody and those leaving other forms of institution (although the underlying return does so).

24. Scottish Government statisticians provided the research team with a further breakdown of HL1 returns which identified prison leavers separately. It was noted, however, that there were likely to be reliability issues with the data. For that reason, this breakdown has not been included in this report.

25. At present, there is no means of identifying the housing outcomes for those who have lost housing as a result of a custodial sentence. This group is not separately identified in relation to, for example, the outcomes of the Housing Options process (described in Annexe 2), nor homelessness applications, nor needs assessments. Nor is this group identified specifically in the HL2 or HL3 returns, which deal with those in temporary accommodation.

26. Although some participants indicated that housing associations accepted referrals for people recently liberated from custody using the Section 5 referral process^[25] as well as by other nomination arrangements, published data does not report those leaving custody as a specific group. There is, therefore, no way of knowing the overall number of those leaving custody who were provided with accommodation by a housing association. Discussions with housing associations, however, suggested that the annual total was likely to be very small.

27. It was clear from the survey and discussions in this research that relatively little information was gathered by housing services beyond that specified in the PREVENT 1 and HL series of returns.

28. Other services indicated that they gathered and maintained information on a case by case basis about housing issues and outcomes. This data, gathered either for operational or monitoring reasons, did not appear to be generally consistent with data gathered by public bodies (for example in terms of the categories or definitions used which may not match those used on the PREVENT 1 and HL returns, or which may not, for example, differentiate between types of accommodation, or between individuals with different sentence lengths).

29. It was noted that much information was qualitative in nature, and could be incomplete (for example, in relation to whether or not identified issues were resolved). It was suggested that it was not routinely analysed, interpreted or used. For all these reasons, this material would be likely to be useful only for indicative, or review purposes.

Evaluations

30. The survey found two evaluations of housing-specific projects: the Community Reintegration Pilot (Scottish Government, 2014d) (pghs A2:76:77); and the SPAN partnership project (Shelter, 2015b), involving Shelter, Sacro and Inverness CAB.

31. Some reintegration organisations indicated that they were in the process of undertaking evaluations of their overall service (including the national PSPs, the Low Moss PSP and some smaller services). While some indicated that housing was a strand of the evaluation, few appeared to focus on the role of, or outcomes relating to, housing alone, or in relation to reintegration and the promotion of desistance.

32. Glasgow City Council noted that it was in the process of carrying out a comprehensive review of its homelessness operations (but this had not been published at the time of the research). Other local authorities indicated that they had internally reviewed their Housing Options approach. A few indicated that reports had been provided to committees within the authority on the performance of various housing-related services, although these were not based on external evaluations.