The Scottish Government

« Previous | Contents | Next »

Listen

Tracking Homelessness: A Feasibility Study

CHAPTER TWO: THE NEED FOR LONGITUDINAL RESEARCH ON HOMELESSNESS

2.1 This chapter clarifies the parameters of this research by outlining the definition of homelessness to which we are working. It also describes the needs of policy for longitudinal research on homelessness, and explains the limitations of existing longitudinal research in this area.

DEFINING HOMELESSNESS

2.2 In this study we have adopted the HTF definition of homelessness (Scottish Executive, 2001). This includes all groups currently defined as statutorily homeless, namely those:

• without any accommodation in which they can live with their families;
• who cannot gain access to their accommodation or would risk domestic violence by living there;
• whose accommodation is 'unreasonable', or is overcrowded and a danger to health; and
• whose accommodation is a caravan or boat and they have nowhere to park it.

2.3 In addition, the HTF defined people experiencing any of the following situations as homeless, whether or not these circumstances fit within current statutory definitions:

• rooflessness - people without shelter of any kind, including those sleeping rough, victims of fire and flood, and newly arrived immigrants;
• houselessness - people living in emergency and temporary accommodation provided for homeless people, such as night shelters, hostels and refuges;
• unsuitable accommodation - people residing in accommodation, such as Bed & Breakfast premises, which is unsuitable as long stay accommodation;
• institutions - people staying in institutions because they have nowhere else to go;
• insecure accommodation - this group includes:

- tenants or owner-occupiers likely to be evicted (whether lawfully or unlawfully);
- people with no legal rights or permission to remain in accommodation, such as squatters or young people asked to leave the family home; and
- people with only a short-term permission to stay, such as those moving around friends' and relatives' houses with no stable base;

• involuntary sharing - people involuntarily sharing accommodation with another household on a long-term basis in housing circumstances deemed to be unreasonable.

2.4 We have endeavoured to develop potential designs for longitudinal research which will encompass as many of these dimensions of homelessness as possible. In addition, we have tried to take account of hidden as well as visible aspects of homelessness in the proposed methodologies. According to Webb (1994), people can be considered 'visibl y homeless' if:

• their homelessness is recorded in official statistics, that is, they have applied to and/or have been accepted by a local housing authority as homeless; or
• they are in contact with homelessness agencies and/or staying in homeless hostels; or
• they are sleeping rough in visible areas or on known sites, for example, in the city centre.

2.5 The hidden homeless are therefore those whose homelessness is not visible in these respects. Households involuntarily sharing accommodation and people experiencing intolerable and/or insecure housing circumstances are the main groups usually referred to as hidden homeless. However, the other categories of homelessness described above can also take hidden forms, e.g. even rooflessness can be hidden if people sleep rough away from known sites.

THE NEEDS OF POLICY FOR LONGITUDINAL RESEARCH

2.6 Homelessness research in the United Kingdom and in the United States has traditionally been cross-sectional in nature. That is, it has provided only a snapshot of homeless people's circumstances at a particular point in time. In the past few years longitudinal research, which tracks the progress of a particular group (or cohort) of people over time, has been identified as a key priority for future work (Fitzpatrick et al., 2000; Anderson & Tulloch, 2000). There are two main reasons why longitudinal research is considered to have particular value for homelessness policy.

2.7 First, longitudinal data is required to aid general understanding of the experience of homelessness, and of what happens before and after homeless episodes. Such data is necessary, for example, to provide robust estimates of the numbers of people who experience short-term, long-term or episodic homelessness. Longitudinal research also enables one to identify more precisely the circumstances which lead households to move in and out of homelessness, with Rossi (1991) and others highlighting the importance of such leverage points in the development of preventative policies. More generally, longitudinal research can help one to understand chaotic lives better because data is collected closer to when significant events happen.

2.8 Second, longitudinal data is required to develop more robust evaluations of particular interventions and initiatives aimed at tackling homelessness, including those introduced as a result of the work of the HTF. The existing evaluative research on homelessness interventions in the UK is very weak, and in particular there are very few longitudinal evaluations which enable long-term outcomes to be assessed (Fitzpatrick et al., 2000; Anderson & Tulloch, 2000). Both the life skills training (Jones et al., 2001) and intermediate accommodation (Kennedy & Lynch, 2001) research conducted for the HTF highlighted the lack of evaluative evidence on the relative merits of different models of resettlement provision. The HTF's 'routes out' research confirmed the need for longitudinal evidence on people's pathways out of homelessness to inform policy priorities and investment decisions:

"Increasingly, research and evaluations emphasise the need for systematic tracking of rehousing and resettlement outcomes, emphasising this is essential to identify the barriers to sustainable resettlement and 'what works' in service provision." (Rosengard et al., 2002:13)

2.9 Leading authors in the US have also stressed the importance of longitudinal evaluations of programmes aimed at tackling homelessness. Rossi (1991, p.1049), for example, comments: "Because change can only be defined in longitudinal terms, the implication is that evaluations are longitudinal observations". He acknowledges the particular difficulties of making repeated contacts with homeless people, but argues that this does not mean that longitudinal research with this group was impossible " …only very difficult and therefore very expensive" (p.1041). In a similar vein, Cohen et al. (1993, p.332) argue: "Obviously research studies need to employ longitudinal designs and short and long-term follow-up periods in order to evaluate effectiveness comprehensively".

2.10 Despite a burgeoning of longitudinal research on homelessness in the US in the 1990s, researchers there continue to argue that there are insufficient long-term studies of resettlement interventions to allow firm conclusions to be reached on their effectiveness (Shinn & Baumohl, 1999). Barrow & Zimmer (1999), for example, contend that studies of transitional housing should include follow-up work extending at least one year, and ideally two, from the point of entry. Similarly, Robertson et al. (1999) argue that more longitudinal research is needed to help develop understanding about the resources required to enable young homeless people to achieve positive outcomes in adulthood.

2.11 In assessing the policy value of any particular design for conducting longitudinal research on homelessness several key points must be considered:

• the breadth of the research questions that can (at least potentially) be addressed;
• the methodological robustness which is likely to be achieved (e.g. statistical representativeness, retention rates etc.);
• which groups within the homeless population would be covered. For example, studies using local authority records will tend to be more successful at picking up families with children, whereas those which draw samples from voluntary sector homelessness services will include mainly single people. Similarly, different strategies are probably needed to identify and track hidden and visible homeless people, and possibly different age groups; and
• whether the tracking methods proposed will most easily pick up those with positive or negative outcomes (i.e. those who have or haven't moved out of homelessness).

2.12 These points are addressed in Chapter 4 in relation to each of the potential research designs.

THE LIMITATIONS OF CURRENT LONGITUDINAL RESEARCH ON HOMELESSNESS IN THE UK

2.13 Some recent research on homelessness in the UK has attempted to incorporate a more dynamic, longitudinal dimension, focusing on people's pathways through homelessness (see Anderson & Tulloch, 2000, for a review). However, the great majority of these studies have relied on retrospective accounts from currently homeless people. This introduces two important limitations into the data produced:

• its retrospective nature prompts concerns about the reliability of respondents' recall of past events, particularly for those with chaotic and/or transient lifestyles; and
• it suffers from right hand censorship because data gathering ceases at a point when the sample is still homeless. This means that much more is known about routes into than routes out of homelessness.

2.14 There have been some recent attempts in the UK to address these gaps through for example, follow-up studies which track homeless people's progress prospectively, or through surveying ex-homeless people. These studies are reviewed in Chapter 3, together with a selection of relevant US literature.

2.15 A further limitation in the UK homelessness literature is the lack of robust statistical studies (unlike in the US where quantitative studies based on large samples are much more common). This means that quantitative research questions, for example regarding the proportion of the homeless population with support needs, remain unanswered. The main exceptions to this dearth of quantitative data are the major Department of the Environment funded studies of single homeless people (Anderson et al., 1993) and homeless applicants to local authorities (O'Callaghan et al., 1996). However, only the latter had any prospective longitudinal element, and is therefore reviewed in Chapter 3.

CONCLUSIONS

2.16 This chapter has defined the parameters of this study by outlining the definition of homelessness employed, and has indicated our intention to address both visible and hidden dimensions of the phenomenon. It has also clarified the policy needs for longitudinal research on homelessness, highlighting its importance in enhancing our general understanding of movement in to and out of homelessness and in developing robust evaluations of policy interventions. Finally, it has pointed to the shortage of homelessness research in the UK which has a prospective longitudinal dimension or is capable of tracing routes out of homelessness.

« Previous | Contents | Next »