Just Measuring the Missing Margins is not Rocket Science

Reflecting on an I-SPHERE report on measuring the personal wellbeing of marginalised groups, here Glen Bramley blogs on these and other findings for including the ‘un-included’ in official statistics.

As we approach the festive season there are many reflections and reminders of people whose situation places them at the margins of society, and particularly vulnerable to deprivation. This may be in terms of homelessness, insecurity, destitution, loneliness, abuse or mental distress. Many of us may be moved to donate to or support organisations or movements that support such people or promote measures to improve their lot. But effective action – whether through altruistic charitable channels or through public policy – can best be judged by evidence. Such evidence must significantly, if not exclusively, involve numbers. We therefore need to know how numerous and severe the problems people experience are; who is experiencing them; where they are; and what are the key factors driving this.

Professor Glen Bramley

I work in I-SPHERE, which is a research group which specialises in some of these overlapping marginal groups, including people experiencing homelessness; severe poverty and destitution; domestic violence; and combinations of complex, severe and multiple disadvantage (‘SMD’) (for example, involving addictions and/or offending with the above problems). It is a characteristic of all of these groups that, compared with a lot of mainstream social science which can draw on rich secondary data sources, they are not well represented in the data routinely used to inform government policy and most of the wider lobbying and campaigning around public policy.

For example, the frequent interventions of bodies like the Institute for Fiscal Studies or the Joseph Rowntree Foundation on poverty in the UK  almost invariably draw on the same data source (the Family Resources Survey) although they may occasionally look at other official household surveys (like the Expenditure and Food Survey). However, all of these surveys have in common that they are household surveys; by definition they do not measure the poverty or wellbeing of people who do not live in private households such as residents of all different kinds of institutions or communal establishments, as well as most of the homeless (not just including rough sleepers).  In fact, the problem is worse than that, because these household surveys don’t collect any information about people who may be staying temporarily in a household without that being their ‘usual residence’ (e.g. ‘sofa surfers’); nor do they often collect adequate information about people temporarily absent. When these omissions are combined with the generally very low response rate to surveys in areas of poverty (combined with transience, private renting and multiple occupation) it seems clear that official surveys may be missing a substantial part of the most troubled populations in the UK.

Administrative data is increasingly hailed as a new solution to some of the problems with surveys, which may be true in some respects. But there are many hurdles to the effective and routine use of administrative data in the UK, particularly owing to the increasingly onerous procedures of data governance surrounding such usage, which are not helped by stories of abuse of personal data by tech companies and malign political movements. Yet, administrative data is only as inclusive as the policy system in which it is embedded. For example, in England until this year, the homelessness system only applied to ‘priority need’ households (primarily families with children), so in effect the administrative records of homelessness omitted most of the, numerically much larger, group of single homeless people. Even in Scotland, where the formal system is most inclusive, approximately 30% of adults reporting homelessness experiences did not apply to a local authority for assistance.

In recent research with Crisis we developed a concept of ‘Core Homelessness’ as an attempt to overcome this limitation and paint a fuller picture of the number of people who are currently experiencing the most acute forms of homelessness or living in short-term or unsuitable accommodation.  This is part of a wider picture involving others who have gained statutory recognition of their homeless status but not yet achieved a settled housing solution, and a much wider group of households in various kinds of need situations who are at tangible risk of homelessness in the near future (Bramley 2017, 2018 forthcoming). While we feel that this is conceptually sound and capable of yielding broadly defensible estimates of rounded numbers of homeless people in the different categories – for example, people rough sleeping, staying in hostels, unsuitable temporary accommodation, squatting or ‘sofa surfing’ – these numbers do not yet have the solidity of official statistics or a granularity of counting at the level of individual local authority areas.

The limitations of administrative data in relation to other dimensions of disadvantage are equally apparent. Offending data from the criminal justice system does not capture offenders who have not been caught or crimes which have not been reported or recorded. Mental health services are known to be inadequate, overstretched and hard to access, and the same is probably true of addiction services.

When we responded to the challenge from the Joseph Rowntree Foundation (JRF) to try to define and measure ‘Destitution in the UK’ ), we quickly came to the view that existing surveys would not adequately capture it. Instead, a more specialist approach was needed to capture the range of marginal groups affected such as undocumented migrants or those with no recourse to public funds; homeless people; complex needs ‘SMD’ groups; as well as targeting mainstream UK households facing extreme and immediate material deprivation. We adapted a survey technique from a previous study based on a structured sample of the users of services providing emergency advice and aid to the range of such groups in case study areas. This enabled national estimates of the scale and profile of the problem of destitution. However, these estimates remain subject to a significant ‘confidence margin’, and it would require a substantial investment in up-scaling to narrow it. Part of the attraction of making such an investment now is that it might also yield better evidence on other groups of strong current policy interest, including rough sleepers and other core homeless groups, and people experiencing complex ‘SMD’ issues.

Fortunately, the interests of JRF and the Office for National Statistics (ONS) have come together in this instance, as the ONS recognise these policy priorities and the shortcomings of existing statistical resources, and are open to examining ways of including the statistically excluded marginal groups within official statistics of both living standards (including poverty and destitution) and personal wellbeing, as well as improving the basic counting of numbers and demographic profiles. We were commissioned by JRF with ONS to undertake a scoping study of this topic, and this week ONS published a summary of our report, the full version of which is published on our website .

A number of recommendations in this report are potentially highly significant for the future measurement and framing of homelessness, severe poverty/destitution and complex needs in the UK. There is an opportunity here, with high level support from the national statistical agency, to bring about a step change in the evidence base on issues which have risen close to the top of the political agenda (leaving aside, please, Brexit). The recent visit by the UN Special Rapporteur on severe poverty Philip Alston and his forthright interim report, as well as other interventions such as from the Archbishop of Canterbury, have served to do this for ‘Destitution’, for example.

The four most important recommendations, in my view, are these

  • ONS should actively explore the feasibility of including the main non-private household population groups within the measurement of living standards and well-being (Rec 9)
  • Government and statistical agencies should consider a measurement framework which recognises the concept of ‘core homelessness’ alongside the existing statutory framework, while also recognising wider groups at high risk of future homelessness (Rec 1)
  • ONS should commission a study of ‘sofa surfers’ and other temporary household members, both through retrospective survey questions and through including them directly in one or more of the major surveys (Rec 8)
  • A medium scale comparison between service-user-based and accommodation-based sampling approaches should be tested in the homelessness and related sectors (Rec 10)

The first of these recommendations actually covers proposed tailored approaches to surveying the populations residing in each of nine different types of ‘communal establishment’, as reviewed in detail in the report. These range from on-line surveys of students in halls; through to piggybacking on existing formal surveys of armed forces personnel and accommodation; to various forms of self-completion or interview surveys in other communal settings. The report recommends dovetailing with the existing UK poverty measurement framework, including the use of consensual material deprivation indicators as well as relative income measures, but with clarification of certain elements. It favours using the ONS’s standard four-question suite on personal well-being but suggests additional areas relating to human and social capital could be included.

This is the best opportunity in a generation to achieve a step change in the statistical inclusion of these groups on the margins of British society. In some areas (e.g. homelessness), Government has been heavily criticised for having neither statistics nor policies which are fit for purpose, and ONS is formally reviewing the whole area of homelessness statistics including definitions. The national statistical agency is ‘on the case’ but it faces dilemmas regarding cost-effectiveness, with strong pressures to substitute administrative data for interview surveys. However, in our view you cannot measure severe poverty and destitution without asking people themselves about material deprivations, and you certainly cannot measure wellbeing without asking them. It will cost more to mount some of these bespoke surveys, or to include non-usual residents in the main surveys. Yet only in this way will we get a true measure of the scale, trends in, and profile of severe poverty/destitution and personal well-being in the UK.

Key References

Bramley, G. (2018 forthcoming) Core and Wider Homelessness across Great Britain: Extent, trends and prospects. Report of Research for Crisis. Edinburgh: Heriot-Watt University.

Bramley, G. (2018) Housing supply requirements across Great Britain for low-income households and homeless people. Report of Research for the National Housing Federation and Crisis.