The Scottish Government

« Previous | Contents | Next »

Listen

Measuring Deprivation in Scotland: Developing a Long-Term Strategy: Final Report

3. Individual and area approaches

One aim of this research was to explore whether individual measures should be developed in addition to, or in place of, area-based measures. From the consultation process, there was a strong consensus on the need for both as these measures serve different purposes. Indeed, feedback from the consultation process stressed the added value that lies in comparing and relating the two sets of measures. This section therefore starts by reviewing the development of the two approaches and their uses. In practical terms, different methodologies are needed to construct these measures because of constraints of data availability. The differences in purpose and methodologies have resulted in some divergence between them in terms of the aspects of deprivation which they measure. Although both approaches start from Townsend's definition, the resulting measures represent slightly different conceptions of deprivation. This must be borne in mind if the Executive intends producing both measures in parallel. Finally the section explores how comparisons between the two approaches might be facilitated.

3.1 Individual measures

There is a long history of measuring deprivation at the individual level which dates back at least to Townsend's work (Townsend, 1979, 1987). This has been further developed in the Breadline Britain and PSE1999 studies (Mack and Lansley, 1985; Gordon and Pantazis, 1997; Gordon, Adelman et al, 2000). All of this work has been conducted by independent researchers. To date, neither Scottish nor British Governments have sought to define and measure levels of deprivation although this may be about to change as Section 7 discusses. The individual approach provides a directmeasure of the living standards enjoyed by each person and hence a direct estimate of the proportion of people in multiple deprivation in a given area. This enables comparisons to be made of the absolute levels of need between areas or of changes in levels of need over time. Through the collection of further information on individual characteristics, this approach allows research to explore the relationships between material or social deprivation and other dimensions of social exclusion such as social, cultural or political marginalisation.

The individual approach relies on the collection of a large amount of information about each individual's living standards, covering their material and social circumstances. The main method to date has been through the use of a household survey. These surveys are relatively lengthy and this tends to limit the number of interviews and hence geographic detail. The PSE1999 had a sample of 1530, enabling it to provide robust estimates of levels of deprivation for Britain and some information on the picture for regions such as Scotland. With sufficient resources, it would be possible to expand the survey approach to provide data for sub-areas of Scotland but it would be very costly to provide data for areas smaller than local authorities in this way.

In theory, it may be possible to obtain direct measures of individual living standards by making greater use of the range of data held by different Government departments, as discussed in the Interim report. There are major technical and legal challenges before such an approach becomes a reality, not least due to data protection legislation and individual rights to confidentiality. This is an area the Executive might usefully monitor and review again in the medium term. For now, the survey approach is likely to remain the only feasible means of identifying individual deprivation and it is the one this report focuses on.

3.2 Area-based measures

In contrast to the individual measure, there is a long history of the Scottish Executive (and its predecessor, the Scottish Office) commissioning area-based deprivation indices for Scotland (Duguid and Grant, 1984; Duguid, 1986; Duguid, 1995; Gibb et al, 1998; Noble et al, 2003). These have their origins in urban policy initiatives. From the late 1960s, there was growing concern across Britain about the persistence of concentrations of deprivation in "inner city" areas and a recognition that the problems were multi-dimensional - declining employment opportunities and high unemployment, poor housing and physical environment, low educational attainment, poor health and failing public services, for example (Atkinson and Moon, 1994; Hill, 2000). This led to a series of initiatives targeted on areas with the greatest concentrations of deprivation (the Urban Programme being one of the first of these, the area-based Social Inclusion Partnerships one of the most recent). As a result, there has been a need to identify areas with the worst concentrations of deprivation in a consistent fashion.

There has been a long and vigorous debate both about the extent to which "deprived people" are concentrated into "deprived areas" and about the appropriateness of using area-based initiatives to tackle problems of deprivation (Holterman, 1975; Glennerster et al, 1999). Nevertheless, concentrations of deprivation persist, as does a concern about the impacts of living in "deprived areas" on individual life chances (Scottish Executive, 2002a). The aim of "narrowing the gap" between deprived areas and the average remains an important one for the Scottish Executive and, increasingly, for organisations involved in Community Planning Partnerships. In consequence, there is still strong demand for measures to identify the locations with the most severe concentrations of deprivation and track these over time. It should be stressed that measuring deprivation at the small area level does not imply that this is the most appropriate scale at which to intervene to try and tackle deprivation - a point also made by Noble et al (2003).

Recent indices in Scotland have been based on wards or postcode sectors (average population around 4000 to 5000) but earlier indices focussed on Census output areas (average population around 120 people). To achieve this level of spatial detail without the cost of very large surveys, some compromises are made. These indices are constructed from a series of indicators each measuring a single characteristic of the population (the proportion unemployed, lacking qualifications or on low income, for example). As a result, this approach cannot identify the number of individuals in multiple deprivation in each area. The combined index is a purely relative measure which ranks areas in relation to each other at a given point in time. It cannot be used to estimate absolute changes over time.

3.3 Conceptual bases for individual and area approaches

The two approaches differ not only in terms of how they measure deprivation but also in terms of what they measure. This reflects both practical constraints on the types of data available but also differences in the conceptualisation of deprivation which underpin each measure. Table 3.1 illustrates this by showing the domains which two current measures cover - the individual approach of the PSE1999 and the area approach of the SIMD2003 (the latter including the three domains which would have been in the index had adequate data been available at the time). The SIMD2003 is already organised into domains. The indicators from the PSE1999 have been grouped under headings taken from Townsend's definition. It is striking how little overlap there is between the two although both cover aspects of financial resources in addition to aspects of deprivation.

3.3.1 Individual approach

The domains in the PSE1999 focus directly on material goods or social relations - food, clothing, housing, household goods and participation in social activities. Access to these "necessities" is directly affected by individual income although, arguably, housing is slightly different as this is an area where, for many on low incomes, consumption is determined to a large extent by state provision (social renting) or subsidies which are separate from income (Housing Benefit). There is no attempt to include measures of area or locational characteristics within this approach. Within the PSE1999, area deprivation is seen as quite distinct, and is measured through a separate set of indicators. Deprivation in this sense is purely individual and determined by financial resources.

Table 3.1: Domains for individual and area approaches

3.3.2 Area approach

In addition to income and housing, the area index is composed of measures of personal characteristics such as employment, health or education, and of area or locational characteristics (geographic access to services, physical environment and crime/social disorder). Other than housing, there are no measures of access to the material goods or social relations considered "necessities" as there is no data on these for small areas.

The justification for including income is obviously its role as the key cause of deprivation. Housing measures are included as this is the one aspect of living standards for which there are adequate measures at small area level. The justification for the other domains requires more explanation.

Education, health and employment

On the one hand, the education domain is seen as measuring "the key educational characteristics of the local area that might contribute to the overall level of deprivation and disadvantage" (Noble et al, 2003: p.20 - emphasis in original). In this perspective, the education indicators are indirect indicators of deprivation as they measure a major cause of low income. The argument might equally have been that educational characteristics are an indirect measure of deprivation as low educational attainment is associated with low household incomes or deprivation. On the other hand, the SIMD2003 also justifies measuring low educational achievement as a deprivation in its own right: "if - for whatever reason - one area has better educational results than another that may be less disadvantaged in other respects, then this area is less educationally deprived" (p.20 - emphasis in original). This sees education as an important in itself rather than as a measure of deprivation. With the health domain, the justification in the SIMD2003 report is again both in terms of its relationship with deprivation and in terms of health as important in its own right: "While ill health is closely intertwined with other aspects of deprivation, it is also an important aspect of deprivation in its own right" (p.16).

While low educational attainment or poor health may be problems in themselves, this does not imply that they should be seen as deprivations given Townsend's definition. We argue therefore that the justification for both these domains should rest much more clearly on their role as indirect indicators of underlying problems of deprivation. In both cases, the domains can be justified as indirect indicators of outcomes and of causes (or contributory factors) which are strongly associated with deprivation.

With the employment domain, the justification in the SIMD2003 is placed solely on unemployment as a problem in itself. It "seeks to measure enforced exclusion from the world of work. The domain does not seek to capture income deprivation to which joblessness leads, since this is tackled in the Income Deprivation Domain" (p.14). This fits well with some conceptions of social exclusion but, as with education and health, it does not fit within our definition of deprivation. (If two people have the same income but one is unemployed, the unemployed person may be seen as more excluded but we would not see them as necessarily more deprived).

The alternative justification for including the employment domain (rejected by the SIMD2003 report) is that unemployment is a major cause of low income and hence deprivation. If there were no other measures of low income, this domain would be a useful addition to the index. As it is, it appears to largely duplicate the function of the income domain but it does so less directly. There are also very significant problems with the definition of this domain as measuring those "who want to work but are unable to do so through unemployment, sickness or disability" (p.14), as discussed in the Interim Report. As a result, we recommend that the Executive considers dropping this domain from the index.

Area or locational characteristics

These domains were seen by many respondents during the consultation process as important components of an area deprivation index, although this does not mean there was unanimous support for every domain. The main justification for including them in the deprivation index is rather different than for the measures of individual characteristic.

It is possible to argue that these domains should be included in the index as indirect measures of individual deprivation. Measures of poor physical or social environment can be seen as indirect indicators of deprivation outcomes on the basis that people with low incomes are more likely to live in such areas than more affluent groups. Poor environment or poor geographic access to services may also be seen as indirect measures of causes of deprivation. Living in an area with a poor environment may contribute to problems such as poor health and hence deprivation (Noble et al, 2003: p.33). Poor geographic access can be seen as leading to higher costs in travelling to access services and hence in reduced disposable incomes and increased deprivation.

From the consultation process, however, it is clear that many people using area deprivation indices see area or locational characteristics as direct measures of problems with the area as a whole. These indicators are not seen as proxies for underlying problems of deprivation but, rather, as problems in their own right which should be the target of specific policies. As noted already, in the PSE1999, area characteristics are treated separately from measures of individual deprivation (the list of "necessities"). In the area indices, these are seen as part of the overall measure of deprivation which combines individual and area characteristics. We recommend that these domains should be included on this basis. They should not be seen as indicators of underlying problems of individual deprivation.

3.3.3 Final set of domains

The domains discussed above reflect those included in the current measures developed for the PSE1999 and for the SIMD2003. The Interim Report set out further items which might be included in the area index either as new domains or as additions to existing domains.

There was support for:

• covering a broader range of factors which determine the levels of financial resources at the disposal of each household in addition to current income;
• extending the geographic access domain to include access to work as well as services, and availability of communications infrastructure; and
• developing a new domain on social relationships or social capital.

A summary of the final list of domains which we recommend the individual and area measures should cover is provided in Table 3.2 below. Although we have argued against the employment domain (as have others such as the Countryside Agency (2003)), there was strong support for retaining it. We have therefore retained it in the list for now.

Table 3.2: Summary of proposed domains

3.4 Comparing results from the two measures

As noted above, there is a long-running debate over the extent to which deprived individuals are concentrated into the most deprived areas and over the suitability of area-based initiatives as responses to problems of concentrated deprivation. By devising individual and area-based measures, this work could provide very valuable insights into these relationships and inform the development of policy in future. As the discussion above illustrates, the measures start from rather different conceptual bases and, to some extent, measure different types of problem. Care needs to be taken when making comparisons between the two.

The process of comparison can be aided in at least three respects. First, the Executive should make available as many of the individual indicators underpinning the area-based index as possible, subject obviously to the need to preserve confidentiality. These would show the proportion of people deprived on a single dimension in each ward and, hence, the extent to which different groups were more or less concentrated into the most deprived areas. One constraint on making available data from the SIMD2003 at ward level has been the fact that some of the data had previously been released for a different set of small areas (postcode sectors). Release of both sets of data might have breached confidentiality rules, due to the possibility of differencing. In future, this suggests a need to co-ordinate release of small area statistics better so that the full set of deprivation indicators can be released.

Second, the individual survey could be used to explore the distribution of deprived individuals between more or less deprived areas. Assuming the individual survey records the precise postcode of each person interviewed, it is possible to link each individual's data to the area in which they live and hence to an area deprivation score. There would not be sufficient data to provide a direct estimate of the number of deprived individuals in every small area but it would be possible to estimate the characteristics of each group of areas (the most deprived 10 per cent of areas, and so on). This could be used to estimate the proportion of deprived individuals living in the most deprived 10 or 20 per cent of areas. Similarly, it could also be used to estimate the proportion of people living in the most deprived 10 or 20 per cent of areas who were deprived on the individual measure.

Third, this approach could be taken a stage further by using modelling techniques to provide estimates of the numbers of people multiply deprived in each individual small area. Knowledge of the characteristics of deprived individuals from survey-based measures could be combined with data on small area characteristics from the Census and/or Scottish Neighbourhood Statistics. A similar approach has been used by Lee et al (1995) to estimate distributions of deprived individuals for small areas and by Bramley and Lancaster (1998) to estimate income distributions for small areas. There would of course be questions of validity with such measures; these might be at least partially addressed by a one-off survey of deprivation in a small number of small areas to cross-check the two methods. If the validity of this approach were to be widely accepted, it offers the possibility of being able to track levels of deprivation for small areas over time - potentially a powerful tool for policy evaluation.

3.5 Conclusions

There is demand for both individual and area measures of deprivation as they are complementary rather than alternatives. Furthermore, relating the two measures to each other may provide particularly useful insights into the nature and distribution of deprivation in Scotland so there are synergies to be gained from developing both. Practical constraints mean that different methods will be needed for each. Individual measures will be based on surveys. Area-based indices will continue to be based on indicators from diverse sources as it would be prohibitively expensive to collect sufficient data through surveys. The practical steps outlined would help to ensure that the potential synergies from developing both measures are realised.

The different origins and uses of the two types of measure mean that they have tended to be based on slightly different understandings of deprivation. The individual approach tends to retain a narrower focus on the "necessities of life" but area measures include both individual and area characteristics. As users clearly expect area measures to have this wider conceptual base, we recommend that this difference is maintained even though it means that direct comparisons between the two approaches need to be made with care.

« Previous | Contents | Next »