Background to the Research

• The government initiative 'New Targeting Social Need' aims to tackle social need and social exclusion. In order to target resources towards those in greatest need, objective measures need to be found to identify people, groups and areas. This paper evaluates the methods used to identify those in social need in Great Britain and the Republic of Ireland.
  

Research Approach

• The authors used secondary data analysis and a range of statistical tests in order to test the validity and reliability of a range of methods of identifying areas and groups in social need.
  

Main Findings

Geographic units of analysis

• This method uses geographic units as the units of analysis - the geographic unit should reflect the geography of deprivation, each unit should be as socio-economically similar as possible. In reality this is usually not the case, as one unit may show a fairly even distribution of deprivation, while another may have a pocket of deprivation within an area of relative affluence.
• In practice, administrative units are often used and it is unlikely that geographical deprivation patterns follow administrative boundaries. Therefore, it is possible that the mismatch of administrative boundaries and unobserved boundaries could lead to an area of relative deprivation not being identified.
• The size of the unit of analysis is also important and the geographical unit varies according to the objective of the analysis. In relation to measuring deprivation in NI, the electoral wards have tended to be the geographical unit most frequently used. A typical ward has a population of about 3,000 individuals living in 1,000 households.

Deprivation Analysis in Northern Ireland

• The Census of Population is the best source of data on all geographical units and geographical deprivation analyses in NI have largely been based on these data.
• The first comprehensive analysis of the geography of deprivation 'Belfast Areas of Special Social Need'- 1976 , centred on an analysis of 101 electoral wards in the BUA. A total of 20 social need indicators (unemployment, free school meals etc.) and 129 social characteristic variables (children as percentage of population, public authority housing etc.) were measured for each ward. Some of the indicators were measured using 1971 Census data, but the majority were derived from administrative sources. Analysis of the data yielded ward maps in which a core set of deprived wards was identified (the inner-city plus parts of north and west Belfast).
• The BUA analysis was repeated using census data only and the analysis extended to all wards in NI. This yielded a map of the relatively deprived wards throughout Northern Ireland.
• In the 1980s, an update of the 1970s analysis was carried out in the BUA, based mainly on the 1981 census data. The analysis was based on 10 indicators, the set of relatively deprived wards within the BUA remained largely unchanged. At the NI level, relative deprivation was examined through 4 census indicators.
• In the 1990s, using data from the 1991 Census, Professor Robson and his team were commissioned by government to take a fresh look at the geography of deprivation in NI and the 'Robson Report' was published in 1994. Robson used a new set of 18 indicators that included indicators such as lack of household facilities and unemployment.
• Unlike previous work in this area in NI, Robsons' analysis indicated that deprivation will not be distributed evenly across geographical areas and is unlikely to fit into administrative units. The Robson measures are based on 3 hierarchical geographical levels; the 26 LGDs, 566 electoral wards and 3,729 Census of Population EDs.
• The Robson method identified 3 measures of multiple deprivation; degree - the average level of deprivation in the area, intensity - assesses the level of the worst deprivation within an area, and extent - measures the percentage of an area's population living in sub-areas which are defined as 'deprived'.
• The Robson index of relative deprivation has been extensively used in the targeting of programmes that target multiple social and economic need.
• The Robson index scores and ranks the geographical units in NI from the most deprived to the most affluent. This has proved to be problematic in the targeting of various programmes as a 'cut-off' point in the deprivation had to be determined resulting in a stark contrast between those areas above the cut-off point (deemed as not deprived) and those below (deemed as deprived).
• The Robson measures, in keeping with most composite scores, are not ratio variable. It has also been argued that the Robson method has a bias against small areas and is slightly influenced by population size; although this effect is small compared to the effect of deprivation.

Recent Experience in England

• In 1995, Professor Robson produced the 'Index of Local Conditions for England', this is very similar to the work reported in NI - except the set of indicators was slightly different. A follow up report that included an updated index was published in 1998 and concluded there was little scope for updating the ILC.
• A comprehensive review of measures of deprivation was carried out by a team from Oxford University led by Mike Noble and published as 'Indices of Deprivation 2000'. A number of relevant indicators are identified and brought together in composite deprivation measures. The use of administrative data sources, the presence of postcodes and advances in IT capability to manipulate large databases, resulted in information that was more up-to-date and accurate, for example the pinpointing of the number of people receiving social security benefits by electoral ward. The idea of domains of deprivation was developed and given more prominence.

Recent Experience in Scotland

• In 1995, the Scottish Office Research Unit published Deprived Areas in Scotland, using Census of Population EDs as the geographical unit of analysis. It identified 12 indicators (reduced to a single composite) across three domains of socio-demographic, economic and housing, all measured through the 1991 Census of population. The worst 10% of Eds were classified as deprived areas.
• 'Revising the Scottish Area Deprivation Index', appeared in 1998 and estimated deprivation scores at postcode sector level. It used the Robson method, with three composite measures (degree, extent and intensity) based on 6 indicators.

Recent Experience in the Republic of Ireland

• The Economic and Social Research Institute produces most of the work on the geographical distribution of deprivation, culminating in the publication of 'Where are poor households' in 1998. This study explored the spacial distribution of a number of indicators such as housing and tenure, and unemployment, and looked for consistent patterns. The study did not create composite measures of deprivation and the author criticises Census of Population data as largely socio-demographic and lacking any measure of income or satisfactory measure of deprivation.

Townsend Index

• This is a composite index based on 4 indicators, unemployment, car ownership, home ownership and household crowding, which were transformed to z-scores and then combined with equal weighs. It has been used widely in NI.

The Northern Ireland Acute Need Index

• This was developed in 1997 using regression modelling to predict hospital utilisation at small-area level and is not a direct measure of deprivation.

The DoE Index of the 1980s

• Developed by the DoE in Great Britain, this index was based on 7 census variables, it had two variants, a basic index and a social index, with the latter giving double-weighting to certain indicators.

Updating the Northern Ireland Measures

• An up-to-date composite measure of deprivation wards in NI was developed by the Statistics and Research Branch in 'Deprivation in Northern Ireland 2000,' in order to inform the DSD's urban regeneration programme. It brought together information on the number of recipients of a range of social security benefits, alongside indicators for unemployment, education and health. These indicators were combined using the principal components analysis method.
• Research to review deprivation measures is currently underway and is being led by the Noble Team. The findings from this research are likely to supersede the DSD work.

Conclusions

• Robust and timely measures are key to the implementation of New TSN objectives.
• Such measures have been developed over recent years, most are based on Census of population and have informed programmes and initiatives.
• Advances in IT now make it possible for more up to date measures to be developed from government maintained adminstrative data sources.
• Current work by the Noble team should enable government to target resources more efficiently in future.