Policy use of the SARs

 

There have been a number of specific policy related uses of the SARs in such areas as labour force forecasts, improving the precision of small area estimates, the use of SARs with survey data for synthetic estimation, and in household projections.

• The use of SARs for greater precision in policy-making
  The SARs can deliver more than the conventional approach by the use of techniques to avoid the 'ecological fallacy' and related problems. For example, the SARs can be used to predict the probability of long-term illness for each age of the 0-9 and 80-89 year olds, controlling for gender, class and tenure etc. which cannot be done via the SAS data (Gardiner, 1996). The results of this analysis show that people living in public housing were consistently more likely to have long-term illness than people in private sector housing; and that age and tenure were highly significant, but not gender.
  
  
• The wide range of issues on which the use of SARs can provide enhanced value
  In Manchester City, the SARs were used to show that, for single households in the city, tenure patterns vary significantly with age, with younger age groups concentrated in private rented property and the older ones in Council renting or owner occupations. The SARs can provide information for policy making which cannot be obtained from LBS/SAS. The Children's Services Division in the City Council wished to have information on the family type and economic circumstances for the 0-8 year olds, but what SAS can provide is the 5-year age bands where children aged 0-8 cannot be grouped. Another example concerned a comparison between cyclists and others. The SAR data showed that the distances to work travelled by cyclists were the same as by non-cyclists; that the cyclists were better educated, and that although half of them had at least one car at home, twice as many cyclists as non-cyclists did not have a car. Such data proved very good background information to the Sustainability Group (Butler, 1995).
  
  
• The use of SARs for labour force forecasts
  To conduct labour force forecasts in multi-ethnic areas, detailed economic activity rates by age and sex for each ethnic group are required, and the information can only be obtained from the SARs (Bradford City Council, 1996). The imprecision of estimates, due to the problem of small cell size, can be improved by using the univariate or bivariate distributions of key variables from the 100% Small Area Statistics (SAS). Simpson (1998) describes how the marginal numbers in each age category obtained from the sample data can be scaled to give numbers consistent with the numbers obtained from the 100% count available for age using Iterative Proportional Fitting (IPF).
  
  
• The use of SARs with survey data for synthetic estimation
  The fact that the SARs cover a wide range of socio-demographic variables enables the use of synthetic estimation whereby data from the SARs and from social surveys can be used in combination to obtain estimates at local authority level. Charlton (1998) uses regression coefficients derived from a General Practitioner (GP) morbidity survey to project the uptake of GP services at a local authority level. The predictions were applied to each individual in the 2% SAR and the probability was obtained of their having a serious illness in each of the SAR areas.
  
  
• The use of SARs for household projections
  Given the importance of household projections with regard to future housing requirements in England, King and Bolsdon (1998) argues that the SARs offer actual and potential opportunities to add value to traditional projections. The availability of the SARs enables testing definitional sensitivity of projection outcomes, assisting further detailed disaggregation of projected components, assisting the matching of household projections to dwelling supply, and offering scope to explore via data linkage the relationships between household projections and 'backlog' housing needs, affordability, dwelling size, and tenure. The SARs can also assist in less conventional approaches to household projection including gross formation modelling, dynamic modelling, and microsimulation.