2.1. Overall methodology
In 2005, the Robertson Centre for Biostatistics at Glasgow University formally evaluated the methodology used in the SIMD 2004. The methodology passed its 'health check' and was approved as fit for identifying areas of concentrated multiple deprivation. A full report of the evaluation, which includes an Executive Summary and Glasgow University's recommendations, along with the Scottish Executive's response were published online on 3 November 2005 (see Annex c).
The Scottish Executive accepted Glasgow University's recommendation to remove shrinkage from the methodology and, where indicators are age-sex standardised, to change from the Direct to Indirect Standardisation method. Further information on these methodological changes and their impact on the SIMD are given below.
Aside from removing shrinkage and changing the standardisation method, the overall methodology for the 2006 SIMD remains the same as the 2004 index. Within each domain, the data source or methodology for creating individual indicators has changed for a number of indicators. These changes have been kept to a minimum, however changes are inevitable due to improvements in data quality or indicator availability, these are described in full in the relevant chapters of this technical report.
A full technical report for the SIMD 2004 was published in October 2003 (see Annex c). The individual methods for creating the domains and overall index are described fully in the 2004 technical report. Changes and additions to this methodology for the SIMD 2006 are described below.
The statistical techniques used to compile the SIMD are carried out using SAS statistical software. Codes to run the programmes were adapted from those created by the Scottish Executive in the calculation of the SIMD 2004. After the SIMD 2006 had been calculated, the SAS codes were independently reviewed by Alex McConnachie at the Robertson Centre for Biostatistics at Glasgow University. The SAS codes were found to be fit for purpose although some recommendations for improving the codes were made. The Scottish executive will consider these recommendations in time for the next SIMD update. The evaluation of the SAS codes has been published on the Scottish Executive website (see Annex c).
Shrinkage has not been applied to any indicators included in the SIMD 2006. In the SIMD 2004, shrinkage was applied to indicators in the health and education domains. This procedure was used with the intention to improve the quality of the small area data in indicators where populations were small. Shrinkage involves 'borrowing strength' from a more robust value, in this case the Local Authority mean. Briefly, it can be thought of as a weighted average of the data zone score and the mean of the data zone scores in the same local authority. Thus the shrinkage procedure moves data zone scores towards their local authority mean score. The degree of movement depends on the size of the weight.
Glasgow University, in their evaluation of the SIMD 2004 methodology, recommended that shrinkage should be removed in the SIMD 2006. From their analysis, shrinkage was shown to have little effect on the resultant indices and, by shrinking towards local authority means, introduces a very small bias against data zones in otherwise less deprived areas. Glasgow University also felt that the application of shrinkage within some domains but not others did not constitute a consistent approach, and that the use of Factor Analysis results implicitly in a degree of shrinkage. Removing shrinkage has considerably simplified the methodology used to construct the SIMD.
2.3. Age-sex standardisation
The aim of standardisation is to provide a summary 'adjusted' rate to take into account underlying differences (age, sex, deprivation etc) of a study population relative to a 'reference' population (in the case of the SIMD this is Scotland).
A minor recommendation made by Glasgow University was to change the age-sex standardisation method used in the Comparative Illness Factor, Comparative Mortality Ratio1 and the Working age adults with no qualifications indicators from the direct to the indirect method.
A Directly Age Standardised Rate is a theoretical rate, which would have occurred if the age-sex specific rates in the actual study (data zone) population were applied to a 'reference population'. A disadvantage of direct standardisation is that it can be influenced by unstable/unreliable rates due to small numbers. This is not the case with indirect standardisation.
The Direct method used in the SIMD 2004 was calculated using the formula:
Indirect Age standardisation is a comparison of observed to expected numbers of cases by applying age-sex specific rates from a 'Standard Population' i.e. Scotland to the population of interest (data zone).
The Indirect Standardisation which has been used in SIMD 2006 uses the formula:
Indirectly standardised ratios are of the observed numbers of events to the expected numbers in each data zone, given the national age-sex distribution of events. This is a more widely used summary of event rates, and was found to have limited impact on the resultant SIMD 2004 ranks compared with the direct standardisation methodology used in 2004. For example, the working age adults with no qualifications indicators produced by the two different methods are highly correlated (Pearson Correlation coefficient of 0.99).