3.17. Although annual births and deaths data are available for SLAs, the absence of migration data for non-census years means that it is not possible to use the component method to update SLA population totals. Instead, population for most SLAs are estimated by applying a regression model.
3.18. Regression techniques first establish a relationship based on past data between population growth and the growth in 'symptomatic indicator(s)'. Symptomatic indicators are any available set of data which in some way relate to changes in population size. The choice of symptomatic indicators varies across the States. Some examples are the numbers of: dwellings, Medicare enrolments, drivers licenses and electricity connections. The relationships between population growth and symptomatic indicators are expressed mathematically in terms of 'regression coefficients' and, with the knowledge of the growth in the indicators for the current time period, they enable population growth to be estimated. The regression technique used is the 'difference correlation' method which is explained in Appendix 8.
3.19. Characteristics such as population growth rates may vary quite considerably between SLAs. In acknowledgment of these differences, stratification is applied to the modelling procedure. This involves separating the SLAs within a State into subsets based on factors such as location (urban or rural), population growth (high or otherwise), and house per dwelling ratio (number of houses in relation to number of total dwellings). More accurate estimates may then be calculated based on relationships existing within these subsets of more homogenous SLAs.
3.20. A range of alternative procedures based on established relationships between changes in symptomatic indicators and population change can be also used to estimate and/or validate SLA population estimates obtained from the regression technique. In addition, adjustments are made to take account of local knowledge.