CONSTRUCTING THE INDEXES
This section outlines the key details of the process used to derive each of the four SEIFA indexes: IRSD, IRSAD, IER and IEO. Each index is a weighted combination of different Census variables. A more detailed description of the index construction is given in Chapter 4 of the Technical Paper (available from the Downloads tab).
CHOOSING THE VARIABLES
Variables are included in the initial variable list if they relate to the concept of advantage and disadvantage that SEIFA is trying to capture. The main constraint on the initial variable list is that the variables can only be sourced from Census data. To learn more about the variables used, please see Chapter 3 of the SEIFA 2016 Technical Paper (available from the Downloads tab).
DATA
Proportions: The indexes are calculated using 2016 Census data at the base level of Statistical Area Level 1 (SA1). All of the variables included in the indexes are calculated as proportions at the SA1 level. For example, the proportion of unemployed persons is calculated as the number of unemployed people divided by the total number of people in the labour force in each SA1.
Areas without SEIFA Scores: Due to non-response and low population levels in some SA1s, not every area can be given a meaningful index score.
If an SA1 meets one or more of the following criteria, the area does not receive an index score:
- has no usual addresses
- the usual resident population is less than or equal to 10
- the area is classified as off-shore or migratory
- there are fewer than 6 employed persons
- there are fewer than 6 classifiable occupied private dwellings
- the proportion of people in private dwellings is less than or equal to 20%
- the denominator of a variable in the index is less than 6
These criteria aim to ensure that the largest number of areas receive a score, while maintaining an acceptable level of quality and confidentiality.
THE METHOD
The indexes are a weighted combination of Census variables. The method used to determine the weights and final variable composition is called Principal Component Analysis (PCA). The weights describe the relative importance of each variable. If the weight created by PCA for a particular variable is too low, the variable is dropped from the index. These variables do not meaningfully contribute to the indexes.
Note that the variable weights are calculated by a data-driven method. They are not determined subjectively by the ABS.
If you wish to learn more about PCA, please refer to Chapter 4 of the SEIFA 2016 Technical Paper (available from the Downloads tab).
Once the variable weights have been finalised, they are used together with the variable proportions to create index scores for SA1s across Australia. The interpretation of the index scores and associated output is discussed in the "SEIFA Measures" section of this guide.