The Australian Standard Geographical Classification (ASGC) has provided the ABS statistical geography framework since its inception in 1984. A review of the ASGC in 1996 revealed some shortcomings with the application and ongoing maintenance of the ASGC, and these were reinforced in a further review in 2007. The main issue identified was with the provision of data over time as the areas within the ASGC were not stable, with annual updates largely driven by changes in Local Government Area (LGA) boundaries to which the ASGC had to conform. This requirement to conform to LGAs compromised the ABS' ability to maintain accurate time series datasets, and required them to be recast on to the latest boundaries, which was an ongoing process. Other issues identified included inconsistent area sizes and populations, and variable statistical quality in very small areas. In addition, in 2004 the ABS had commenced the design of a new very small area-level geography - the Mesh Block.
The outcome of the 2007 review resulted in the creation of a completely new statistical framework, the Australia Statistical Geography Standard (ASGS). The design of the ASGS addressed many of the issues previously identified with the ASGC. Primary among these was stability of design and a concerted effort to standardise the design of statistical regions at all levels, optimising each level of the new geography for consistent and comparable quality statistical outputs. Additionally, the ASGS incorporated the Mesh Block as its basic building block. The ASGS is designed to be stable over 5 years, with updates to the ASGS promulgated in the lead up to each Census of Population and Housing (the Census).
Units within the ASGS are internally consistent in population size, and based on functional and geographic criteria. Each level of the main structure of the ASGS is optimised for the release of key data collections - the Statistical Area Level 1 (SA1) is optimised for the release of Census data, the Statistical Area Level 2 (SA2) for the release of demographic Estimated Resident Population statistics, and the Statistical Area Level 4 (SA4) optimised for the release of Labour Force Survey data.
The ABS released the first data on the ASGS in August 2011. The 2011 Census, conducted on 9 August 2011, was the first Census to be collected on the ASGS, and Census data will become available in June 2012. As the ASGS will be the basis for the release of all geographically based statistics from the ABS, there is an opportunity for other organisations to modify their own statistical compilation methods to align with the ASGS. The advantage of doing so includes access to a wider variety of geographically comparable statistics that will inevitably lead to improved information for decision making and policy formulation.
The change to a new geographic standard has created a need for users to move historical and existing time series datasets onto the new geography to enable the ongoing comparison of data. There are a number of strategies that can be applied to aid this process, and the aim of this information paper is to look at the options that are available to users when converting data from other geographical classifications, including the ASGC, to the ASGS. This paper will examine options such as address coding and correspondences and will discuss the concepts behind these methods of converting data, as well as discussing some of the issues that the use of these methods can present to users. These options are not only available for ABS data collections, they can also be applied to data held by other agencies and organisations.
The options available to users to convert unit record data will depend largely on the information contained within those records. If these records contain full address information, or latitude and longitude, then it will be possible to geocode them to the finer levels of the ASGS such as Mesh Block or SA1. However, if only partial address information is available then coding indexes can be used to convert data, but this will only be possible at the higher levels of the ASGS such as SA2 and above.
If the collection data contains records that already include Mesh Block or SA1 codes, then allocation tables can be used to aggregate data to higher levels of the ASGS. It needs to be noted though, that an allocation table cannot be used to convert data from higher levels of the ASGS to the finer levels.
If no address information is retained with the unit records then geocoding and indexes will not be options for the conversion of data and correspondences will need to be utilised. There are many issues associated with using correspondences, most notably that attempting to convert data from large scale geographies to the finer levels of the ASGS will not work well and the data returned will not be reliable. The type of data being converted also needs to be considered as the weighting unit used can impact on how well data will be converted. All of these issues are discussed in detail.