New ABS Geography part 3. Replacing SLAs with SA2s
Probably the most radical change in the new ABS geography is the move to SA2s (“Statistical Area Level 2” – another imaginative name…). These replace Statistical Local Areas (SLAs), which were always a bit misunderstood. This is the third part in my series on the new ABS geography.
Jump ahead: summary of ‘what is an SA2?’
Before you start, you may like to catch up with the previous posts on this topic.
You may also like to read the other posts on the new ABS geography:
Basically, the SLA has been for many years the main data output geography for the ABS. The only catch was that not many people knew what they were. SLAs have always been based on Local Government Areas, but SA2s will change all that.
A lot of SLAs were just single local government areas. E.g: Ryde (C) NSW, West Wimmera (S) Vic, Gawler (T) SA – where the letter in brackets represents the type of LGA – C for City, S for Shire, A for area, T for town, DC for District Council etc. That made it nice and easy as they were the main form of data output.
Larger LGAs were often split into multiple SLAs – e.g. Yarra Ranges (S) – Lilydale OR Dubbo (C) – Part A OR Stirling (C) – Coastal OR Bayside (C) – Brighton. These correspond to broad regions within the LGA but not necessarily to units relevant to local government.
SLAs are the basic output unit for a lot of Census data, but perhaps more importantly, for other ABS datasets. For instance, you can get Estimated Resident Population, Tourist Accommodation, Building Approvals, Births and Deaths data at the SLA level from the ABS website, as well as a lot of unpublished data from other surveys on request.
One of the reasons for changing is the lack of consistency in SLAs across the country. The City of Brisbane, for instance, was split into 158 suburb-sized SLAs, for about a million people, while the City of Blacktown, NSW, had only 3 SLAs for almost 300,000 people. This inconsistency meant that the resolution of the additional data series listed above has been far better in Queensland and the ACT than NSW, Victoria and WA, for instance.
So the ABS is replacing SLAs with SA2s. These basically replicate the Queensland model for the rest of Australia, giving suburb/town sized areas across most regions. So one of the great things about them is that you will be able to get standard non-Census output, such as the Estimated Resident Population, at more detail in the other states. SA2s in urban areas will correspond roughly to suburbs, or small groupings of suburbs.
The groupings are generally named for one dominant suburb, which could cause a little confusion in knowing which suburb is combined with a neighbour. Though probably no more confusion than existed with SLAs in the past.
See the example below in metropolitan Perth. Each of the named areas is an SA2.
In regional areas, small towns (< 5,000 people) are generally included as an SA2 with their surrounding region (eg. Stawell Vic, Temora NSW, Katanning WA) while larger towns have an SA2 covering just the town area, and another for the outskirts and surrounding region. Eg. Cowra (NSW), Cowra Region).
Towns over about 20,000 population are split into multiple SA2s in many cases, which break up into named suburbs, or along directional lines eg. Wagga Wagga – North, South, East and West, which seems a little arbitrary in some cases.
Metropolitan areas get far more detail at the SA2 level than at the old SLA level. The City of Blacktown, mentioned earlier, has 3 SLAs, but they are replaced by 17 SA2s (which don’t quite align to the LGA boundary, see below).
Despite far more detail at the SA2 level in metropolitan areas than currently exists with SLAs, the total number of SA2s nationally is only about 50% higher than SLAs (the 2010 version of the classification had 1,389 SLAs, while the 2011 ASGS has 2,214 SA2s). The reason for this is that there is less detail in some of the more sparsely settled rural areas. This is particularly the case in Western Australia, which has a large number of very small LGAs, each of which has been its own SLA. Many of these are now combined into a single SA2 – eg. the SA2 of Wagin, WA now contains the former SLAs of Cuballing (S), Wagin (S), West Arthur (S), Williams (S) and part of Narrogin (S). This is due to a need for consistency in the size of SA2s, each needing a minimum population of around 1,000 people.
This brings us to the other interesting point about SA2s – even in metropolitan areas they don’t necessarily align to Local Government Boundaries. ABS have prioritised concepts such as communities, catchment boundaries, official localities and geographical barriers over aligning to LGAs. In fact part of the reason for bringing in the whole new classification was to remove the need to align to LGA boundaries in the first place, which can be changed by State Governments without consulting the ABS.
See the example below for suburban Adelaide – the pink boundaries are LGA boundaries, while the darker lines are SA2s and lighter ones SA1s. In this area you won’t be able to add up SA2s to get a total for Tea Tree Gully, Salisbury, or Port Adelaide Enfield councils.
Now ABS have said that they will continue to produce population estimates after the 2011 Census for Local Government Areas, and for the Census itself, .id will continue to be able to produce community profiles for Local Government by using the same splitting and alignment methods as we’ve used before. But the outcome is less clear for the extra datasets such as Births, Building Approvals and Tourism data.
The other issue is that, even from the Census, we will lose the neat progression of being able to aggregate up whole Collection Districts to SLAs and LGAs – SA2s don’t neatly add up to LGAs and never will. So while there will be more data resolution in many areas (and less in some), if you work in Local Government, there is quite a bit more work in making sense of the data and getting lower geographic levels to add up to your LGA.
- SA2s are replacing SLAs and will overall provide a greater resolution across the country, particularly for non-Census datasets currently produced at SLA level.
- For metropolitan areas, there is far more resolution, with suburbs or small groups of suburbs being the basis for SA2s.
- In rural areas, heavily populated districts are likely to have better resolution, with a split between a large town and its region the most common.
- However, more sparsely settled areas have less resolution, with multiple LGAs combined into a single SA2 for data output. This particularly occurs in Western Australia and parts of rural NSW and QLD.
- SA2s do not aggregate to LGA boundaries. ABS have said that they will continue to support LGA level data in the future, but there is no longer a neat progression as lower areas do not sum to LGAs.
- If you are a .id client, we will make sure that your geographic areas in profile.id and forecast.id remain comparable over time, despite the geography changes.
If you enjoyed this article, sign up to get updates via email or twitter (above), and feel free to share it.