Data quality
Identification of Aboriginal and Torres Strait Islander people
The accuracy of statistics about Aboriginal and Torres Strait Islander (First Nations) people depends on the accuracy of the information on Indigenous status in the underlying data.
Self-reporting is the way that Indigenous status information is collected from individuals (both Aboriginal and Torres Strait Islander people and non-Indigenous) in most data collections. A standard question is used to collect this item in the ABS Census of Population and Housing, ABS household surveys and in national administrative datasets. There is also a national standard for the recording of responses to this question. Answering the standard question is voluntary. To collect accurate data, the standard question must be asked of the respondent as it is worded, and the response must be recorded correctly.
Indigenous identification is often incomplete or is inconsistently reported across data sets. This can occur, for example, because there is no Indigenous status question for the collection; where the question is not asked (and left blank); where a person chooses not to identify in specific circumstances, or where the question is not asked but the answer is presumed.
Incomplete and inconsistent reporting of Indigenous status usually results in under-identification of First Nations people in data sets. The degree of under-identification can vary across different jurisdictions, by remoteness, sectors and service providers. Under-identification can create difficulties in assessing accurately the extent of the issue being investigated, and issues for accurately monitoring progress over time.
There are several methods used to assess the quality of Indigenous identification in key data sets. For example, the AIHW has assessed the quality of Indigenous identification in hospitals data (AIHW 2013). An estimated 88% of First Nations patients were correctly identified in Australian public hospital admission records in 2011–12. As a result of this work, total hospitalisations are adjusted by 12% to provide a more accurate estimate of the ‘true’ number of hospitalisations for First Nations people.
Similarly, data linkage showed that mortality for First Nations people was under-estimated by 18% in 2006–2010, and the data were adjusted prior to producing mortality and life expectancy estimates (AIHW 2017).
Coverage of the population
Survey estimates are based on information obtained from a sample of the population. There are many different methods that can be used to obtain a sample.
In First Nations-specific surveys, such as the ABS National Aboriginal and Torres Strait Islander Health Survey (NATSIHS), the sampling is specifically designed to produce representative estimates for the First Nations population, nationally, by state and territories and by remoteness.
However, for total population surveys, random sampling at the total population level will not result in a large enough population of First Nations people to allow for adequate disaggregation.
Another issue is the lack of full coverage of the First Nations population in some data collections. This can result in undercounting and/or a possible ‘biased’ representation of the First Nations population. For example, the 2018 Survey of Disability, Ageing and Carers excluded people living in Very remote areas, as well as households in discrete First Nations communities, so this survey is not representative of all First Nations people.
References
ABS (Australian Bureau of Statistics) Indigenous Status Standard, ABS, accessed 22 October 2024.
AIHW (Australian Institute of Health and Welfare) First Nations people: Aboriginal and Torres Strait Islander identification, AIHW, accessed 22 October 2024.
AIHW (2013) Indigenous identification in hospital separations data– Quality report, AIHW, accessed 16 October 2024.
AIHW (2017) Trends in Indigenous mortality and life expectancy, 2001–2015: evidence from the Enhanced Mortality Database, AIHW, accessed 16 October 2024.