IMPACTS OF SAMPLE SIZE ON ESTIMATE VARIABILITY
The smaller a population is in proportion to the total population, the higher the associated errors. Further disaggregation of data for small populations produces even higher errors proportionately. For example:
- In the 2015 SDAC, 23.9% of the total Aboriginal and Torres Strait Islander population had disability (excludes very remote areas and discrete Aboriginal and Torres Strait Islander Communities);
- The same survey shows 7.3% were classified as having a profound or severe limitation. The 95% confidence interval for this measure is ±1.4%. This means that we can be 95% confident that the actual proportion of people with a profound or severe limitation lies between 5.9% and 8.7%;
- Variability of this measure becomes much greater when the sample is limited further. In this example, the sample will be limited to those aged 35-44 years of age;
- In the 2015 SDAC, 5.6% of Aboriginal and Torres Strait Islander people aged 35-44 years had a profound or severe limitation; and
- The 95% confidence interval on this smaller sample is ±5.1%. This means that we can be 95% confident that the actual proportion of 35-44 year olds with disability, who have profound/severe limitation, lies between 0.5% and 10.7%.
Understanding small samples and their associated errors is particularly important when looking at smaller population groups. For further information on the potential effects of sample survey methodology see ‘Understanding Statistics’ on the ABS website.