SAMPLE DESIGN AND SELECTION

Sample design

The 2012-13 National Aboriginal and Torres Strait Islander Health Survey (NATSIHS) and National Aboriginal and Torres Strait Islander Nutrition and Physical Activity Survey (NATSINPAS) sample was designed to provide reliable Aboriginal and Torres Strait Islander estimates for the whole of Australia, and for remote and non-remote areas. 'Remote' is referred to as the aggregation of the Australian Statistical Geography Standard (ASGS) Remoteness Areas of 'Remote Australia' and 'Very Remote Australia', defined from Statistical Area Level 1s (SA1s). In addition, the NATSIHS (and the combined sample of NATSIHS and NATSINPAS, referred to as the Core) was designed to provide reliable estimates for state and territory and for the Torres Strait Area. The sample size, distribution and method of selection was based on the aims of the survey, the topics it contained, the level of disaggregation and accuracy at which the survey estimates were required, and the costs and operational constraints of conducting the survey.

In order to achieve the design objectives the in-scope Aboriginal and Torres Strait Islander population was divided (as far as possible) into persons residing in discrete Aboriginal and Torres Strait Islander communities in remote areas, referred to as the community frame, and the remainder of the in-scope population, referred to as the non-community frame. For both NATSIHS and NATSINPAS, samples were drawn from both the community frame and the non-community frame using different sample designs. This approach was taken because of operational and sampling issues.

The Indigenous Community Frame (ICF) consisted of a list of discrete Aboriginal and Torres Strait Islander communities (including any outstations associated with them) in remote areas of New South Wales, Queensland, South Australia, Western Australia and the Northern Territory. The ICF was constructed from a list of communities using the 2006 Census population counts and the Discrete Indigenous Communities Database to identify potentially new communities as well as those communities that no longer exist. The non-community frame consisted of a list of SA1s and utilised estimates of the number of Aboriginal and/or Torres Strait Islander dwellings of each SA1 as determined by Census counts. Where available, 2011 Census counts were used for the non-community frame, while the 2006 Census counts were used for the ICF and the remaining non-community frame.

In New South Wales, Queensland, Western Australia, South Australia and the Northern Territory, the sample design differed by community and non-community areas (and non-community areas was further broken down into non-remote and remote non-community). In Victoria, Tasmania and the Australian Capital Territory the sample design was the same throughout, corresponding with the method used in non-community areas of the other states and territories.

To achieve the design objectives, state/territory by non-remote/remote sampling fractions for responding persons were set as shown in the following tables, which also depict the corresponding expected fully responding persons.

NATIONAL ABORIGINAL AND TORRES STRAIT ISLANDER HEALTH SURVEY, State/territory by remoteness sample
	NSW	Vic.(a)	Qld	SA	WA	Tas.(b)	NT	ACT	Aust.
Non-remote
Responding sampling fractions	1/153	1/51	1/146	1/39	1/57	1/28	1/31	1/15	1/79
Expected fully responding persons	1 250	900	1 000	750	850	850	400	400	6 400
Remote
Responding sampling fractions	1/30	0	1/43	1/15	1/36	0	1/36	0	1/35
Expected fully responding persons	300	0	800	400	900	0	1 450	0	3 850
(a) The Victorian remote population benchmark has been used as part of the non-remote sampling fraction and has been applied to the non-remote benchmark for weighting.
(b) The Tasmanian non-remote sample includes sample from remote areas, due to the small sample size in this state. The non-remote sampling fraction therefore includes the remote population benchmark. However, data from the respondents in remote Tasmania will contribute to remote estimates in output products.

NATIONAL ABORIGINAL AND TORRES STRAIT ISLANDER NUTRITION AND PHYSICAL ACTIVITY SURVEY, State/territory by remoteness sample
	NSW	Vic.(a)	Qld	SA	WA	Tas.(b)	NT	ACT	Aust.
Non-remote
Responding sampling fractions	1/244	1/398	1/259	1/277	1/289	1/198	1/299	1/200	1/260
Expected fully responding persons (2 years and over)	748	109	536	101	161	114	40	28	1 837
Remote
Responding sampling fractions	1/31	0	1/36	1/45	1/73	0	1/47	0	1/46
Expected fully responding persons (2 years and over)	282	0	915	128	432	0	1062	0	2 819
(a) The Victorian remote population benchmark has been used as part of the non-remote sampling fraction and has been applied to the non-remote benchmark for weighting.
(b) The Tasmanian remote population benchmark has been used as part of the non-remote sampling fraction and has been applied to the non-remote benchmark for weighting.

Actual numbers of responding persons are available in the Response rates section of this chapter.

Sample selection

Both the community and non-community selections were performed using stratified multistage sampling. Participants in the 2011-12 National Health Survey (NHS)/National Nutrition and Physical Activity Survey (NNPAS) may be included in the 2012-13 NATSIHS/NATSINPAS non-community sample. Where possible, the sampling methodology attempted to reduce the likelihood of this occurring, although respondents moving between the survey periods and being selected in their new residence is a possibility. However, Very Remote areas of Australia were excluded from the NHS and NNPAS samples to avoid overlap occurring in this geographic region. Discrete Aboriginal and Torres Strait Islander communities were also excluded from the NHS and NNPAS to avoid overlap in the community sample.

Non-community

In the non-community sample frame, Statistical Area Level 1 (SA1) (or CDs in the Northern Territory (NT)) formed the first stage selection units. SA1s/CDs were first stratified by state, remoteness area and the number of Aboriginal and Torres Strait Islander dwellings in the SA1/CD based on 2006 or 2011 Census data. Meshblock level information within SA1s/CDs were also used to assist in targeting smaller areas expected to contain Aboriginal and Torres Strait Islander dwellings. Where available, 2011 Census data was used, however this was not feasible for all selections due to the timing of availability of 2011 Census data.

A sample of SA1s/CDs were then selected from each strata with probability proportional to the number of Aboriginal and/or Torres Strait Islander households in the SA1/CD. This meant that SA1s/CDs with a higher number of Aboriginal and/or Torres Strait Islander households had a greater chance of selection. A random selection of dwellings within selected SA1s/CDs were then screened to assess their usual residents' Indigenous status. Where a dwelling contained one or more Aboriginal and/or Torres Strait Islander usual residents a random selection process was undertaken of the Aboriginal and/or Torres Strait Islander usual residents.

The number of persons randomly selected to participate in the survey from a dwelling depended on their remoteness area and the survey enumerated in. For NATSIHS, non-remote areas had up to two Aboriginal and/or Torres Strait Islander adults (18 years or over) and up to two Aboriginal and/or Torres Strait Islander children (0 to 17 years of age) selected, while remote areas only had up to one adult and/or one child selected. For NATSINPAS, up to one adult and/or one child (2 to 17 years of age) were selected in both non-remote and remote areas.

Community

The remote community sample (as detailed previously) was obtained from a random selection of discrete Aboriginal and Torres Strait Islander communities (and their associated outstations) using information collected in the 2006 Census. Communities were selected with probability proportional to the size of the community, so that larger communities had a higher chance of selection. Within selected communities and outstations, a random selection of dwellings was made. Within selected dwellings, up to one Aboriginal and/or Torres Strait Islander adult (18 years or over) and/or one Aboriginal and/or Torres Strait Islander child (0 to 17 years for NATSIHS, 2 to 17 years for NATSINPAS) were randomly selected to participate in the survey.

Previous Page

Next Page