The 2002 NATSISS collected information from a maximum of three people aged 15 years or over in each selected private dwelling. The 1994 survey collected information from people in non-private dwellings, such as hospitals and prisons, as well as from people in private dwellings. In addition, the 1994 survey collected information about each person, including children, in each selected private dwelling.
The selection, for the 2002 NATSISS, of a sub-set of residents of a dwelling was designed to manage the provider load on any one household. The detailed questionnaire for this survey would otherwise place a burden on large households. Some information was collected from a household spokesperson on the composition of the entire household, including information on the number of children, so that households with children can be identified in output.
The restriction of the number of people sampled from any one household enabled the total 2002 survey sample to be spread across a large number of households. The 1994 survey sampled 17 800 people in total, 9400 of whom were aged 15 years or over and resident in 4000 private dwelling households. The 2002 NATSISS sampled about the same number of people aged 15 years or over in 5900 households. As a result, the 2002 NATSISS was less clustered than the 1994 survey and there were consequent reductions in sampling error.
While people in non-private dwellings may have characteristics that are different from the rest of the population overall, their exclusion from the 2002 NATSISS was judged to have only a minimal effect on the representativeness of results. By restricting the sample to private dwellings only, a larger number of people could be sampled within the survey budget. Also, the size of the population in non-private dwellings is relatively small. Based on census results, about 4 per cent of the Indigenous population may be resident in non-private dwellings at any point in time.
In addition, there is now a much broader range of administrative statistics available on people in prisons and other corrective service institutions than was available in 1994. In conjunction with the corrective service sector, the ABS publishes information on Indigenous people in prison from the annual prisoner census and quarterly movements in the population in corrective service institutions (ABS 2004e). Furthermore, in recognition that the prisoner population is not static, the 2002 NATSISS collected retrospective information from those respondents who had been incarcerated in the previous five years.
In all data comparisons between the 1994 and 2002 surveys that have been published by the ABS, the population of the 1994 survey has been restricted to align exactly with the 2002 population, namely people aged 15 years and over living in private dwellings.
The 2002 NATSISS was designed to produce reliable estimates for all eight Australian States and Territories, resulting in higher quality estimates at the State and Territory level (particularly for Victoria and the Australian Capital Territory) than was the case for the 1994 survey (e.g. see the employment data in Fig. 3.2). In addition, robust estimates by ABS remoteness categories are available at the national level. While the 2002 survey was not designed to produce other regional data, reasonably robust estimates for some data items for some ATSIC regions can be produced. In contrast, the 1994 sample was spread across the then 35 ATSIC regions and the Torres Strait area to provide some estimates for each region. Under the 1994 strategy, estimates at the regional level were nevertheless associated with relatively high sampling errors.
