Assessing the utility of 1996 Census data on Indigenous Australians

Author/editor: Hunter, B
Year published: 1998
Issue no.: 154


The large non-biological, or non-childbirth-related, increases in the Indigenous population cast doubt on how much confidence can be placed in 1996 Census data on Indigenous Australians. The credibility of analysis of 1996 Census data on Indigenous Australians hinges on who the people are who have changed their Indigenous identification between the last two censuses. This paper uses three techniques to indirectly examine this question.

Change in Indigenous identification in last two censuses

The Post-Enumeration Survey (PES) conducted in non-remote areas after each census provides an important insight into the sort of people who are likely to change Indigenous identification.

  • The size of the 'marginal' Indigenous population, defined as those who changed identification in the three week period between the census and the PES, fell substantially between 1991 and 1996. Indigenous people appear to be more confident about identifying themselves as Indigenous in the 1996 Census.
  • The percentage of population who identify as Indigenous in either the census or the PES was identical in 1991 and 1996. The number of people who identify as Indigenous is more stable than the prima facie evidence indicates.
  • After accounting for the fact that the PES is not conducted in remote areas the potential Indigenous population, defined as those who identify as Indigenous in either the PES or the census, increases from 314,800 to 383,600 between 1991 and 1996. The 1991 number corresponds quite closely to Gray's estimate of the base Indigenous population (313,500) that must have existed in that year, given the 1996 census results and latest data on fertility and mortality. Therefore is no need to resort to explanations which rely on bogus identification to explain the large increases in the Indigenous population.
  • The age profiles for both the 'marginal' and more consistent Indigenous identifiers are concordant with the higher adult mortality that is characteristic of Indigenous populations.

Cohort analysis of Indigenous and non-Indigenous populations, 1986-96

The second method uses cohort analysis of people born in a particular year to examine changes in education among people who identify as Indigenous.

  • There is no significant change in the proportion of early school leavers for Indigenous identifiers for the last three censuses. That is, there is no evidence of changes in Indigenous characteristics resulting from an influx of relatively well educated individuals.
  • There was a small increase in proportion of the Indigenous population with post-secondary qualifications between 1991 and 1996. However, this may be a result of education policy and active labour market programs in this period as much as a change in the composition of the Indigenous population.

The geography of increased Indigenous identification

The third method looks at changes in identification rates across both urban and rural areas to explore whether there are any obvious spatial concentrations of increased Indigenous identification.

  • Increased identification occurred in similar areas to those areas traditionally identified as Indigenous (low socioeconomic status and high unemployment rates areas).

An excessive focus on whether this newly identified population are Indigenous is undesirable. The continuing high level of disadvantage among the Indigenous population means that self-identification signifies that one is, more than likely, disadvantaged. While it is difficult to say with absolute certainty that census statistics only reflect the economic status of the Indigenous population, they are sufficiently credible, in the opinion of this author, to be taken at face value.

ISBN: 0 7315 2595 7

ISSN:1036 1774

Updated:  16 June 2009/Responsible Officer:  Centre Director/Page Contact:  CASS Marketing & Communications