COMPLETENESS OF CENSUS COUNTS
Censuses can miss people, or even count people more than once, and while many procedures are used to help ensure such problems are minimised some miscounting is inevitable. A method used to assess the completeness of census counts, and to inform the extent to which any adjustments should be made, is a post-enumeration survey. These surveys, conducted immediately after the census, determine whether or not people were properly enumerated and serve to measure the proportions of people (by age, sex, state of residence) who were missed or double counted. Along with the adjustments to census counts made so that they refer to usual residents of Australia, the information from the post enumeration survey is used to adjust census counts to produce official census year population estimates. Australia has typically achieved high coverage in its population censuses with under-enumeration rates of less than 2% in recent censuses.5
NET OVERSEAS MIGRATION AND CATEGORY JUMPING
Although Australia maintains a good system of recording the movements of people to and from its borders, obtaining accurate information about net gains (or losses) of people within any period, is affected by the reliability of the information collected from travellers in regard to their intended duration of stay, whether 'permanent', 'long term' or 'short term'. These distinctions are important because short term arrivals and departures (those of less than 12 months duration) are omitted when producing population estimates: only people moving on a permanent or long-term basis are taken into account. One of the major issues associated with the use of this information relates to the discordance between the intended duration of stay and the actual period of stay among some travellers.
Category jumping is the term used to describe changes in travel intentions from short-term to permanent/long-term or vice versa. The extent of the shifts between categories and their net effect in reducing or increasing the numbers of people who should have been included in population estimates can be estimated by comparing information on travel period intentions with subsequent data on actual outcomes. For example, in regard to Australian residents departing to an overseas destination in a given reference quarter, it is possible to retrospectively look at the numbers who did and did not return within 12 months of their departure date and compare this with the numbers of people who said they would (or would not) return within 12 months. For any quarter the difference between those who said they would return within 12 months and those who did return within 12 months gives the extent of category jumping. The extent of category jumping among overseas visitors to Australia can be similarly determined.
A major component of population change for small areas arises due to internal migration, yet aside from statistics collected from the census such data is difficult to obtain, especially below the State/Territory level. For the purpose of producing intercensal population estimates the ABS has been using Medicare transfers (members of Medicare who register their change of address) as the key source of data about the numbers of people who move between the States and Territories.
While Medicare theoretically covers all Australian usual residents as well as those non-Australian residents granted temporary registration, there are a range of Australian usual residents who do not access the Medicare system, primarily due to access to alternative health services. Such people include some Indigenous persons, defence force personnel, prisoners and persons eligible for Department of Veterans' Affairs Health Services. Furthermore, there are also those individuals who simply do not register their change of address with Medicare when they move, even though they continue to access the Medicare system. As such, Medicare data on interstate movers have a degree of under coverage and various modelling techniques must be used to compensate for these problems.
RECONCILING INTERCENSAL ESTIMATES
Taking the census date population estimates from one census and adding to it the net population gains that arise from the various components of growth, namely births, deaths and migrations (as described in the balancing equation) to give post censal estimates up to the time of the next census rarely produces the same count as that provided by the estimates obtained from the next census. The extent of such intercensal discrepancy is influenced by data limitations: both the data on the components of growth obtained from administrative sources and the census counts themselves are likely to contain some errors. Comparisons of such coincident estimates for recent census dates show that the discrepancies are small (an average of about 0.3% over the last four census years for estimates related to Australia's total population) but that they tend to be larger for estimates produced for each of the States and Territories (close to 3% for the Northern Territory for the estimates relating to the 1991 and 1996 censuses).6 The larger discrepancies associated with the States and Territories than for Australia as a whole indicate that it is with estimating levels of internal migration that, as might be expected, estimation errors are most likely to occur.
POPULATION ESTIMATES OF INDIGENOUS PEOPLE
Obtaining up-to-date population estimates relating to Aboriginal and Torres Strait Islander people in Australia is particularly important because various social indicators (birth rates, death rates and so on) which might be used to compare their wellbeing to the total population, or to monitor their wellbeing over time, depend on reliable population estimates. They are also important, in their own right, in terms of their use for government funding decisions. However, there are a number of issues that affect the production of reliable estimates.
A major issue relates to the fact that available census counts of Indigenous people are known to be substantially affected by the changing propensity of people to say, on the census form, that they are of Aboriginal/Torres Strait Islander origin. Analysis of past census counts have shown increases in numbers of Indigenous people that are larger than would be expected from levels of fertility, mortality and migration. The accepted interpretation for the increased population figures is that there has been an increased propensity to identify as Aboriginal and/or Torres Strait Islander. Thus, there is much less certainty that any post-censal population estimates produced for the Indigenous population would correspond with an actual population count if it were to be taken than is the case for population estimates of the total population. Further compounding the difficulty of producing reliable post-censal estimates and projections of the Indigenous population is poor data quality for Indigenous births and deaths in a number of jurisdictions. Data to establish internal migration are not available because there is no Indigenous identifier on Medical Benefits Service records. Despite these problems the ABS currently produces 'experimental' estimates and projections of the Indigenous population. The ABS advises that these be used with caution when analysing changes in the Indigenous population over time and that the analysis of any changes in rates which have been constructed using these estimates should also be undertaken with great caution.