Population estimates by marital status
2.24 Prior to 30 June 1998, annual estimates of the population by marital status were compiled quarterly and published annually as at 30 June by single year of age and sex for Australia as a whole. They were based on Census year estimates and were updated using the component method (i.e. using data from births, deaths, marriages, divorces and migration records).
2.25 The procedures used to create population estimates by marital status prior to 30 June 1998 can be found in the previous edition of this publication, Demographic Estimates and Projections: Concepts, Sources and Methods, 1999 (cat. no. 3228.0).
2.26 From July 1998, data on the marital status of overseas arrivals and departures were only provided on visa applications (and only for certain visa classes). Since these data were not available for a high proportion of movers, mainly Australian and New Zealand citizens, the 1997-98 distribution of overseas arrivals and departures by age, sex and marital status was used for the overseas migration component of ERP estimates by marital status for 30 June 1999 and 30 June 2000.
2.27 From 30 June 2001, estimates by marital status were only calculated after each Census. State and territory-level estimates were also calculated for Census years only.
2.28 The marital status categories for which the estimates are produced (i.e. single, married, widowed and divorced) refer to 'registered' marital status. People who state at the Census that they are 'separated but not divorced', for example, are classified in the estimates as 'married'. People in de facto relationships were classified according to their response to Question 6 "What is the person's present marital status?" in the most recent Census (i.e. never married, widowed, divorced, separated but not divorced, married).
Census year population estimates by marital status
2.29 In 2006 population estimates by marital status, single year of age and sex were produced using the following steps:
2.30 Census counts of residents are compiled by single year of age and sex for each marital status category. ('Census counts of residents' means that overseas visitors who are counted in the Census are excluded.) Marital status is imputed for people who did not respond to this question on the Census form. (In 2006 marital status was imputed from other details supplied on the Census form or, failing that, information provided by the Census collector.)
2.31 Net undercount rates derived from the PES for each marital status category, are applied to the Census counts in Step 1 and scaled to state/territory or Australian-level PES-adjusted Census counts for each age and sex.
2.32 In 2006, RTOs by age and sex for each marital status were added to the adjusted Census counts derived in Step 2 to give Census date age/sex marital status population estimates. These RTOs were synthetically derived from RTOs by state/territory, sex and age, 2006 PES data by marital status and sex and 2006 PES data by state/territory, age and sex, using an iterative proportional fitting procedure (see Appendix 5 - The iterative proportional fitting procedure).
2.33 Census data for ages 15 to 19 are particularly susceptible to anomalies from such things as stray marks on the Census form, imputation procedures and mischievous responses. When compared with data from the state and territory Registrars of Births, Deaths and Marriages, numbers of young divorcees, widows and married persons in the Census appear high. To overcome this, demographic estimates are used for this age group. This involved 2001-06 registration data by age for brides and grooms, divorces and deaths. Age of surviving spouse (i.e. widows' age) is derived using a Census matrix of relative spouse ages. The marital status of recent overseas migrants to Australia aged 15 to 19 is also included after minor demographic adjustment.
2.34 As the Census does not fall on 30 June, a further adjustment is necessary to backdate these estimates to 30 June. For example, the 2006 Census was held on 8 August, and after Steps 1 to 4 were finalised the population estimates at 8 August were back-dated to 30 June. This was accomplished using death registrations and interstate migration data for the period 1 July to 8 August 2006. Finally the figures were constrained to state/territory and Australian-level 30 June population estimates by age and sex. This also helps overcome the lack of overseas migration data for backdating and introduces the age adjustment for persons with birthdays between 30 June and Census night described earlier.
Population estimates by country of birth
2.35 Quarterly population estimates by country of birth are compiled and published annually as at 30 June for Australia as a whole. These estimates, produced by single year of age and sex, classify the population according to countries of birth.
Census year population estimates by country of birth
2.36 Population estimates by country of birth at 30 June in the Census year are compiled as follows:
2.37 Census counts of residents (i.e. excluding overseas visitors) are compiled by single year of age and sex for each country of birth. Country of birth is imputed for those people who did not respond to this question on the Census form. This is achieved using the distribution of country of birth from the PES.
2.38 Net undercount rates are derived from the PES for the 10 countries with the highest Census representation (including Australia), plus a remainder category. These rates are applied to the results from Step 1 and constrained to total Australian age and sex PES-adjusted Census counts. This yields Census counts adjusted for undercount and not-stated country of birth. The eleventh, that is the remainder, country category then serves as a constraint for scaling all other (i.e. non top-10) country of birth Census counts.
2.39 RTOs by sex and country of birth are added to the adjusted Census counts derived in Step 2 to give Census year country of birth population estimates. These RTOs are derived from aggregated passenger card information originally provided by the Department of Immigration and Citizenship (DIAC).
2.40 As the Census does not fall on 30 June, a further adjustment is necessary to produce estimates as at the nearest 30 June reference date. For example, the 2006 Census was held on 8 August, and after Steps 1 to 3, the population estimates at 8 August was back-dated to 30 June. This was accomplished using data on birth and death registrations and overseas arrivals and departures for the period 1 July to 8 August 2006.
2.41 Data from Step 4 are converted to financial year of birth then forced (see Appendix 5 - The iterative proportional fitting procedure) to add to the 30 June population estimates of all residents by financial year of birth and sex (as described in paragraph 2.20).
2.42 Population in non-specific country of birth categories (e.g. "United Kingdom - Not Further Defined") are redistributed pro rata across the countries in that region.
2.43 For 30 June estimates in Census years only, population estimates by country of birth, age and sex are also cross-classified by state/territory. This was achieved using a similar approach to that outlined above, and included an interstate migration backdating component and simultaneously constraining by age and sex to both Australian-level country of birth estimates and state/territory level estimates. Post-censal country of birth estimates by state and territory are not available as there is insufficient data on interstate migration by country of birth.
Post-censal population estimates by country of birth
2.44 Quarterly population estimates by country of birth for post-censal years are compiled by updating the Census year estimates in accordance with births, deaths and overseas migration. Each component of change is first converted to financial year of birth. The population for each country of birth by birth cohort are then updated as follows:
For the Australia-born population
For all other countries of birth
2.45 For each component, persons with non-specific country codes are distributed pro rata across the countries in their region.
2.46 Prior to the introduction of the new NOM method (see Chapter 6 - Estimating net overseas migration
), there had been concern about the reliability of long-term overseas migration data, including the manifestation of some countries with many long-term arrivals but far fewer departures. Therefore between
2001 and 2006 overseas migration by country of birth was derived from permanent flows only, constrained to net Australian levels.
Experimental estimates of the Indigenous population
2.47 After each Census, estimates of the Indigenous (Aboriginal and Torres Strait Islander) population by age and sex for the states and territories and Australia are compiled for the Census year and ten or fifteen preceding years. Therefore while estimates are available from 1986, at the time of publishing only 1991-2001 are on the same basis (2001 Census-base), and 1991-2006 on the 2006 Census-base are expected to be released later in 2009.
2.48 The estimates are 'experimental' in that the standard approach to population estimation is not possible because satisfactory data on births, deaths, overseas and internal migration are not generally available. Furthermore, there is significant intercensal volatility in Census counts of the Indigenous population, thus adding to the problem of estimating the true Indigenous population. This volatility can in part be attributed to the change in propensity for persons to be identified as Indigenous as recorded on a Census forms.
2.49 Since 2001, Census year Indigenous estimates have been produced concurrently with statistical local area (SLA) population estimates. This is so the Indigenous estimates better reflect undercount rates by Indigenous status and that SLA estimation by Indigenous status is not a post-hoc process. This is described in greater detail in Chapter 3 - Estimating population for statistical local areas
2.50 The process involved apportioning usual residence Census counts of persons with not-stated Indigenous status to Indigenous or non-Indigenous, by SLA, age, sex and Census form-type. Applied to these were synthetic net undercount adjustments by state/territory, age, sex and Indigenous status, then age structures adjusted to correspond to 30 June. SLA by Indigenous status estimates were finally derived through constraining by age and sex to synthetic capital city/balance of state/territory population estimates and SLA age-sex totals, adjusting for RTOs and backdating population components from Census date to 30 June.
2.51 For 30 June 2001, Australian and state/territory Indigenous estimates were 'survived' back to 30 June 1991 using life tables. This method is known as the reverse survival method and assumes zero net internal and zero overseas migration.
2.52 For more information see Chapter 8 - Estimating the Indigenous population