3228.0 - Demographic Estimates and Projections: Concepts, Sources and Methods, 1999  
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 30/08/1999   
   Page tools: Print Print Page Print all pages in this productPrint All  

Census estimates

2.4. The following two sections discuss how census date estimates and the subsequent post-censal estimates are derived.

Census date population estimates

2.5. Census date population estimates are calculated for each State by sex and single years of age (from 0 years to 102 and over).

2.6. There are four steps involved in arriving at these estimates, the results of which are detailed in Table 2.1:

Step 1.

2.7. Census counts of residents are compiled for each State by single years of age and sex. ('Counts of residents' necessarily means that people counted in the census who usually reside overseas are excluded, such as visiting tourists). The question in the 1996 Census from which these data were derived was as follows:

What is the person's usual address?

      • 'Usual' address is that address at which the person has lived or intends to live for a total of 6 months or more in 1996.
      • For persons who now have no usual address, write 'no address'.
      • For boarders at boarding school or college, give address at boarding school or college.

Step 2.

2.8. The census counts of residents are then adjusted upwards to compensate for census net undercount. In 1996, about 1.8 per cent of the population were not counted by the census, while 0.2% were counted more than once. The precise degree of adjustment is based on estimates of undercount from the Post-Enumeration Survey and to a lesser degree comparisons between census results and independent 'demographic' and other estimates of the population (see Appendix 1: Demographic Estimates of Census Under-counting).

2.9. The Post-Enumeration Survey is a sample survey conducted 3 weeks after the census to estimate the number of people (and their characteristics) who did not complete or were not included on a census form. It also detects instances of multiple counting of individuals but the number of such cases is far outweighed by the number of people who are not counted. The net undercount is therefore the excess of the undercount (people not counted) over the number of instances of multiple counting. Details of the 1996 Post-Enumeration Survey are available in Census of Population and Housing: Data Quality - Undercount, 1996 (2940.0).

2.10. The 'demographic' estimates are an annual population series for Australia as a whole, compiled solely from registered births and deaths and overseas migration data from 1925 onwards (ie. they are compiled irrespective of census counts, see Appendix 1).

2.11. A detailed description of the 1991 and 1996 adjustments for net undercount is contained in Chapter 4.

Step 3.

2.12. Estimates of the number of Australian residents temporarily overseas on census night are obtained from passenger card statistics for those Australian residents returning in the twelve month period subsequent to the census date who indicated that they were overseas on census night. These are added to the adjusted census counts derived in Step 2 to give census date population estimates. The population estimates are then converted to financial year of birth, which for the 6th August 1996 Census-date required adjusting all cohorts for the 37 day period back to 30 June by taking 37/365 of each cohort and adding it to the next one.

Step 4.

2.13. If the census does not fall on 30 June, a further adjustment is necessary to arrive at estimates for that date. For example, the 1996 Census was held on 6 August, and after Steps 1 to 3 were finalised, the population estimates at 6 August had to be back-dated to 30 June. This was accomplished using data from births and deaths registrations, overseas arrivals and departures data and estimates of interstate migration for the period 1 July to 6 August.


2.1:
Components of Estimated Resident Population, 30 June 1996 (a) ('000)

NSW
VIC
QLD
SA
WA
TAS
NT
ACT
Aust.(c)

Components as at 6 August 1996
Census Count, Place of Enumeration
6,038.6
4,373.5
3,368.9
1,427.9
1,726.1
459.7
195.1
299.2
17,892.4
plus
Residents absent interstate
64.1
86.2
26.5
25.7
13.7
10.2
4.1
9.7
240.4
less
Interstate visitors
53.5
26.1
98.3
11.0
20.8
4.3
18.2
7.9
240.4
less
Overseas visitors
43.2
19.4
49.7
5.4
13.1
1.1
5.7
2.1
139.6
equals
Census Count, Place of Usual Residence
6006.1
4414.2
3247.4
1437.2
1705.9
464.5
175.3
299.0
17752.8
plus
Adjustment for undercount (b)
91.4
74.0
57.3
19.3
28.1
6.6
5.7
3.4
285.8
plus
Residents Temporarily Overseas
114.4
77.4
41.2
17.9
34.9
3.3
1.7
6.1
296.9
equals
Estimated Resident Population as at 6 August 1996
6,211.9
4,565.6
3,345.9
1,474.4
1,769.0
474.5
182.7
308.5
18,332.5

Components from 1 July to 6 August 1996
less
Births
8.5
6.2
4.8
1.9
2.5
0.6
0.4
0.5
25.3
plus
Deaths
5.7
3.8
2.8
1.4
1.4
0.4
0.1
0.1
15.7
less
Net Interstate Migration
-1.8
-0.9
2.8
-0.7
0.5
-0.3
0.5
-0.1
. .
less
Net Permanent and Long-term Overseas Movement
6.0
4.0
2.3
0.4
2.0
0.1
0.1
0.1
15.0
less
Category Jumping
0.1
0.1
-
-
-
-
-
-
0.2
equals
Estimated Resident Population as at 30 June 1996
6,204.7
4,560.2
3,338.7
1,474.3
1,765.3
474.4
181.8
308.3
18,310.7

(a)
Final estimates.

(b) On a usual residence basis.

(c) Includes 'Other Territories'.


Post-censal population estimates

Estimates of total population

2.14. Using the census year (ie. 30 June) population estimates as the initial base population, post-censal estimates at the national level are compiled in accordance with births, deaths, and overseas migration. At the State level, an additional item, interstate migration, is included.

2.15. This updating process can be expressed mathematically as follows:


Pt+1=Pt + Bt, t+1 - Dt, t+1 + Ot, t+1 + I t, t+1

where:
Pt=estimated population at the end of period t
Pt+1=estimated population at the end of period t+1
Bt, t+1 =births during period t,t+1
Dt, t+1 =deaths during period t,t+1
Ot, t+1 =net overseas migrants during period t,t+1
It, t+1 =net interstate migrants during period t,t+1

2.16. The resulting updated population estimates (ie. ) then become the base populations in the calculation of estimates for the following period. Estimation of the separate components of population change is detailed in the appendices: (Appendix 2 - Estimating Births and Deaths; Appendix 3 - Estimating Overseas Migration; and Appendix 4 - Estimating Interstate Migration).

Estimates by age

2.17. This basic procedure, outlined above, is used not only to update the total population, but also to update the age disaggregation of the population. Population estimates by single years of age are compiled quarterly by updating the census date estimates by age. These annual updates require data for the components of population change to be classified by financial year of birth in order to match each event to its corresponding cohort in the population. (Population estimates by single years of age at 30 June in effect are a classification of the population according to financial year of birth.) However, classification by financial year of birth is only available for two of the three components of change ie. births and migration. It is not available for deaths because, apart from a few exceptions (each State has its own notification form), death certificates only require notice of the deceased person's age at the time of death; they do not require date of birth. Knowledge of a person's age is not sufficient information to determine whether that person's birthday has already occurred in the current financial year.

2.18. Since ABS cannot currently for all States capture information on date of birth of the deceased from death certificates, financial year of birth must be estimated from age at death. This is done by assuming an even distribution of deaths across the year for all ages except ages 0 (under 1) and 1. Infant deaths have a heavily skewed distribution (a large proportion of deaths at age 0 occur in the first four weeks of life), and the conversion of age 0 to financial year of birth takes this into account. The conversion of age 1 to financial year of birth also takes into account the slightly skewed distribution of deaths at that age.

2.19. The conversion from age to birth cohort is based on the following relationship:

where:
=deaths at age x to birth cohort c born in year t to t+1
=deaths at age x in year t to t+1
='separation' factor for deaths at age x

2.20. For deaths aged 0, the separation factor calculated from Australian data is around 0.85, indicating that of the deaths aged 0 in a particular year, 85 per cent were born in that year and 15 per cent were born in the previous year.

2.21. For deaths aged 1, the separation factor is around 0.60. For deaths aged 2 and over, the separation factor is about 0.50, indicating an even distribution of deaths over time at those ages.







Previous PageNext Page