4.8. The level of coverage of the Australian Census is considered to be excellent, and compares favourably with censuses in other countries. Census data are used for a variety of purposes, without prior adjustment for undercounting. However, because population estimates are used in important ways such as government funding and electoral representation (paragraphs 1.17, 1.19), and given that the level of undercounting is related to important variables such as geographic area, age and sex which are used for population estimates, it was decided that as from 1971, for the purpose of estimating population, adjustments should be made to compensate for undercounting.

4.9. Census undercounting is measured primarily by the Post-Enumeration Survey (PES), a sample survey conducted immediately after the Census. While the first PES was run in 1966, the 1976 PES was the first to be used for population estimates. While population estimates since 1971 have taken account of the undercount, due to inadequacies in the 1971 PES the 1971 estimate was derived by working back from the 1976 estimate using the intercensal data on births, deaths and net overseas movement.

4.10. Details of the 1996 survey are available in Census of Population and Housing: Data Quality - Undercount, 1996 (2940.0). Estimates of census undercount derived from the PES are subject to adjustments based on demographic analysis (demographic adjustments). These demographic adjustments are explained in more detail in Appendix 1 - Assessing estimates of net undercount.

Census Undercount in States

4.11. As shown in Table 4.2, the remarkably consistent rate of undercount for Australia as a whole for the last four censuses - it was between 1.6 and 1.9 per cent in each case - mask variations in the rates for the individual States. New South Wales, Victoria, South Australia and the Australian Capital Territory have shown relatively small variations. In Queensland there has been a considerable and consistent decline - from 3.0 per cent in 1981 down to 1.7 per cent in 1996. In the Northern Territory the rate was very high in 1981 (4.2%) and 1986 (5.5%), but fell to 2.9 and 3.1 per cent in 1991 and 1996 respectively. The Northern Territory has consistently shown the highest rate of undercount of all States.

	4.2: Undercount Rates by State of Usual Residence
		1981 Census		1986 Census		1991 Census		1996 Census
	State	Rate	Standard	Rate	Standard	Rate	Standard	Rate	Standard
		%	error	%	error	%	error	%	error
	NSW	1.8	0.1	1.5	0.2	1.9	0.1	1.5	0.2
	VIC	1.5	0.1	1.8	0.2	1.8	0.1	1.6	0.3
	QLD	3.0	0.2	2.4	0.2	1.8	0.1	1.7	0.3
	SA	1.6	0.1	1.6	0.3	1.6	0.1	1.3	0.3
	WA	1.6	0.1	2.2	0.3	2.1	0.2	1.6	0.3
	TAS	1.0	0.3	1.1	0.3	1.7	0.2	1.4	0.4
	NT	4.2	1.3	5.5	1.4	2.9	0.7	3.1	1.6
	ACT	1.4	0.5	1.6	0.8	1.4	0.2	1.1	0.3
	Australia	1.9	0.1	1.8	0.1	1.8	0.1	1.6	0.1

4.12. The rates of census undercounting by age and sex show that the undercount is greatest in the young adult range, from approximately 15 to 35 years. This is so for both males and females and is shown in Table 4.3. At all ages undercounting is greater for males than for females, averaging 2.0 per cent and 1.1 per cent respectively, for example, in 1996. The highest rates observed in 1996 were for males aged 20-29, averaging 4.1 per cent. The highest rate for females was 2.7 per cent for the 20-24 year age group.

4.13. Census undercounting by age and sex is also assessed by comparison between population estimates derived from that census (ie. census counts adjusted for PES, Australian residents temporarily overseas (RTOs) and backdating to 30 June), and 'demographic' estimates of the population. 'Demographic' estimates (see Appendix 1) are population estimates for Australia as a whole compiled solely from data on demographic events ie. births, deaths and overseas migration. They do not take into account any census data. For 1996, they were compiled from all events since 1925, resulting in estimates of the population age 0-69 years at 30 June 1996. The comparison of the first draft population estimates with the 'demographic' estimates indicated that the PES had not fully detected the extent of undercounting for males aged 20-29 and females 20-24 aged years. Demographic estimates for older ages are deemed less reliable due to the cumulative effect of any small annual errors in overseas migration estimates (particularly category jumping).

4.14. The third means by which age/sex census counts were assessed in 1991 was comparison with the sex ratio (males per 100 females) in each age group of people enrolled for Medicare. This is especially useful at younger ages before the reliability of Medicare data can diminish due to delays in removing deceased persons from the register.

4.15. Table 4.3 summarises adjustments that were made to compensate for undercounting in the 1996 Census in five-year age groups for both males and females. The three measures by which undercounting was assessed, namely the PES, the 'demographic' estimates and Medicare were each given weight ie. none was allowed to over-ride the others where adjusting might have led to unrealistic results. The final undercount was then used in deriving the final population estimate.

	4.3: Comparison of Census Counts, Demographic Estimates and Population Estimates, and the Undercount Rate, by Age and Sex, 30 June 1996, ('000) (a)
	Age group (years)	Census count at 6 August	First draft population estimate	Demographic estimate (b)	First draft minus demographic estimate	Final population estimate	Final undercount rate (%)
	M a l e s
	0-4	648.7	665.8	658.6	7.3	665.6	1.6
	5-9	657.0	669.2	664.3	4.9	669.3	1.3
	10-14	660.4	670.2	661.1	9.1	670.2	1.0
	15-19	639.3	650.4	649.9	0.5	655.3	2.2
	20-24	666.6	705.0	720.2	-15.2	708.9	4.4
	25-29	671.1	711.0	716.2	-5.3	710.5	3.7
	30-34	686.8	732.2	713.7	18.5	720.7	2.4
	35-39	695.7	726.9	713.2	13.7	726.7	2.5
	40-44	651.0	675.9	652.4	23.5	676.1	1.9
	45-49	629.7	651.4	643.2	8.2	654.2	1.7
	50-54	498.3	517.2	533.2	-16.0	517.5	1.4
	55-59	404.8	420.0	419.0	1.0	419.9	1.1
	60-64	340.2	351.8	352.3	-0.5	353.8	1.1
	65-69	325.8	335.8	337.6	-1.7	337.4	1.0
	70-74	268.7	276.1			276.1	0.9
	75+	337.6	345.7			345.7	1.5
	F e m a l e s
	0-4	615.9	635.6	624.8	10.8	631.4	1.5
	5-9	626.0	636.8	632.4	4.4	636.8	1.2
	10-14	629.0	638.0	630.0	8.0	638.0	0.9
	15-19	610.2	623.7	615.9	7.8	623.8	1.8
	20-24	655.6	686.0	691.5	-5.5	688.0	2.7
	25-29	681.9	707.8	697.1	10.7	707.6	1.6
	30-34	702.9	723.6	700.4	23.1	723.8	1.1
	35-39	716.0	729.7	702.0	27.8	729.3	0.6
	40-44	664.0	678.6	643.3	35.3	678.9	0.8
	45-49	625.7	640.2	616.1	24.1	639.7	0.6
	50-54	483.4	497.0	492.4	4.7	497.4	0.9
	55-59	395.1	407.9	391.0	16.9	407.5	0.7
	60-64	343.6	356.4	337.3	19.1	356.7	1.0
	65-69	344.9	354.9	340.5	14.3	354.7	0.7
	70-74	318.7	326.9			327.0	0.9
	75+	555.0	562.0			562.0	0.7
	(a) All figures are on a usual residence basis. (b) Based on post-1925 demographic events. This restricts age-grouping comparisons to under 70 years of age.

4.16. The 1986 and 1991 undercount rates and their standard errors for the marital status categories and for selected countries of birth are shown in Table 4.4. To a certain extent the rates reflect the age structures of the various categories. For example, the never married category, which has the highest undercount rate in the marital status classification, includes a higher proportion of people in the 15-19, 20-24 and 25-29 year age groups, for whom the rate of undercounting is higher. Similarly, in the country of birth classification, the population of the older source (of immigrants) countries, such as Italy and Greece, have older age profiles and lower undercount rates. New Zealand-born immigrants, on the other hand, include a higher concentration in the 15-29 year age groups.

	4.4: Undercount Rates by Marital Status and Country of Birth
			1986 Census			1991 Census			1996 Census
			Rate	Standard		Rate	Standard		Rate	Standard
			%	error		%	error		%	error
	Marital status
		Never married	2.2	0.1		2.8	0.1		3.0	0.2
		Married	1.4	0.1		1.3	0.1		1.1	0.2
		Divorced	(a)	n.a.		(a)	n.a.		1.8	0.3
		Widowed	1.6	0.2		1.7	0.2		1.0	0.4
	Country of birth
		Australia	1.8	0.1		1.8	0.1		1.6	0.1
		New Zealand	4.5	0.7		3.5	0.6		2.8	0.7
		U.K. and Ireland	1.6	0.2		1.6	0.2		1.7	0.3
		Germany	1.8	0.6		0.9	0.5		2.1	0.8
		Greece	1.0	0.4		1.0	0.4		(a)	n.a.
		Italy	0.4	0.2		1.3	0.3		1.2	0.8
		Netherlands	1.3	0.5		1.1	0.6		1.6	1.1
		Former Yugoslav Republics	2.0	0.5		0.7	0.4		0.8	0.6
		Other Europe	1.8	0.6		1.5	0.3		0.8	0.8
		Vietnam	1.0	0.5		(a)	n.a.		2.2	1.0
		Other Asia	2.1	0.5		2.2	0.4		1.8	0.5
		All others	3.0	0.3		2.2	0.4		2.1	0.5
		(a) Rate not considered to be a reliable measure of undercounting.

	4.5: Undercount rates by Indigenous Origin, 1996 Census
		Estimate	Standard error	Relative Standard error (%)
	Indigenous	7.1	1.5	21.5
	Non-Indigenous	1.5	0.1	7.1