3108.0 - Demography Working Paper 1979/1 - Population Estimates in Australia: A Discussion Paper, 1979  
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 28/07/2001   
   Page tools: Print Print Page Print all pages in this productPrint All

Foreword

This 1979 paper has been placed on the ABS web site given its historic importance in the development of Australia's population estimates. Following discussion associated with the 1979 paper, population estimates were placed on a usual residence basis as outlined in Population Estimates: An Outline of the New Conceptual Basis of ABS Population Estimates (ABS Catalogue No. 3216.0) released on 29 March 1982. At the State and Territory level population estimates based on this new concept were constructed back to 1971. More information on ABS concepts is in Demographic Estimates and Projections: Concepts, Sources and Methods.

In a note to Dr Frank Horner of 19 September 1979, Dr Choi acknowledges that 'much of the paper was originally written by you'. He also notes that 'The ABS has decided to produce resident-based population from 1981. Mr Cameron (the then Australian Statistician) has decided that we cannot avoid to have an 'official' set'.

John Paice
Director
Demography
26 June 2001


C.Y. CHOI
Central Office
Australian Bureau of Statistics
July 1979

Views expressed in this paper are not necessarily those of the Australian Bureau of Statistics.
Demography and Social Branch
R series: Demog-1


This paper discusses the background to the Post-1976 changes made to population estimates and describes the methodology employed in the current series. It raises significant problems surrounding the adequacy of the current series and the options for future estimates.

Readers are invited to comment on the issues raised and in particular to express opinion on the desirability or advisability of the options.


GENERAL

Information regarding the size of the population of an area can be obtained by periodic census enumerations, records of births, deaths and migration, and in some countries by a continuous population register. Such information can be supplemented by various other data relating to population size, eg school enrolments, number of electors registered, number of children receiving child-allowance. In countries where population censuses are regularly conducted and where there is a reliable registration of births, deaths and migration, the basic process of estimating population is to update a 'stock' of population (a census count) using 'flow' statistics of births, deaths and migration.

For Australia, where both population stock and flow data are available regularly, the basic procedure is to add to the population at the beginning of a period, natural increase (births minus deaths) and net overseas migration (arrivals minus departures). For State population estimates, an additional item, interstate movement, is included. This basic procedure is used not only to estimate the total population, but also to estimate the ages of the population. Thus, in the calculation of age estimates, the procedure is applied to each age to give the estimate for the next higher age in the next year.

Much of the work involved in population estimation is in the assessment of the accuracy of each of these components of population change used in the basic formula. Once these components and their adjustments are determined, the estimation task becomes a mechanical process.


POPULATION ESTIMATES UP TO 1971

The normal practice, up to and including 1971, in the estimation of the population at the national and subnational levels had been to accept the census count as the best estimate of the population. Adjustments had been found necessary in many cases to "correct" mis-statement of ages, usually by the smoothing of heapings at ages ending in particular numerals such as fives and zeros, but there had not been any attempt to adjust the total. In the 1962 Year Book of the Commonwealth of Australia, it is noted that "the results (of the 1961 Census enumeration) attain a very high degree of accuracy and may be accepted without reservation." In 1966, a post-enumeration survey was conducted after the census, and results showed that under-enumeration was small, about 0.5 per cent for the national total.

Using the census as a benchmark, estimates for subsequent years were made applying the basic formula until the next census was conducted. The normal practice of amending previous population estimates following a census involves substituting the census figure for the current estimate at the census date. The difference between the census figure and the estimated figure (this difference is sometimes called the intercensal adjustment) is then interpolated for all the intercensal years back to the date of the previous census. The following table shows these differences for periods before 1971:


POPULATION FIGURES 1921-1971, Australia


Census
Census count(a)
Estimated from
previous census
Difference

'000 PERSONS

1911
4,454
4,525
-70
1921
5,435
5,447
-13
1933
6,630
6,619
11
1947
7,579
7,560
19
1954
8,987
9,002
-16
1961
10,508
10,515
-7
1976
11,550
11,574
-23
1971
12,756
12,812
-56

(a) as enumerated; no adjustments made.

For periods before 1933, the reason for intercensal adjustments was considered to be primarily an under-count of overseas migration. Net migration statistics were therefore adjusted to incorporate the discrepancies. In the periods before 1921, under-count was considered to be particularly serious with departure statistics, and this necessitated downward adjustments to net overseas migration figures.

In the period 1921-33, when the required adjustment was of a different direction - upwards rather than downwards - it was recognised that other factors might also have contributed to the discrepancy. The Statistician's report on the 1933 Census (p.26) mentioned possible deficiencies in birth registration and overseas arrivals, and the likelihood of a higher standard of accuracy in the results of the 1933 Census than in those of the 1921 Census. However, the past practice of absorbing the intercensal discrepancy by revising net migration statistics was continued.

The 1947, 1954 and 1961 Statistician's Reports recognised similar sources of deficiencies. In addition, the 1947 Statistician's Report mentioned the possibility of a greater degree of error in overseas migration records because of the disturbances of the 1942-45 War years. For periods after 1933, intercensal adjustments were shown separately, not incorporated in the migration statistics series. Nevertheless, the rather large discrepancies for the 1933-47 and the 1947-54 periods were thought to be mainly caused by the inadequacy of overseas migration statistics and inaccuracies of census counts.

While there had been some mention of inaccuracies in vital registrations (births and deaths), these had not been considered serious enough to cause the extent of the discrepancy experienced in population estimates. Neither were adjustments for census under-enumeration considered. This situation continued for the 1961-66 period. The intercensal discrepancy remained low, although there was an increase in the 1961-66 period.

In 1971, the post-enumeration survey indicated that census under-enumeration was around 1.35 per cent. The intercensal adjustment required to make the recorded population increase during the 1966-71 period agree with the population change indicated by the censuses was now 56,000. This was a large increase over the intercensal adjustment of only 7,000 for the 1954-61 period. The increase in the discrepancy from 1954-61 onward suggests growing under-enumeration at the censuses, as the collection of overseas migration statistics had improved and the registration of birth and deaths were considered to remain satisfactory and acceptable. Despite the known under-enumeration at the 1971 census, no upward adjustment was made to the 1971 Census count, and the census result was the figure used to describe the population size at 30 June 1971. The discrepancy was absorbed into the population series for 1966-71 by interpolating over the period in a way similar to the practice of previous periods.

Estimates of the population of States and Territories were complicated by the increasing difficulty of estimating movements across State and Territory boundaries. For periods prior to June 1966, net interstate movements were estimated from records of all movements by sea, air and rail. From June 1966, internal movements were estimated using records of interstate change of address on child endowment (now family allowance) records and electoral rolls. Holiday, business and other short-term movements were thus no longer taken into account. Correspondingly, overseas arrivals and departures were allocated to each State and Territory according to their State of intended or previous residence in lieu of their State of embarkation or disembarkation. The change was made because it was recognised that the estimates based on air, sea and rail travel were inadequate as measure of total net interstate movements (interstate travel by road was not recorded) and that measuring change of residence would avoid seasonal fluctuations in the population of some States (for example, the increase each year prior to this change in the December population of Tasmania and the June population of Queensland). This has meant that while the State population figures at census time measured the number of persons actually in the State including overseas and interstate visitors (ie the de facto population), the estimates of changes in the population reflected residential interstate movements. The overseas migration component, however, continued to include short-term movements.

This method of estimating State populations was introduced in 1967 and estimates back to 1961 were revised accordingly.


POST-1976 CENSUS ESTIMATES

After the 1976 Census, the ABS made two important decisions which changed the previous population estimation procedure. One was the decision to adjust the population for under-enumeration at the census, and the other was the exclusion of short-term overseas movements in the calculation of migration gain from overseas. (Short term movement is movement for an intended or actual stay of less than 12 months).

To explain the reasons for these changes, it is necessary to describe a number of new developments which have complicated the task of estimating Australia's population.

a The 1976 post-enumeration survey, carried out immediately after the census as a check on its coverage and accuracy, revealed a relatively large degree of under-enumeration (2.71 per cent compared with the apparent 1.35 per cent in 1971). A study was made after the 1976 Census to examine the accuracy of the 1971 and 1976 Census figures by comparing them with independent population estimates for particular age groups. These independent sources included birth, death and migration statistics (for the construction of the population at census dates for persons under 19 years of age), school enrolments, and child endowments (family allowance). The investigation supported the results of the post-enumeration surveys which showed a considerable growth of under-enumeration at the census since 1966. It also provided some confirmation that the extent of overall under-enumeration at the 1976 Census was of the order revealed by the 1976 Census post-enumeration survey, ie 2.71 per cent. It further indicated that the 1971 post-enumeration survey may have been deficient, ie under-estimated the extent of under-enumeration at the 1971 Census. Appendix 1 contains some of the results of this investigation.

b Since the 1971 Census, the number of short-term overseas arrivals and departures has become so great, and their seasonal movement from quarter to quarter so marked, that the Australian population estimated in the usual way has fluctuated widely from quarter to quarter. In recent years, this fluctuation has resulted in the June quarter population being lower than the March quarter population. The timing of these short-term movements is unfortunate both for the June estimate which is used for numerous purposes and for the December estimate which is used as a benchmark for annual grants to the States by the Commonwealth. There is a build-up of departures by Australian residents in the four months prior to June 30, and in the month of December there is another, more concentrated exodus. (this amounted to 81,000 in December 1976). Appendix II shows the increase of short-term movements in recent years and their seasonal fluctuations.

c To a large extent as a result of the build-up of short-tem passenger movements, there was a net short-term loss of population between the 1971 and 1976 Census dates of 79,000. Australian residents departing for less than a year in 1970-71 numbered 377,000; in 1975-76 they numbered 951,000. In each intervening year the short-term departures of Australian residents exceeded the numbers returning, and the cumulative loss on this account for the five fiscal years ended June 1976 wag 67 000. It is clear that the lag effect due to the build-up in the short-term movement of Australian residents must have played a substantial part in the net loss of population by short-term movement during the 1971-76 period. (There is, of course, the contribution from some Australian residents who departed short-term but did not return, and those who stayed abroad longer than they had intended so that they return in the long-term category).

The post-censal amendments to the population take into account these new developments.

It was decided that the under-enumeration in the 1976 Census of 2.71 per cent was large enough to make it advisable to adjust the 1976 Census benchmark to include it. This was the first time this had been done in Australia. (Other countries which conduct post-enumeration surveys usually find an appreciable amount of under-enumeration - the figure for the United States Census in 1970 was 2.5 per cent - but it is not customary for the enumerated figures to be adjusted for it). Appendix III discusses under-enumeration at the 1976 Census.

The problems of the apparent fluctuation of population due to short-tem overseas movements, and of the apparent loss of population due to the build-up of short-term departures of Australian residents, are consequences of the de facto basis on which Australian population estimates has been traditionally made. The Census is collected on a de facto basis - that is, by enumeration of all people who spent the Census night in Australia - and consistently with this the post-censal population estimates in the past have been moved forward by net immigration figures which included short-term movement as well as long-term and permanent movement. One solution to these problems, therefore, would be to adopt a resident basis for the population estimates instead of a de facto basis. The term 'resident based population' is used here to refer to the population which is defined by their usual place of residence rather than by where they happen to be at a particular time. Census enumerations based on the principle of usual residence, eg in the US, are sometimes known as 'de jure' censuses.

(In this Paper, the term 'de jure' is avoided because it contains a legal connotation which is not implied in the discussions. For a description of the definitions of the concept of population, see M. Spiegelman, Introduction to Demography, 1968, Harvard University Press, p. 9-13. Also United Nations, Handbook of Population Census methods , 1955, Vol III, p. 55-59.)

To do this for the nation as a whole, the census, after adjustments for under-enumeration if required, will need to be put on a full resident base, ie by adding to it the estimated number of Australian residents temporarily overseas and subtracting from it the estimated number of overseas visitors in Australia. This resident population base is updated by adding to it natural increase and net long-term and permanent overseas movements. For State estimates, an additional step is required to it interstate visitors recorded at the census back to their usual place of residence. State estimates are updated using natural increase, net permanent and long-term overseas migration (by State) and net interstate migration.

In many countries, short-term movements are inadequately recorded and are not included in the measure of net gain from international movements. Appendix IV describes the current methods of population estimation in Great Britain, the USA and Canada. In these countries, the lack of adequate international migration statistics makes it necessary to conduct de jure censuses to include residents temporarily overseas in the resident population base.

Although it is possible to estimate the net short-term absentee population at Census date 1976 from the present statistics of short-term arrivals and departures, more work needs to be done on these estimates before ABS will be ready to change the conceptual basis of the population estimates from de facto to resident. As already mentioned, people who intend to be absent for less than a year may stay longer, and the estimates may be affected by other kinds of change of category ('category jumping'). Among these are cases where people who have not changed their minds about their period of absence put themselves in different categories when completing their arrival and departure cards. It is believed that the extent of this has diminished since the new passenger card was introduced in January 1974, but the matter needs more investigation.

It is also necessary to ascertain the extent to which overseas visitors have extended their short-term stays. Information from the 1976 Census about short-term visitors in Australia will need to be analysed and compared with the estimates from overseas migration statistics.

While it was impracticable to put the population benchmark onto a full resident basis immediately after the 1976 Census, the exclusion of short-term overseas movements for the updating from the census de facto benchmark avoids the fluctuations in the population series caused by the seasonal ups and downs of short-term visitor movements; it also avoids the apparent loss of population due to the build-up of short-term departures of Australian residents.

The amended population estimates published after the 1976 Census and revised back to 1971, therefore do not embody a change to a full resident population concept; however, they do measure net migration gain by permanent and long term movement only.

The effect of the above change is that changes in the population are measured on a resident basis, although the 1976 Census benchmark is a de facto figure.

For the States and Territories, the post-1976 population series based on long-term and permanent overseas migration is more conceptually homogeneous than past series. Instead of being de facto in concept for overseas migration and resident in concept for interstate movements, it is now resident in concept for both.


BACK REVISION TO 1971

For Australia as a whole, back revision to 1971 adopts the same method as updating from 1976, ie accepting the 1976 Census figure adjusted for 2.7 per cent under-enumeration as the base, and backdating this using natural increase and net long-tem and permanent overseas migration.

The following table presents the calculations involved:


'000

Census 1976 adjusted for under-enumeration
13,915.5
. Less natural increase 1971-76*
697.2
. Less net migration 1971-76 (permanent and long-tem)
281.2
Estimated 30 June 1971 population
12,937.2

* Includes adjustment to balance the number of those aged 0-4 years in the adjusted 1976 Census figure with the number accounted for by births, deaths and net migration in the period.

On this basis the estimated 1971 population for Australia is 12 937.2 thousand which is some 10 thousand higher than the 1971 Census figure adjusted for 1.35 per cent under-enumeration as revealed by the 1971 post-enumeration survey (12, 927.8 thousand). This difference of about 10 thousand is due partly to:

1) the revised 1971 population containing an implied estimate of under-enumeration at the 1971 census which is higher than that measured by the 1971 PES, and

2) the contribution from the change in conceptual base.

The contribution from the change of conceptual base consists of the difference between 1971 and 1976 in the net number of overseas visitors, ie Australian residents temporarily overseas minus overseas visitors in Australia.

The 1971 State and Territory population estimates are calculated by making 2 adjustments to the 1971 Census population as enumerated. These are:

1) adjustments for 1971 Census under-enumeration and

2) adjustments to discount effects of overseas short-term movements during 1971-76. This brings the 1971 to 1976 population estimates onto the same basis as the Post-1976 estimates.

The following table shows the steps taken in the estimation of 1971 population by States and Territories.


1971 POPULATION FIGURES

State
1971
As enumerated
*Under-enumeration
adjustments
**Adjustment for
short-term overseas
movements
1971 estimated population

'000 PERSONS

NSW
4,601.2
101.7
-23.5
4,679.4
Vic
3,502.4
56.9
-38.9
3,520.4
Qld
1,827.1
48.8
+5.5
1,881.4
SA
1,173.7
18.8
-7.2
1,185.3
WA
1,030.5
22.7
-10.0
1,043.1
Tas
390.4
2.0
-2.2
390.2
NT
86.4
6.9
-1.4
91.9
ACT
144.1
3.0
-1.5
145.6
Australia
12,755.6
260.9
-79.3
12,937.2

* These estimates of 1971 under-enumeration are obtained by backdating the estimated 1976 population for Australia (which includes adjustment to under- enumeration) to 30 June 1971 using natural increase and net total overseas migration, and comparing this with the 1971 Census count. This gives an estimated under-enumeration for Australia, both sexes, of 260.9 thousand or 2.04% of the 1971 Census count. This is a higher estimate of under-enumeration than the one measured by the 1971 Census post-enuneration survey (172.2 thousand, or 1.35 per cent of the census count). The ratio of the two estimates (calculated separately for males and females) was then multiplied to the amount of under-enumeration measured by the 1971 post-enumeration survey for each State and territory.
** This is the net overseas movement of short-term duration to each State and Territory during the period 1971-76. Short-term movements are defined as movements for an intended or actual stay of less than 12 months.


EFFECT OF THE CHANGES MADE

In summary, the effects of the changes made since June 1976, described above are:

a The whole population series is determined by the 1976 Census figures adjusted for under-enumeration, which are considered to be the best estimates of the Australian and State populations for 30 June 1976.

b From June 1976 on, estimates are based on changes in population on a resident basis, both for Australia and the States, although the Census 1976 benchmark is de facto in concept.

c Between June 1971 and June 1976 for Australia as a whole estimates of the increase in population are also on a resident basis.

d For States and Territories, the method used in producing both the 1971 and 1976 estimates implies the inclusion of interstate visitors at both dates. Estimates for each State and Territory in the 1971-76 period takes into account net long-term and permanent overseas migration, natural increase and net interstate movements estimated from changes of residence in family allowance and electoral roll records. Adjustments are also made to correct for incomplete recording of interstate movements and to take into account the difference in the net number of visitors in each State and Territory between 1971 and 1976.


POPULATION ESTIMATES FOR LOCAL GOVERNMENT AREAS (LGA)

Population estimates at the LGA level are produced once a year, as at June 30. While they are a subset of the population estimate series, they are estimated using a different methodology. At the LGA level, the basic formula which relates natural increase and net migration to the census benchmark cannot be easily adopted. This is because movements across LGA boundaries are often the most important, sometimes the overwhelming component of LGA Population change, and these movements are not recorded or measured accurately.

The normal practice in updating LGA census benchmark has been, and still is, to estimate the increase or decrease of the population using relevant indicators of which increases and decreases in the stock of private dwellings are of which major importance. The utilization of dwelling figures takes cognizance of the change in occupancy rates over time and also of the variation in these rates between suburbs. Census data as well as information of local developments are also used in the estimation of population growth. In some ABS State Offices (eg NSW) systematic records are kept of information relating to population growth and decline in each LGA.

More recently, in order to use the various indicators more fully and systematically, a regression model has been developed and is being introduced to examine the strength and the stability of relationships between population growth and the growth of various indicators. Indicators include school enrolments, number of children endowed (family allowance), number of age pensioners, stock of dwellings, births and deaths. The use of the regression technique lessens the reliance on stock of dwellings and occupancy rates, and appears to be an improvement over previous methods.

The resulting updated LGA estimates are forced to sum to the given State total which is independently derived using the basic formula. Because LGA estimates are forced to add to a given State population, they take on the characteristics implied in the methodology used in estimating national and State populations. The current mixed concept on which national and State populations are based is therefore also reflected in LGA estimates.


FUTURE OPTIONS

The post-1976 changes made to the procedure of estimating population as described in the above paragraphs represent important departures from previous practice.

Firstly, they represent a recognition that population censuses contain an important element of under-enuneration which needs to be allowed for in population estimates. This means that a clear distinction must now be made between census population data and population estimates. Because of the problem of differing under-enumeration in successive censuses, the ABS recognises that census data 'as recorded' is not always suitable for the calculation of population growth.

Secondly, they indicate that the large temporary movement of persons overseas for short-term visits can no longer be included as a measure of population change without distorting the more stable underlying trend of growth of the population. This situation applied equally to interstate movements and to overseas movements. The current procedure of excluding short-term overseas and interstate movements avoids this distorting effect but introduces conceptual complexities in relating series of essentially resident based vital and migration statistics to a de facto census base.

The adoption of a full resident population base appears to be a logical and natural extension of changes made so far. The ABS is examining the steps needed to produce a fully resident base population series soon after the 1981 Census. To place the 1981 Census itself on a resident base will involve considerable effort. It will be necessary to determine the age, sex, and LGA of usual residence of all Australian residents temporarily overseas at the time of the Census. A straightforward extension of the current system that gives State level figures, leads to a massive coding operation. For persons in Australia, but away from their usual residence, it is a simple matter to return them to their LGA of usual residence if they state it on their Census form, but not otherwise. A resident base also compounds the problem of adjusting for under-enumeration as the population missed by the Census could be expected to be proportionately more mobile. The ABS is investigating ways of overcoming these problems.

The adoption of a full resident population series is an option which will avoid the distortions introduced by short-term movements, and will put Australia's population statistics on a consistent basis, overcoming the conceptual difficulties which the current series suffers. This will also bring Australia to the forefront of developed nations in their attempt to estimate population on a full resident basis.

It can be further argued that a resident population series is more suited than a de facto one for the determination of representation in the political system. Temporary overseas visitors do not have the right to vote in Australian elections and they should, therefore, not be included in the determination of representation in national and State governments. A similar argument can be extended to the treatment of interstate visitors who vote in their State of usual residence.

However, the use of population data is not limited to the distribution of electoral representation; population data are basic to various economic and social analyses including research for forward planning to meet future requirements. In Australia, populations of States and local government areas are important elements in the calculation of income tax sharing allocations. It appears reasonable to suggest that the nature of particular inquiries, research and analyses may be such that population series on a full resident basis would not be the most appropriate.

If the distortions introduced by short term movements are to be avoided but a de facto base required then the present mixed concept could be retained. There may also be some purpose for which a fully de facto population series, even though it contains seasonal fluctuations, may be superior to the fully resident series or the mixed-base series. (However, it should be noted that while currently available data allow the compilation of a fully de facto population series at the national level, this is not possible at the State level because interstate migration is currently measured in reference to residential changes).

Given the above considerations, it may be desirable from the point of view of satisfying a variety of users for the ABS to produce alternative population series. It will, then, became the user's responsibility to judge the appropriateness of alternative population series for their application. The availability of more than one population series is not uncommon, and is a practice adopted by Great Britain and the USA.

The conceptual distinction between a resident based and a de facto based population series is of theoretical importance, although in the past, the practical effect has not been considered critical. The practical aspect will become more important as short term travel increases both internationally and within the country. The recent introduction of low air fares can be expected to accentuate this trend.


APPENDIX I: COMPARISON OF CENSUS RESULTS, ORIGINAL AND ADJUSTED, WITH INDEPENDENT ESTIMATES

The attached table presents a comparison of the 1971 and 1976 Census figures with independent estimates of the population. The table summarises data derived for age groups 0-4 to 15-19 from births, deaths and immigration statistics, for age group 5-9 from school statistics, and for age groups 5-9 and 10-14 from child endowment statistics in 1976.

The table illustrates the following points:

1. It confirms the existence of under-enumeration in both the 1971 and the 1976 Censuses.

2. It provides some confirmation of the PES adjustment in 1976.

3. The schools figures appear to confirm that the 1971 Post-enumeration survey (PES) was deficient.

4. The child endowment figures (for 1976 only) tend to confirm the PES adjusted Census figures for age groups 0-4 and 5-9, but fall below them for the 10-14 age group.


TABLE 1A: COMPARISON OF CENSUS RESULTS, ORIGINAL AND ADJUSTED, WITH INDEPENDENT ESTIMATES

Age Group/Year
0-4
5-9
10-14
15-19
Estimate
1971
1976
1971
1976
1971
1976
1976

Census as enumerated
1,223.4
1,199.1
1,220.3
1,256.0
1,226.4
1,245.3
1,218.1
Census as adjusted by PES
1,251.2
1,235.0
1,229.8
1,279.2
1,234.2
1,266.2
1,259.9
Estimate from natural increase and net immigration
1,236.9
1,223.2
1,229.1
1,254.3
1,232.5
1,244.9
1,242.4
Estimate from child endowment
n.a.
1,235.3(a)
n.a.
1,278.8
n.a.
1,254.9
n.a.
Estimate from schools plus census "not at school"
1,237.8
1,295.8

(a) Children endowment figure for 0-4 years is 1,201,800, but this is understated owing to time-lag in new babies being enrolled. The figure has been adjusted by substituting for the 201,100 age 0 in the children endowment figures a figure of 288,300 estimated from natural increase and net immigration.
NOTE: The child endowment figures include 17,972 children in institutions, for whom no age distribution is available. This figure has been distributed proportionately over ages 0-15 in the above table.


TABLE 1B: PERCENTAGE DIFFERENCES BETWEEN ESTIMATED AND CENSUS RESULTS AS ENUMERATED

Age Group/Year
0-4
5-9
10-14
15-19
Estimate
1971
1976
1971
1976
1971
1976
1976

Census as adjusted by PES
2.27
2.99
0.78
1.85
0.64
1.68
3.43
Estimate from natural increase and net immigration
1.10
2.01
0.72
-0.14
0.50
-0.03
1.99
Estimate from child endowment
n.a.
3.02(a)
n.a.
1.82
n.a.
0.77
n.a.
Estimate from schools plus Census "not at school"
n.a.
n.a.
1.43
3.17
n.a.
n.a.
n.a.

(a) Children endowment figure for 0-4 years is 1,201,800, but this is understated owing to time-lag in new babies being enrolled. The figure has been adjusted by substituting for the 201,100 aged 0 in the children endowment figures a figure of 288,300 estimated from natural increase and net immigration.
NOTE: The child endowment figures include 17,972 children in institutions, for whom no age distribution is available. This figure has been distributed proportionately over ages 0-15 in the above table.


APPENDIX II: SHORT-TERM OVERSEAS MOVEMENTS 1971-1978
(Short-term movements are defined as movements of less than 12 months of intended or actual stay.)

A. SHORT-TERM ARRIVALS

Period
Australian Residents
Visitors
Total
As a percent of all Arrivals

1971March Qtr
105,513
111,299
216,812
77.1
June Qtr
84,687
96,966
181,653
78.3
Sept. Qtr
118,818
93,546
212,364
79.1
Dec. Qtr
103,580
130,582
234,162
78.9
1972March Qtr
123,799
113,288
237,087
83.1
June Qtr
98,657
82,626
181,283
80.3
Sept. Qtr
144,597
92,406
237,003
83.1
Dec. Qtr
123,909
138,083
261,992
83.4
1973March Qtr
153,122
118,924
272,046
85.1
June Qtr
126,745
95,537
222,282
83.3
Sept. Qtr
181,439
104,993
286,432
85.8
Dec. Qtr
159,536
152,670
312,206
84.3
1974March Qtr
205,421
135,483
340,904
87.3
June Qtr
158,956
112,708
371,664
84.7
Sept. Qtr
210,796
119,614
330,410
86.8
Dec. Qtr
177,045
164,878
241,923
84.5
1975March Qtr
243,716
134,721
378,437
91.9
June Qtr
184,374
100,337
284,711
91.4
Sept. Qtr
244,060
115,362
359,422
91.6
Dec. Qtr
208,459
165,603
374,062
90.5
1976March Qtr
281,679
143,722
425,401
92.7
June Qtr
197,527
104,969
302,496
91.3
Sept. Qtr
272,389
115,308
387,697
92.0
Dec. Qtr
216,670
167,869
384,539
89.5
1977March Qtr
283,384
147,639
431,023
90.9
June Qtr
191,287
109,579
300,866
89.8
Sept. Qtr
280,644
121,067
401,711
91.6
Dec. Qtr
218,362
184,996
403,358
89.6
1978March Qtr
273,663
156,759
430,422
91.2
June Qtr
216,355
118,060
334,415
90.9



B. SHORT-TERM DEPARTURES

Period
Australian Residents
Visitors
Total
As a percent of all departures

1971March Qtr
82,470
130,188
212,658
86.2
June Qtr
117,817
107,905
225,722
87.8
Sept. Qtr
98,605
97,849
196,454
86.4
Dec. Qtr
115,025
114,080
229,105
87.0
1972March Qtr
97,177
133,268
230,445
86.8
June Qtr
144,396
98,491
242,887
87.0
Sept. Qtr
124,779
95,281
220,060
87.0
Dec. Qtr
138,167
114,280
252,447
88.4
1973March Qtr
113,788
144,853
258,641
88.1
June Qtr
181,480
109,254
290,734
89.3
Sept. Qtr
164,649
106,096
270,745
89.9
Dec. Qtr
178,224
121,698
299,922
91.0
1974March Qtr
156,562
148,720
305,282
90.6
June Qtr
224,672
120,274
344,946
92.0
Sept. Qtr
186,417
115,874
302,291
91.5
Dec. Qtr
201,999
130,510
332,509
90.5
1975March Qtr
200,064
155,921
355,985
91.0
June Qtr
236,410
114,253
350,663
92.1
Sept. Qtr
227,870
109,698
337,568
92.7
Dec. Qtr
247,471
126,582
374,053
93.2
1976March Qtr
215,493
162,452
377,945
92.5
June Qtr
260,456
112,329
372,785
92.7
Sept. Qtr
243,492
110,036
353,528
93.0
Dec. Qtr
254,358
127,651
382,009
92.9
1977March Qtr
205,116
167,083
372,199
92.6
June Qtr
262,693
120,486
383,179
93.3
Sept. Qtr
250,701
116,523
367,224
93.8
Dec. Qtr
252,743
136,851
389,594
94.1
1978March Qtr
207,089
184,957
392,046
93.4
June Qtr
294,399
125,920
420,319
94.2



C. NET SHORT-TERM MOVEMENTS

Period
Australian Residents
Visitors
Total

1971March Qtr
23043
-18889
4154
June Qtr
-33130
-10939
-44069
Sept. Qtr
20213
-4303
15910
Dec. Qtr
-11445
16502
5057
1972March Qtr
26622
-19980
6642
June Qtr
-45739
-15865
-61604
Sept. Qtr
19818
-2875
16943
Dec. Qtr
-14258
23803
9545
1973March Qtr
39334
-25929
13405
June Qtr
-54735
-13717
-68452
Sept. Qtr
16790
-1103
15687
Dec. Qtr
-18688
30972
12284
1974March Qtr
48859
-13237
35622
June Qtr
-65716
-7566
-73282
Sept. Qtr
24379
3740
28119
Dec. Qtr
-24954
34368
9414
1975March Qtr
43652
-21200
22452
June Qtr
-52036
-13916
-65952
Sept. Qtr
16190
5664
21854
Dec. Qtr
-39012
39021
9
1976March Qtr
66186
-18730
741456
June Qtr
-62929
-7360
-70289
Sept. Qtr
28897
5272
34169
Dec. Qtr
-37688
40218
2530
1977March Qtr
78268
-19444
58824
June Qtr
-71406
-10907
-82313
Sept. Qtr
29943
4544
34487
Dec. Qtr
-34381
48145
13764
1978March Qtr
66574
-28198
38376
June Qtr
-78044
-7860
-85904



APPENDIX III: UNDER-ENUMERATION AT THE 1976 CENSUS - PERSON COVERAGE CHECK OF THE POST ENUMERATION SURVEY

The Person Coverage Cheek (PCC) was conducted as part of the Post Enumeration Survey. This survey was run 2 weeks after Census night and was designed to produce estimates of net under-enumeration of persons. The PCC was a 0.67% sample of private dwellings across Australia.

Persons living in non private dwellings (eg hotels, motels, hospitals) and sparsely settled areas were excluded from the post enumeration survey because of operational difficulties in conducting follow up interviews. However, these amount only to about 5 per cent of the population and hence any underenumeration of them is unlikely to have a significant effect on the overall level of underenumeration.

The post enumeration survey sought only a limited amount of information from sample households, ie number of persons by sex, age, marital status, country of birth and employment status. The results relating to total number of persons are subject to a standard error of approximately .04 per cent at the Australian level, less than .1 per cent at the State level, and less than .5 per cent in the ACT and the NT.

450 trained interviewers approached the randomly selected houses to determine the number of persons staying at each household on Census night. The estimated net under-enumeration was derived by comparing the number of people enumerated on Census night as stated on the Census schedule with the number of people living at that household on census night as enumerated by the interviewer.

While every effort is made to minimise under enumeration in the Census, some inevitably remains for various reasons:

i inadvertent omission of very young children;

ii persons missed because the dwelling was missed by the collector (out-of-date maps, ill-defined LGA boundaries, etc.);

iii treatment by the collector of an occupied dwelling as unoccupied;

vi persons in occupied households not wishing to be included on the census household schedule.


RESULTS FROM THE PERSON COVERAGE CHECK

Overall the PCC revealed that 2.7% of people were missed in the 1976 Census.

Some general conclusions can be drawn from the analysis of the PCC file.

1. The under-enumeration rate for males was higher than for females (3.0 versus 2.4%)

2. The under-enumeration rate varies between different age groups and ranges from 1.7%o for the 10-14 year age group to 4.7% for the 20 -24 year age group. In fact, the rate peaked for the 20 year old age group with 6.2%.

3. Divorced people (5.3%) (closely followed by people who are permanently separated) had a higher underenumeration rate than married people (2.1%) .

4. Persons born outside Australia except the United Kingdom had a higher under-enumeration rate (3.2%) than for persons born in Australia.

5. Unemployed persons had an underenumeration rate of 6.4% compared with 3.0% for employed persons.


CENSUS UNDERENUMERATION AS SHOWN BY THE POST-ENUMERATION SURVEY, BY AGE-GROUP NAD SEX, 1976

Age (years)
Males
Females
Persons

0-4
2.91
3.08
2.99
5-9
1.97
1.73
1.85
10-14
1.82
1.54
1.68
15-19
3.64
3.21
3.43
20-24
5.55
3.90
4.73
25-29
3.88
2.55
3.23
30-34
3.63
1.43
2.56
35-39
2.90
1.70
2.31
40-44
2.51
1.82
2.18
45-49
2.57
2.03
2.30
50-54
3.22
2.67
2.94
55-59
2.87
2.43
2.65
60-64
1.61
2.19
1.91
65-69
1.99
2.47
2.25
70 and over
2.82
2.73
2.76
All ages
3.03
2.40
2.71



ESTIMATION OF POPULATION IN GREAT BRITAIN, UNITED STATES AND CANADA


INTRODUCTION

Estimates of population for all three countries are prepared on a partially residential basis. However, there are problems in obtaining details of residents temporarily outside the country at the time of the Census. All the countries take their Censuses at dates other than the mid-point of the year, so the census population has to be brought forward to 30 June for a base population. None of them adjust the Census population for under-enumeration.

Lack of adequate overseas migration statistics in these countries necessitates indirect estimation of this component of population growth from survey and other data. The conceptual definition adopted for the overseas migration component is of a permanent and long-term nature.

I. GREAT BRITAIN

Estimates are prepared separately for England, Wales, and Scotland and for the following combination:

i England and Wales

ii Great Britain (England, Wales and Scotland)

Census Base

The population census of Great Britain is based on a de jure concept; it includes all persons enumerated in Great Britain on Census night and also usual residents temporarily outside of Great Britain. Census populations are tabulated on different bases. Some household composition tables include all usual residents of an area, irrespective of their actual location on census night. Usual residence tables however, do not include persons usually resident in Great Britain who were not in England, Wales or Scotland on census night.

Population Estimates

For England and Wales, 3 different estimates are prepared.

1 The home population which is the population resident in England and Wales as defined in the census.

2 The total population which is the home population plus members of H.M. Forces belonging to England and Wales serving outside England and Wales minus the forces of other countries stationed in England and Wales.

3 The civilian population is the total population minus all members of H.M. Forces, whether or not in England and Wales.

Intercensal estimates are made using the conventional method of adding natural increase and net migration onto a census population base. Common series of natural increase and net migration are used to compile estimates on all 3 different bases.

International migration: The Census is updated for international migration using more or less permanent and long-term data obtained from the International Passenger Survey (IPS). The definition of an immigrant used in the IPS is a person who, having lived abroad for at least twelve months, intends to reside in the United Kingdom for at least twelve months, and vice-versa for an emigrant. Movement to and from the Irish Republic is not covered by the survey, but alternative estimates of migration are used for population estimates. The resulting component is referred to as net civilian migration.

Natural Increase: This is derived from registration of births and deaths.

Other adjustments: To produce estimates of "total" and "civilian" population, adjustments are made for changes in the number of armed forces in the area.

Reference Sources

Office of Population Censuses and Surveys, Census 1971, England and Wales, Usual Residence Tables, General Explanatory Notes.

Office of Population Censuses and Surveys, Series PP1 No 2, Population Estimates 1975 (revised) 1976 (provisional), England and Wales.

CSO Social Trends 9, 1979 edition, page 38.

Office of Population Censuses and Surveys, Series MN No 2, International Migration.

Office of Population Censuses and Surveys, Population Trends 14, Winter 1978, page 32.


II. UNITED STATES

Population estimates on 4 different bases are made for each of the 50 States and the District of Columbia. The Commonwealth of Puerto Rico and US Territories in the Pacific are excluded.

1970 Census Base

The US Census is a de jure census, ie people are counted at their usual place of residence, where they live and sleep most of the time. People without a usual place of residence were counted where they were enumerated.

Special rules were adopted in the treatment of the armed forces, naval personnel and merchant crews on ships, and in the treatment of persons in institutions.

Persons who were overseas for an extended period (in the Armed Forces, working at civilian jobs, studying in foreign universities, etc) are not included, but special efforts were made to count persons temporarily abroad on vacations, business trips, and the like, and to include them in their usual place of residence in the US. These efforts included obtaining information about them from members of their households. Americans leaving the United States for temporary travel abroad in the month before the Census were asked to complete a form if they indicated there would be no one at home to report them in the Census. A matching process was used to avoid duplication.

Citizens of foreign countries having their usual residence in the United States, not living at an embassy etc, who were working, attending school, etc, and their families were included. Citizens of foreign countries visiting or travelling in the United States or living on the premises of an embassy, ministry, legation, chancellory or consulate were not enumerated.

Population Estimates

Population estimates on four different bases are prepared.

1 Resident population: conforms to the Census, and includes residents of the 50 States and the District of Columbia.

2 Civilian population: resident population less the Armed Forces (both foreign and US) stationed in the United States.

3 Total population including armed forces overseas: includes the resident population and the US Armed Forces stationed in foreign countries and the outlying areas, but not their dependents.

4 Total population, including armed forces and government employees overseas and their dependents: This population includes the following US civilians overseas in addition to the Armed Forces stationed overseas.

a dependents of members of the Armed Forces

b civilian citizen employees of the Federal Government and their dependents.

Components of Intercensal Population Change

i International Migration

This component is derived from four sources.

a Alien immigration is based on published statistics of the number of aliens entering who have had their residence status in US changed from temporary to permanent. Separate estimates for Cuban and Vietnamese refugees are made.

b Net arrivals from Puerto Rico (supplied by the Planning Board of the Commonwealth of Puerto Rico).

c Net immigration of civilian citizens affiliated with the US Government was derived from data on births in military hospitals overseas and changes in the numbers of civilian citizen employees of the Federal Government overseas and of citizen dependents of Federal employees and servicemen.

d Emigration estimates are based on information supplied by the Social Security Administration, and immigration data presented in statistical publications of various countries. The estimates are knows to be understated because statistics on permanent arrivals from the United States are not available for many countries, and because little information is available on alien emigration.

ii Natural Increase

The births component used for population estimates is based on births to the resident population. Prior to 1970, they were adjusted for under-registration in accordance with the results of the occasional "Birth Registration Tests". This adjustment was discontinued after the 1970 Census, because it was estimated that the under-registration had become negligible (over 99 per cent complete). The same figure for births is used for all series of population estimates.

The deaths component includes deaths to Armed Forces overseas. Deaths of members of the Armed Forces overseas are obtained from the Department of Defence and other sources. Under-registration of deaths has not been considered significant. No adjustments have been made since 1960 to allow for under-registration.

Reference Sources

US Bureau of the Census, Number of Inhabitants, United States Summary, PC(1)-A1.

US Bureau of the Census, current population reports. Population estimates and Projections, Series P-25, Nos 519, 545, 706.


III. CANADA

Census Base

The census counts all persons whose usual place of residence was somewhere in Canada, including Canadian tourists and businessmen temporarily absent, government employees stationed abroad and their families, and crews of Canadian merchant vessels. Not included are government representatives of other countries (and their families) attached to the legation, embassy or other diplomatic body of that country, members of the Armed Forces of other countries stationed in Canada and members of their families who are not citizens of Canada, students attending school in Canada whose usual residence is outside Canada, and residents of another country visiting in Canada temporarily.

Components of Population Estimates

i International Migration

This is defined as movement between Canada and a foreign country which involves a change of residence. Data on the number of immigrants are supplied by the Department of Employment and Immigration. Estimates of the number of emigrants are obtained indirectly by using information from three sources:

a Reports containing data on the number of immigrants to the United States, whose country of last residence was Canada.

b Estimates of the number of persons entering the United Kingdom to reside, whose country of last residence was Canada; these estimates are taken from the International Passenger Survey of the UK.

c Estimates of emigrants to countries other than the United States and the United Kingdom are made on the basis of the 1971 and 1976 Canadian Census data, vital statistics, immigration figures and information provided by the United States and the United Kingdom.

The only data available on the age distribution of emigrants to obtain age distributions of intercensal estimates of the population are five-year age groups for emigrants to the United States. This is converted to single years of age by Sprague multipliers, and the resulting age distribution is applied to the total estimate for all emigrants from Canada.

ii Natural Increase

Natural increase is derived from registration of births and deaths.

Reference Sources

Statistics Canada, Quarterly Estimates of Population for Canada and the Provinces, July 1978.