3132.0 - Demography Working Paper 2002/2 - Estimated Resident Population and Effects of Census Systems Created Records, 2002  
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 27/11/2002   
   Page tools: Print Print Page Print all pages in this productPrint All

Introduction

1. This paper outlines initial findings of a review of processes used in determining preliminary Estimated Resident Population (ERP) for States, Territories and Australia for 30 June 2001 using 2001 Census results. Background information is provided on some 2001 Census processes and System Created Records for non-contact dwellings. This paper also lists release plans for final ERP based on the results of the 2001 Census.


System Created Records

2. System Created Records are created during census processing for people for whom a census form has not been received and the collector believes that they have been missed from the Census count.

3. For private dwellings, most System Created Records (SCRs) are created where the collector has not been able to make contact with the household, yet believes that the dwelling was occupied on Census Night. Smaller numbers of System Created Records are due to situations where people indicate a desire to mail back a census form but do not do so, and where people refuse to complete a census form. The term 'non-contact' dwelling is used in this paper to refer to all these situations. The number of person records created for the non-contact private dwellings is based on the average number of people in private dwellings for that collection district.


4. In non-private dwellings (such as hotels, motels, etc), System Created Records are created where a person is listed on the dwelling summary sheet as being present on Census Night, but no census personal form has been received.

5. Systems Created Records have values imputed for age, sex, marital status and usual residence. Values for other variables are set to not stated or not applicable, depending on the imputed value for age. Almost 95% of System Created Records are created in non-contact dwellings. Compared with 1996, the proportion of 'non-contact' dwellings more than doubled from around 0.9% of all dwellings in 1996, to 2.0% in 2001.

TABLE 1 System Created Records (SCRs): 2001 and 1996 Censuses

2001
1996
Increase in SCRs, 1996 to 2001


State/Territory
Total Persons
SCRs
% of SCRs
Total Persons
SCRs
% of SCRs
percentage point increase

NSW
6,371,745
173,878
2.7
6,038,696
88,961
1.5
1.3
Vic.
4,644,950
97,618
2.1
4,373,520
57,388
1.3
0.8
QLD
3,655,139
63,340
1.7
3,368,850
44,033
1.3
0.4
SA
1,467,261
17,959
1.2
1,427,936
14,487
1.0
0.2
WA
1,851,252
36,551
2.0
1,726,095
24,909
1.4
0.5
Tas.
456,652
5,983
1.3
459,659
5,007
1.1
0.2
NT
210,664
7,869
3.7
195,101
7,383
3.8
0.0
ACT
311,947
6,608
2.1
299,243
3,954
1.3
0.8
Other Territories
2,740
17
0.6
3,323
70
2.1
-1.5
TOTAL
18,972,350
409,823
2.2
17,892,423
246,192
1.4
0.8



Reviewing System Created Records - utilising information from improvements in field work

6. The ABS publication How Australia Takes a Census (cat. no. 2903.0) describes 2001 Census field operations and administration. Improvements in field procedures introduced for the 2001 Census resulted in basic demographic information being collected from around 8% of non-contact private dwellings. In such circumstances, a resident of the dwelling has provided the census collector when delivering the form with a count of the number of males and the number of females who reside in the dwellings.

7. Using this information a review of procedures for measuring Census net undercount conducted in September 2002 has shown that on average the number of persons in non-contact dwellings is substantially lower than in other private dwellings. A consequence of this finding is that the practice of creating person records using the average number of people in private dwellings tends to over-estimate the number of people living in those dwellings.

8. Preliminary analysis has shown that the creation of persons records for non-contact private dwellings has resulted in a Processing Induced Overcount of person records of between 50,000 and 100,000 persons.


Further analysis for Processing Induced Overcount

9. Before a revision can be made to Estimated Resident Population to account for the extra persons imputed, further analyses of System Created Records need to be completed. Average household size is known to vary across regions and between dwelling structure types. Using characteristics of dwellings, and information regarding social and economic characteristics of geographic regions, it is possible to assess the impact of Processing Induced Overcount at the state of usual residence by sex by capital city/balance of state level. When this has been determined it can be incorporated with other final components in determining ERP for 30 June 2001.


Finalising September 1996 through 30 June 2001 Estimated Resident Population

10. Following each Census the ABS finalises population estimates for 30 June of the Census year back to 30 September of the year of the previous Census. This process requires:

final Census counts on a usual residence basis;
final data on Census net undercount;
final data on Residents Temporarily Overseas on Census night;
final births data (on a date of occurrence basis) for the intercensal period;
final deaths data (on a date of occurrence basis) for the intercensal period;
final interstate migration data for the intercensal period;
final net permanent and long term overseas migration data for the intercensal period; and
final category jumping data for the intercensal period.

11. For the 30 June 2001 ERP, a final measure of Processing Induced Overcount, as a consequence of imputing person records from non-contact dwellings, needs to be derived at the five year age group by sex by state by part of state level.

12. Additionally, in determining final 30 June ERP for the Census year, population estimates must be "backdated" from Census night to 30 June. For the 2001 cycle this means backdating ERP from 7 August 2001 to 30 June 2001, a period of 38 days (or around 10% of a year). This backdating process requires births, deaths, net interstate migration data, net permanent and long term overseas migration data, and category jumping data. To measure category jumping, passenger card data from the Department of Immigration and Multicultural and Indigenous Affairs (DIMIA) are needed for all movement months up to and including September 2002.

13. The ABS expects to complete the time series of passenger card data in late December 2002 with the release of April - June 2001 supplementary edition of Overseas Arrivals and Departures (cat. no. 3401.0).

14. Only when all these components are finalised will it be possible to determine the 2001 Census based estimated resident population for 30 June 2001. At this stage it will also be possible to measure the National and State intercensal discrepancy for the 1996-2001 period.


Release of final ERP September 1996 through 30 June 2001

15. National and state quarterly population estimates for the period 30 June 2001 through 30 September 2002, using final data from the 2001 Census, will be released in a new electronic publication in late February 2003.

16. The September 2002 edition of Australian Demographic Statistics (cat. no. 3101.0) will be published on 20 March 2003. This publication will contain final national and state ERP for the intercensal period September 1996 through June 2001, and preliminary ERP (using the 2001 Census as a base) for the quarters September 2001 through September 2002.

17. Final single year of age ERP at the State level for 30 June 1997 through 30 June 2001, and preliminary data for 30 June 2002, will be published in March 2003 in the publication, Population by Age and Sex (cat. no. 3201.0).

18. Details of final Census net undercount, with a discussion of gross undercount and gross overcount, including Processing Induced Overcount and the impact of System Created Records, will be published in Information paper : 2001 Census Data Quality - Undercount (cat. no. 2940.0) on 1 April 2003.



Demography Section
Australian Bureau of Statistics
27 November 2002


Fact Sheet: Effect of Census Processes on Non-response Rates and Person Counts