1.1 Introduction
A census of population and dwellings is invariably the largest data gathering exercise in any country. In principle, it requires counting everybody in the country on census night. The census yields a wealth of valuable information for analysing changes in the socio-demographic profile of the population, and for monitoring, planning, and decision-making at the national and local level, by government, business and the general community. It is also integral to the derivation of reliable post-censal population estimates and for charting future demographic trends.
Given the strategic significance of the census data and its diverse applications, Statistics New Zealand, like other national statistical organisations, makes concerted efforts to ensure the universal coverage of its five-yearly census. Yet, the censuses everywhere tend to miss some people. In their article titled Census Undercount: The International Experience, Zitter and McArthur (1980) noted:
"A review of the practises of most of the developed countries of the world relative to census undercount suggests that the United States does not have a monopoly in its concern for complete coverage of the population in conducting its censuses and in measuring and evaluating their accuracy."
The incomplete coverage may result from, for example, inadvertent omission of young children, difficulty in enumerating people on the move and those living in apartments, as well as people not willing to cooperate with census enumerators.
There are a myriad of statistical procedures that demographers and others use to check the accuracy of census coverage (Shryock and Siegel, 1973). These include: (a) checks against demographically derived estimates, (b) comparison of census figures with administrative records and other sources, and (c) a post-enumeration survey (PES).
In principle, a post-enumeration survey is a special survey undertaken shortly after the census to evaluate the completeness of census coverage. It involves an independent re-enumeration of a statistically designed sample of all dwellings covered by the national census. Post-enumeration surveys are an essential feature of census-taking in many countries.
In New Zealand, the first post-enumeration survey was held after the 1996 Census of Population and Dwellings (Statistics New Zealand, 1998a). The second survey was held after the 2001 Census. Like the first survey, its expressed objective was to measure the level of census coverage, ie the extent of undercount/overcount of New Zealand’s population.
This report describes and discusses the salient features of the 2001 Post-enumeration Survey, including its scope, methodology, the information gathered and the key results. Discussion of the survey results also covers implications for post-censal population estimates and other demographic series.
1.2 Organisation of the report
The report is organised into four chapters and two appendices. The highlights, which precede the main body of the report, summarise the survey results. This chapter, Chapter 1, outlines the significance of the Post-enumeration Survey, its objective, its brief history, as well as the known sources of census miscount. Chapter 2 presents the technical details on the survey methodology, including the survey population, sample design, data collection procedures and estimation method, as well as sampling and non-sampling errors. The main findings are described and discussed in Chapter 3. The results include an analysis of age, sex, ethnic and broad geographical differentials, and a comparison of New Zealand's undercount results with the latest available figures for Australia, Canada, the United Kingdom and the United States.
Chapter 4 summarises and discusses the 2001 PES results, and also includes an outline on the adjustment of the base population for deriving post-censal population estimates and projections. The PES results are a vital element in this statistical process. Appendix 1 includes the Post-enumeration Survey form, while further technical details such as dwelling weights, person weights and non-response adjustment are set out in Appendix 2.
1.3 Miscount and its sources
In such a large and complex exercise as a census, it is inevitable that some people and dwellings will be missed. Reasons for people being missed are many and varied and include:
- failure of census field procedures
- dwellings entirely missed by enumerators
- occupied dwellings misclassified as vacant
- people deliberately avoiding the census – refusing or unwilling to respond (for fear that information given will be used against their interests)
- people being reluctant to open their door to strangers
- people shifting from one house to another around the time of the census
- multiple households living at the same address
- people being away temporarily (eg work, school)
- people having no usual residence (eg transients, street kids)
- newborn babies being overlooked.
On the other hand, a person may be counted more than once. Such ‘double counting’ may involve:
- students living away at school or university (and also being counted at the home of their parents)
- children under joint custody
- people living away from home while working
- people shifting from one house to another around the time of the census
- people living in institutions
- people with dual residences
- erroneous enumeration of deceased persons, babies born after census night, emigrants, etc
- vacant dwellings classified as occupied (non-respondent) dwellings, leading to the creation of substitute forms.
It should be noted that an error in recording the correct geographic location of a person does not constitute a coverage error. For example, a person who is enumerated in the wrong area does not represent overcoverage for the area in which the person was enumerated nor undercoverage for the area in which the person should have been enumerated.
1.4 Why measure the undercount?
In New Zealand there is growing interest in accurately determining the level of coverage of the census. This helps improve both census processes and the general quality of post-censal population estimates. Population estimates are used for a variety of purposes:
- allocation of funds to organisations using population-based weightings
- denominators for the calculation of rates (eg birth and death rates) and per capita time series
- determination of population weights for various surveys
- administration, policy-making and planning, by both central and local government
- demographic, social and economic studies.
The experience of other countries is that those groups who are mobile or difficult to enumerate are most likely to be missed in the census. In New Zealand, indirect evidence drawn from demographic analysis, birth registrations and school enrolments suggests that the census does miss some people and that this under-enumeration varies across different groups.
1.5 History of the Post-enumeration Survey
Before the 1996 Census, Statistics New Zealand had evaluated certain aspects of the general quality of census data but had not attempted to measure the level of undercount or overcount directly, using a post-enumeration survey. In 1990, a pilot test as a preliminary to a PES was conducted, but the survey did not go ahead. In 1994, appropriate funding was approved by Government to allow Statistics New Zealand to undertake a PES in conjunction with the 1996 Census. The 1996 PES was the first survey of its type in New Zealand. The 1996 Census of Population and Dwellings was held on 5 March 1996 and the PES was conducted during 18–31 March, starting after the collection of the census questionnaires had been completed to avoid overlap of census enumerators and PES interviewers in the field (for further details, see Statistics New Zealand, 1998a). The second PES was held in March 2001, shortly after the 2001 Census of Population and Dwellings.