Stats NZ has a new website.

For new releases go to

www.stats.govt.nz

As we transition to our new site, you'll still find some Stats NZ information here on this archive site.

  • Share this page to Facebook
  • Share this page to Twitter
  • Share this page to Google+
Introduction

The New Zealand Census of Population and Dwellings is currently held every five years as legislated by the Statistics Act 1975. The primary role of the census is to provide population and dwelling counts for New Zealand, for regions and territorial authorities, and smaller geographic areas such as area units and meshblocks. The census is also the only comprehensive source of information about the social and economic characteristics of local communities and small population groups (eg Māori and iwi, youth, new migrants).

Population statistics are the most important requirement for a census to provide. Ensuring these statistics are fit for purpose is critical when considering future census models.

At the same time, achieving high quality in the population counts is a major driver of the costs of a census. The feasibility of obtaining sufficiently good population estimates, and this trade-off between quality and cost for population statistics will be key determinants in decisions about the long-term direction of census.

McNally and Bycroft (2015) developed quality standards for population estimates that reflect customer requirements for accuracy. These quality standards were designed to assess the statistical quality of population estimates produced by alternative approaches to census-taking, such as administrative-based models that use linked administrative data to produce population statistics.

Nordic countries and others that already produce their census information from administrative sources base population statistics on national population registers. The population registers serve administrative purposes and are designed to include everyone living in the country. There is typically a unique identifier assigned to each person which is widely used. These ‘ready-made’ administrative systems do not exist in New Zealand. Bycroft 2015 contrasts the administrative data available in New Zealand with that typically found in countries that have moved to register-based censuses.

Gibb and Shrosbree (2014) developed a method for constructing a statistical population list from the linked administrative data sources available in Statistics NZ’s Integrated Data Infrastructure (IDI) at April 2013. Population estimates were then derived and compared with official estimated resident population figures for June 2010. While clear limitations were identified in the administrative sources available at the time of the study, the results showed enough promise to continue with further investigations.

Development of the IDI in the two years to May 2015 overcomes some of the earlier limitations. Birth and death registrations and health data are now available. An extended spine structure in the IDI provides better coverage of the population not paying tax, especially children. Health data provides a much broader source of activity information than before – the previous method had relied solely on payment of tax and enrolment in education.

Gibb and Shrosbree’s investigation was undertaken before the 2013 Census, and the administrative data were only available up to 2010. We are now able to compare results against the 2013 Census and the official 2013 estimated resident population, and administrative sources have been updated to include information at least until 2013.

Aims and scope

The major aim of this paper is to examine the accuracy of national population estimates, produced from linked administrative data available in the IDI. We update the methods applied in 2014 in light of the recent changes to the IDI. The accuracy of the population estimates will be evaluated against the official 2013 estimated resident population using the quality standards developed by McNally and Bycroft.

The IDI is continually being expanded and developed. The analyses in this paper are based on the IDI as it stood in May 2015.

The scope of this paper is population estimates at the national level by age and sex. Statistics NZ also produces official population estimates for subnational geographies and for ethnicity. Subnational estimates are dependent firstly on the quality of national estimates, and secondly on information about where people live. The quality of geographic information in the IDI is examined in Gibb, 2015. The quality of ethnicity information in the IDI is examined in Reid et al, 2016.

The potential for administrative data sources to produce other types of census information (for example, information about education, income, families, households, or housing) is discussed in other work (O’Byrne et al, 2014; Shrosbree, 2015; Suei, in press).

This work is not intended to provide a final evaluation of the feasibility of using linked administrative data sources to produce population statistics in the absence of a full-enumeration census. Rather, we will provide information about the use of linked data sources to identify populations that will guide further work.

The remainder of this paper is organised as follows. Section 3 describes the data sources used: the census and official population estimates, the IDI, and the linked Census-IDI dataset. Section 4 describes the method used to construct an 'administrative' resident population from the IDI, and how we assess its accuracy. Section 5 provides results with comparisons at both the aggregate level and individual level. The paper concludes with a short discussion.

  • Share this page to Facebook
  • Share this page to Twitter
  • Share this page to Google+
Top
  • Share this page to Facebook
  • Share this page to Twitter
  • Share this page to Google+