Volume 4

Contents

1. Who are we? The conceptualisation and expression of ethnicity
2. Regression Analysis of Probability-Linked Data
3. Sampling for Subpopulations in Household Surveys with Application to Māori and Pacific Sampling
4. Improving Regional Level Estimates from National Surveys using Census and Administrative Data: A case study using the New Zealand Health Survey

1. Who are we? The conceptualisation and expression of ethnicity (PDF, 609KB)
Paul Callister, Robert Didham, & Anna Kivi

Abstract: Drawing on a range of national and international literature, this scoping paper explores some aspects of ‘who are we’ within a New Zealand context. Throughout the world the conceptualisation of ethnicity, or in some countries race, is undergoing significant change. There are many drivers including high rates of international migration, a vibrant indigenous rights movement in some countries, growing incidence and recognition of ethnic intermarriage, a concern about racism and discrimination, and, connected with this, a debate around how to support and, in some situations, integrate increasingly diverse populations into the wider community. Against this background, the paper draws together some ideas in the following areas: ethnogenesis; the official construction of ethnicity in New Zealand; ethnic intermarriage, and related to that, the transmission of ethnicity to children and multiple ethnicity; ethnic mobility; indigeneity; genetics, the human genome project and ethnicity; and the recent growth of New Zealander responses in the New Zealand census. It is not possible to draw many strong conclusions from the burgeoning national and international literature except there is a very vigorous and complex debate taking place about identity in most countries around the world. While there are some commonalities in the international debates, many local factors affect country specific discussions. However, some of the very broad threads that can be drawn from the literature include: while it is clear that for some people, and in some situations, ethnicity is a critical and daily part of their identity, for others ethnicity can be a minor part of identity or, in particular contexts, ethnicity might have little meaning; that identities are always emerging; that ethnicity is multi dimensional and that, in most countries, there is some fluidity of ethnicity; that identity is not only expressed but is perceived and observed and this perception of others can forge, reinforce and, at times, restrict identities; that the division of the world population, or the population of specific countries, into neat, non-overlapping groups is increasingly problematic; and that official definitions of ethnicity officially may not match all that closely ideas of ‘who we are’ personally. All of this suggests that, from a New Zealand policy perspective, while ethnicity can be an important analytical variable, much care is needed in its use.

Keywords: Identity, ethnicity, ethnogenesis, ethnic mobility, genetics.

2. Regression Analysis of Probability-Linked Data (PDF, 815KB)
Ray Chambers

Abstract: Data obtained after probability linkage of administrative registers will typically include errors due to the fact that some linked records actually contain data items are sourced from different individuals. Such errors can induce bias in standard statistical analyses if ignored. In this report we describe some approaches to eliminating this bias in the case of linear regression analysis and, more generally when inference is based on an estimating equation, with an emphasis on logistic regression. Simulation results that illustrate the gains from allowing for linkage error in linear and logistic regression analysis are presented, as are extensions of the approach to situations where a sample is linked to a register and to where the linked registers are of unequal size.

Keywords: Record matching, linkage errors, linear regression, logistic regression, estimating equations, measurement error.

3. Sampling for Subpopulations in Household Surveys with Application to Māori and Pacific Sampling (PDF, 4.7MB)
Robert Clark, Mike Doherty, Angela Forbes, & Robert Templeton

Abstract: Many New Zealand national household surveys have a requirement to produce statistics with adequate precision both for the whole of New Zealand and for important subpopulations, particularly the Māori and Pacific populations. General population surveys which make no special provision for these groups would generally not achieve sufficient precision for Māori and Pacific estimates when these are of particular interest. Kalton and Anderson (1986) described a range of strategies for sampling subpopulations, where the aim is only to produce statistics about the subpopulation and not national statistics. The purpose of this Official Statistics Research fund project is to extend the methods of Kalton and Anderson (1986) to better incorporate some of the complexities of household surveys, with particular reference to sampling the Māori and Pacific populations in New Zealand.

Keywords: Māori, Pacific, sample, subpopulations, screening, sampling.

4. Improving Regional Level Estimates from National Surveys using Census and Administrative Data: A case study using the New Zealand Health Survey (PDF, 418KB)
Robert Templeton

Abstract: The project investigated small area methods and their use in improving national survey estimates at a regional level for the New Zealand Health Survey. It had four specific aims:

• ensure methods and results are explained in lay terms as much as possible
• be able to incorporate data from administrative sources
• be able to produce measures of quality of the resulting estimates
• implement the methods in SAS.

In this case we illustrate the use of a small area estimator in the New Zealand Health Survey to produce estimates at the District Health Board (DHB) level. A statistical model (namely a generalised linear mixed model) can be used to produce the DHB estimates and assess their quality. A set of small area diagnostics are also examined to help assess the validity of the results. The modelling and diagnostic tools are those developed by Ray Chambers and other authors, who in various papers (referenced) illustrate the methods as applied to the production of small area statistics from the UK Labour Force Survey. The procedures used in this case study are implemented in SAS programs.

Keywords: Small area estimation, composite estimation, random effects, generalised linear mixed models.