Official Statistics Research Abstracts 2008/2009

Official Statistics Research funded 4 projects in the 2008/09 year.

Geovisualisation and policy: exploring the links | Sexual Orientation Data in Probability Surveys | More for less? Using statistical modelling to combine existing data sources to produce sounder, more detailed, and less expensive Official Statistics | Statistics for Sustainable Development Strengthening New Zealand's SDI Framework

OS Research will be inviting the public to scheduled Officials Statistics System Seminars throughout 2008 and 2009, as well as coordinating expert workshops. In the workshops, researchers will come together with government departments and help them apply the research findings to their data. It is anticipated that all of the research and reports that come from the projects will be available after successful review in the Official Statistics Research Series.

For any seminar or expert workshop inquiries, please contact OS Research at: osresearch@stats.govt.nz.

1. Geovisualisation and policy: exploring the links

Project summary

Recent developments in the geospatial arena in New Zealand (such as the creation of a national geospatial strategy and geospatial office), and globally (e.g. the development and widespread use of online mapping tools such as Google Maps), have highlighted the geospatial nature of data.  Users of statistical datasets in particular can benefit greatly by spatially visualising and exploring their data.

Much of the data (e.g. census data) collected by Statistics New Zealand is intrinsically spatial.  Each unit (individual, household or meshblock level) is recorded at a fixed spatial location or for a given region.  An example is the Census of Population and Dwellings which we use as an illustrative data set in this project.  Currently much of the census output is available to users via Table Builder—a menu driven method of constructing cross tabulations based on selected variables (including areas) at different scales.  Table Builder does not allow visualisation of data in map form.

What we are proposing here are illustrations of the capability of map and associated geovisualisation and analysis tools.  The aim is to demonstrate to OSS analysts how geovisualisation can expand the way in which existing data can be used by representing values over geographic domains.  We will also demonstrate to stakeholders the way in which statistical analysis can be productively interfaced with mapping to quantify the magnitude of particular spatial configurations, to measure changes in the distribution of phenomena and quantify notions such as concentration, dispersion and segregation.

The four illustrations we propose will be tailored to address the three audience tier model already used by SNZ:  technical, professional and public.  We wish not simply to illustrate the capabilities of geospatial representation but to understand and assess the ways in which the various outputs possible are understood or comprehended by different categories of users and the implications this understanding has for both the data supplied by SNZ and the way it is configured for users.  Issues associated with data and its configuration for geovisual exploration will be covered in each of the four case studies.

Research aims

1. To illustrate the potential of geovisualisation to help explore, analyse and communicate important patterns and trends in statistical data, to a range of existing and potential users—for research, education, and government.  The emphasis of this project is to expose users of statistical data to new ways to visualise, explore and interpret selected locationally-based data.
2. To show the range of possible applications via four policy relevant case studies for which data is presently available.  Existing geovisualisation tools from the GeoVISTA Center (www.geovista.psu.edu) will be repurposed for these tasks, and will be employed to help answer the policy and research questions described above.  The visualisation applications created—and related findings—will be made available to SNZ and others, to create opportunity for researchers to ask additional, deeper questions of the phenomena.
3. The three audiences—technical, professional and public—do not ask the same questions of the same data set.  We will mimic these audiences and illustrate the different ways in which geovisualisation can enhance the understanding of all three groups with respect to each of the four substantive areas of enquiry.  We will seek to capture, measure and assess the different levels of reception and understanding which geovisualisation generates within these three audiences.

Project Sponsor: Statistics NZ
Project team: Victoria University of Wellington, University of Auckland, and Statistics NZ

Back to top

2. Sexual Orientation Data in Probability Surveys

Project summary

Sexual orientation is a topic of interest for public policy nationally and internationally. There is growing evidence of disparities in outcomes between heterosexual and gay, lesbian and bisexual (GLB) populations in New Zealand. Currently there is very limited data collected on these populations in the official statistics system (OSS), no reliable denominator data to validate samples in surveys and no overriding conceptual or methodological framework that guides the OSS in this area.
This project will:

• formulate a conceptual framework for sexual orientation that supports statistical measurement and data analysis
• review New Zealand and international research to consider which methodological issues impact on the collection of robust data on GLB populations in probability surveys
• consider these issues alongside the pragmatic requirements of a New Zealand OSS measurement framework
• develop recommendations on how to address them (in a collaborative process with stakeholders), to enable high quality, consistent and comparable data collection across the OSS.

Given the lack of baseline data, the presumed small size, fluid and hidden nature of the collective GLB population we need to use existing data to estimate the characteristics of these groups. Some official statistics surveys have already collected sexual orientation data. In New Zealand these are Te Rau Hinengaro – the Mental Health Survey 2003/4, the Drug Use Survey 2003 and the Alcohol Use Survey 2004. This project will:

• analyse these data sets, and
• combine survey information to improve estimates of the demographic profile of GLB populations in New Zealand.

It will be informed by overseas models and findings of other surveys. It will take issues such as misreporting and non-response into account. This model can inform the development of sampling strategies for future probability surveys.

Research Aims

To build capability to collect and analyse robust, high quality data on sexual orientation in New Zealand by:

• developing a coherent and theoretically robust conceptual framework for sexual orientation measures in New Zealand
• developing a framework for the robust and effective collection of sexual orientation measures in OS probability surveys
• assessing the capability of existing New Zealand OS to provide reference data for sexual orientation
• developing a model for the estimation of GLB population groups that can provide reference data for public sector information needs and inform the development of sampling strategies for surveys

This will increase the statistical capability of the state sector and a range of other stakeholders, including researchers. It will provide a basis for improving the coverage, reliability and general quality of data available to the OSS and thus wider society, and assist the state sector to develop and implement quality policy that is inclusive of GLB populations. This will ultimately contribute to the social cohesion of New Zealand society and the full participation of all its members in social and economic life. 

Project Sponsor: Ministry of Social Development
Project team: Independent Researcher

Back to top

3. More for less? Using statistical modelling to combine existing data sources to produce sounder, more detailed, and less expensive Official Statistics

Official Statistics has long recognised the potential for leveraging administrative data to achieve more accurate statistics. The use of existing Official Statistics data collections and sound statistical modelling has the potential to save money, provide more accurate statistics at finer levels, and reduce future data collection costs and respondent burden. More recently, the focus has expanded to include the possibilities of combining administrative and/or census data with survey data via statistical modelling, e.g. by using mass imputation, spatial microsimulation and certain types of small area and small domain estimation. These methodological developments have varying levels of theoretical backing, but are nevertheless the first and last are already being used within New Zealand government departments and ministries on what are now extensive and wide-ranging electronic databases containing social, economic, health, epidemiological and other information. A course on the second, spatial

There is increasing general awareness that emphasis needs to move from data collection to better data utilisation, but the statistical theory behind and links between these three methods requires further research, especially for modelling methods that combine data as anonymously as possible to limit confidentiality concerns.

The principal aim of this research project is to explore how best to combine survey and census data, or survey and administrative data, using sound statistical models to produce finer-level statistics for variables collected only by sample survey, without raising such confidentiality issues by formally linking individual records. Essentially, the methods work by imputing or predicting variables of interest that are not collected or are partially missing in the larger dataset, via models that use variables common to the smaller dataset (in which the variables of interest are also present); the fitted model then provides multiple predictions for all missing census observations, which when combined give area or subpopulation estimates together with estimates of accuracy. The statistical techniques considered include mass imputation, spatial microsimulation, and certain types of small area and small domain methods which, while similar in intent, differ in the form of the underlying statistical model used to connect data sources. Details of research methods are given below in the "Research methodology" section. The aim of the research is to compare and contrast the available statistical techniques for producing census-level datasets by modelling, to assess when and why such methods work well, and when they cannot be recommended. Conclusions will be linked to existing Official Statistics projects in a range of government departments and ministries. While anticipated outputs and outcomes include a report and publications in international journals and conference presentations, they will also include a series of brief case studies of research programmes within a number of the government agencies using or producing Official Statistics, and offer ongoing support to viable programmes, although generally without becoming directly involved in the day-to-day aspects of these government projects. The case studies will nevertheless connect the theoretical research with particular situations encountered in the New Zealand Official Statistics System.

Research aims

The principal aim of this research is to establish when and how Official Statistics data sources, particularly surveys and censuses or surveys and administrative sources, can be combined using statistical models based on mass imputation, spatial microsimulation, and small area and domain estimation, to produce cost-effective, accurate, fine-level statistics. The research will consider and further develop a range of statistical methods, and consider their application in principle to set of case studies in sociology, economics, and health from a variety of government departments and ministries.

Project Sponsor: Statistics New Zealand
Project team: Massey University

Back to top

4. Statistics for Sustainable Development: Strengthening New Zealand's SDI Framework

Project summary

This project brings together expertise within Statistics New Zealand and within the AERU Research Unit of Lincoln University to investigate two research questions: how can equity be incorporated into an integrated capital-based approach to sustainable development indicators (SDI); and what insights into a national sustainable development indicators framework can be derived from the way New Zealand includes cultural well-being in its definition of sustainable development and includes cultural indicators in its SDI framework?

The project will begin with a period of detailed research design in collaboration with four officials from participating government agencies other than Statistics New Zealand. The research on equity will involve a systematic review of the New Zealand literature over the last 20 years that addresses the question of in what sense is social equity for the current generation accepted as important for New Zealanders’ perceptions of sustainable development, which will provide the basis for considering potential indicators for equity, adopting an approach developed by the Swiss Federal Statistical Office. The research on cultural well-being will draw on the international literature to explore how cultural capital can be integrated with all the indicators adopted by New Zealand’s SDI framework.

A report on Statistics New Zealand’s Sustainable Development project is scheduled for publication in June 2009. The research project in this proposal will complete two research reports, also in June 2009. Taken together, the publication of three major pieces of work on New Zealand’s sustainable development indicators framework will represent a landmark event for the Official Statistics System, and will contribute to the following outcomes:

1. A stronger conceptual framework underlying Statistics New Zealand’s work on measuring sustainable development, particularly by better understanding of the potential roles of equity and cultural capital in influencing sustainable development.
2. Better alignment of New Zealand indicators of sustainable development at a national and sub-national level with the capital-based indicator framework being recommended as international best practice.

Greater awareness in international policy discussions about the importance of cultural well-being indicators in measuring sustainable development as a result of understanding New Zealand’s unique experience in this field.

Research Aims

New Zealand has been participating in the work of the Joint UNECE/OECD/ Eurostat Working Group on Statistics for Sustainable Development (WGSSD), established by the Conference of European Statisticians in 2005 to provide a theoretical and conceptual framework and to better structure work on indicators. A draft working paper by WGSSD (dated 20 December 2007) is available, entitled Statistics for Sustainable Development: A Framework for Sustainable Development Indicators. The working paper is attractive, because it offers a single framework for integrating social, economic and environmental indicators, based on the capital approach to measuring sustainable development. Like most international studies in this area, however, it does not include specific attention to cultural indicators.

Professors Caroline Saunders and Paul Dalziel in the AERU Research Unit at Lincoln University have also been undertaking research on the capital approach to measuring sustainable development. This began with three different reports on economic indicators for six local government districts in North Canterbury, Selwyn and the West Coast of the South Island (a list of their joint publications is appended below). The AERU research programme has included cultural well-being as an essential component of its approach; indeed a paper on this topic received an Award for best paper at an Australasian conference in 2006.

Statistics New Zealand has recently commenced a project to develop and publish a conceptual and statistical framework to measure sustainable development in New Zealand. This will leverage off the work of the WGSSD, and build on Statistics New Zealand’s 2002 publication, Monitoring Progress Towards a Sustainable New Zealand: An Experimental Report and Analysis. The initial conceptual and statistical framework is anticipated to be completed by July 2008, and so will be available for the project described in this proposal. Research on specific topics and indicators will proceed throughout the rest of 2008, with publication of the results scheduled for June 2009. The publication will report on national measures of sustainable development. The regional and local aspects will be covered through looking at the contribution of different geographic areas to the national picture, in topics where this is relevant. The aims of the project outlined in this full proposal are centred on the current Sustainable Development project. The overall objective is to strengthen that work through collaboration by the research teams in Statistics New Zealand and the AERU Research Unit. This overall objective has resulted in three aims for the research:

1. To complement work already being done within Statistics New Zealand’s Sustainable Development project in order to strengthen the conceptual framework underlying New Zealand’s sustainable development indicators framework.
2. To create new knowledge and report on the conceptual issues that must be addressed if equity is to be incorporated into an integrated capital-based sustainable development indicators framework.
3. To create new knowledge and report on the impact that including cultural well-being has on all the statistical indicators adopted in an integrated capital-based sustainable development indicators framework.

Project Sponsor: Statistics New Zealand
Project team: Lincoln University, Statistics New Zealand