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+
Official Statistics Research Series, 2010

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

Stephen Haslett, Geoff Jones, Alasdair Noble, & Dimitris Ballas

Abstract

The principal aim of this research has been 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 considers, summarises and further develops a range of statistical methods, and considers their application in principle to set of case studies in sociology, economics, and business from a variety of New Zealand government departments and ministries. Guidelines, based on this research, for use of these three core techniques are also provided.

Available files

Read the research paper online, or download and print the PDF from 'Available files'. If you have problems viewing the file, see opening files and PDFs.

Published February 2010

ISBN 978-0-478-31520-2 (Online)

Keywords

Mass imputation, small area estimation, spatial microsimulation.

Sexual orientation data in probability surveys: Improving data quality and estimating core population measures from existing New Zealand survey data

Frank Pega, Alistair Gray, & Jaimie Veale

Abstract

There is an increasing awareness amongst policy makers, service providers, and community organisations of the need for official statistics on sexual orientation. Principal information needs are apparent in the broad categories of enumeration, discrimination, and social well-being. As a result, the Official Statistics System of New Zealand (OSS) has commenced collecting sexual orientation data in some official probability surveys. A number of official statistical agencies in other countries have collected sexual orientation data for some time, providing valuable knowledge and experience in this area.

The OSS is not currently guided in its collection of sexual orientation data by an overarching conceptual and measurement framework, and this poses a potential risk to the quality of data collected. In addition, the lack of reliable denominator data makes it difficult for the OSS to validate drawn samples. The Ministry of Social Development established the Sexual Orientation Data Collection Study (SODCS), in close collaboration with Statistics New Zealand and the Ministry of Health, with the dual aims of developing robust conceptual and measurement frameworks to improve the quality of sexual orientation data collected in New Zealand and estimating core population measures from existing New Zealand survey data on sexual orientation. Independent researchers and statisticians were commissioned to conduct the study on behalf of the Ministry of Social Development.

A multi-source, multi-method research approach was taken. First, we reviewed and analysed the existing body of conceptual and methodological literature in the area. Second, we held focus groups and key informant interviews with takatāpui, fa‘afafine, lesbian, gay, and bisexual individuals, as well as producers and consumers of official statistics. Third, we analysed and modelled sexual orientation data from existing official New Zealand surveys. Throughout the project, we drew on the expertise of a 14-member advisory board and additional experts, including producers of official statistics.

The SODCS has produced four specialist reports (Appendices 1–4), and the key findings from each of these are summarised in the current document.

SODCS Report 1: Sexual Orientation Conceptual Framework is a guiding conceptual document for sexual orientation data collection in the OSS. The conceptual framework discusses the culture and gender frames of sexual orientation in New Zealand, the three key conceptual components of sexual orientation (ie, sexual attraction, sexual behaviour, and sexual identity), sexual orientation as a continuum, and fluidity of sexual orientation. Finally, a working definition for sexual orientation as a statistical topic and definitions for the three key measurement concepts (sexual attraction, sexual behaviour, and sexual identity) are proposed.

SODCS Report 2: Issues in Sexual Orientation Measurement and Data Collection discusses prominent methodological issues in relation to sexual orientation as an official statistic. These include the conceptualisation and definition of sexual orientation, the design of sexual orientation questions, size and demographic distribution of sexual minority populations, sampling populations defined by sexual orientation, survey mode, misreporting and non-responding, and acceptability of sexual orientation survey items. The report outlines strategies for addressing the methodological issues discussed.

SODCS Report 3: Current Best Practice in Sexual Orientation Data Collection makes recommendations about the praxis of collecting sexual orientation data, drawing on the findings and conclusions of the conceptual and measurement frameworks. This document includes preliminary sexual orientation questions and statistical classification standards for sexual orientation which could potentially be used as a starting point for further development of sexual orientation data standards and classifications.

SODCS Report 4: Sexual Orientation Data in New Zealand Probability Surveys – Technical Report assesses the potential of existing New Zealand survey data to improve current estimates of sexual minority populations. Based on careful analysis of currently available data, we conclude that these are of limited suitability for this purpose. Potential sampling strategies to increase sexual minority sub-samples are explored. We also investigate characteristics of refusals, don’t know, other; and missing responses in existing New Zealand survey data for the purpose of determining the percentage of these response types likely to have been elicited from heterosexual and sexual minority respondents respectively. A robust statistical model for the estimation of population characteristics of sexual minority population groups is presented, which includes adjustments for the misreporting and non-response phenomena.

Principal conclusions from the SODCS are that:

  • The conceptual framework developed by the SODCS and presented here provides a solid foundation for improving the quality of sexual orientation data collected in the OSS.
  • The identified measurement and data collection issues relating to sexual orientation data are all amenable to resolution to a degree that would ensure the collection of timely, accurate, reliable, comparable, and high-quality sexual orientation data in New Zealand; the SODCS recommends strategies to address each of these issues.
  • An important further step will be to develop standards for sexual orientation as a statistical topic; the SODCS presents some of the key information and tools needed to produce such a statistical standard.
  • The preliminary sexual attraction, sexual behaviour, and sexual identity questions developed by the SODCS could potentially be used as standard questions by the OSS, pending further testing, development, and trialling along the lines suggested. The approach taken by the Office for National Statistics (United Kingdom) could guide this process.
  • Although sexual attraction and sexual behaviour data are needed to address certain information needs, significant additional development of the proposed preliminary questions on sexual attraction and sexual behaviour is needed before these can be included in OSS surveys. A conceptually sound, pre-tested sexual identity question tailored to the New Zealand and OSS contexts is available, and would require only minor additional testing and development before being fully fit for use in OSS surveys. As demonstrated in the United Kingdom, sexual identity questions can be effectively included in the demographic core of official surveys to provide data on a range of key information needs related to sexual orientation.
  • Sexual orientation data collection should aim to meet the best practice standard of collecting data on all three measurement concepts (sexual attraction, sexual behaviour, and sexual identity). However, prioritisation of the sexual identity concept for inclusion in OSS surveys (possibly in the demographic core) represents an acceptable transitional step until data on all concepts can be collected. To overcome the limitations associated with the prioritisation of a question on sexual identity, the OSS would then need to concurrently work towards standard sexual attraction and sexual behaviour questions for OSS use and establishing a statistical standard for the sexual orientation statistical topic, eventually achieving the best practice of collecting data on all three measurement concepts at the earliest possible point in time.
  • The approach to producing robust estimates of populations defined by sexual orientation developed by the Office for National Statistics (United Kingdom) is an appropriate model for use in the OSS. That is, development and implementation of standard sexual orientation questions in large-scale OSS surveys and, in turn, pooling data from sexual minority samples to achieve large enough data sets for robust analysis and imputation. In addition, the SODCS proposes a model for adjusting data to account for misreporting and non-responses, based on current best practice in this area.

Available files

Read the research papers online, or download and print the PDF from 'Available files'. If you have problems viewing the file, see opening files and PDFs.

Published February 2010

ISBN 978-0-478-35360-0 (Online)

Keywords

Sexual orientation, conceptual framework, measurement, data collection, probability survey, small area estimation.

 

  • 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+