Official Statistics Research funded 7 projects in the 2006/07 year.
Classifying the ethnicity of families and households | Investigation of the suitability of ACC data | Improving regional level estimates from national surveys | Identification and Integration of Child Abuse and Neglect Statistics | Estimating Samoan, Cook Island Māori, Tongan and Niuean mortality rates | Feasibility Study: Applications of Multivariate Mixture Models to Official Statistics | Case Study: Create Design-Standards for NZ Police Administrative Data
OS Research will be inviting the public to scheduled Officials Statistics System Seminars throughout 2006 and 2007, 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. Classifying the ethnicity of families and households
The Sustainable Population website states that ethnicity is a personal characteristic and cannot be ascribed to families, households, businesses or other social constructs. However, it goes on to note that there is a widespread perception that considerations of fictional entities such as 'European families' or 'Pacific households' may add value in some way to analysis. The Review of the Measurement of Ethnicity undertaken by Statistics New Zealand (2004) also found that ethnicity was a personal attribute that could not be attributed to a group. Statistics New Zealand has, nevertheless, continued with a programme to collect data on ‘Māori businesses’ which is a group based measure. In addition, the Ministry of Health is commencing a project to see if it can conceptualise and measure whanau.
Ethnicity of individuals is an important predictor of behaviour such as household composition, fertility, health status and mortality by age and employment by industry or occupation which affect demand for services and infrastructure. Therefore, ethnic composition of households may be a predictor of demand characteristics of households and families. Based on a mixture of theoretical consideration, empirical investigation and consultation, the discussion paper will explore a number of issues. These include:
- What are the uses made of ethnic family/household measures in research and policy making in New Zealand?
- Are these common enough uses to warrant an official measure and if so, what sort of measure should be used?
- Alternatively, should a range of measures, including perhaps a family/household ethnicity scale, be developed to suit a variety of uses?
- What sorts of results are obtained using different methods of classifying the ethnicity of families and households applied to census data?
Project Sponsor: Families Commission
Team:
- Victoria University at Wellington, Institute of Policy Studies
- Statistics New Zealand
- Monitoring and Evaluation Research Associates Ltd.
Back to top
2. Investigation of the suitability of ACC data for describing the epidemiology of serious non-fatal work-related injury
Background
Many Government agencies are interested in reliable statistics to describe the size and nature of, and trends in, the work-related injury problem. There have been problems identifying and describing work-related injury in New Zealand on an ongoing basis. There are two main issues:
- the quality of data relating to key fields (eg. circumstances and diagnosis of injury);
- the biasing effects on ascertainment of cases and trends in incidence rates of extraneous factors that have nothing to do with incidence rates (eg. changes in provision of services, unemployment rate – which affects the probability of an ACC claim. There are two national administrative data sets, ACC data and NZHIS NMDS of hospitalisation, that are potential source of these statistics.
Aims
To present an accurate picture of the epidemiology of serious injuries In this project, injuries are described as serious if they are associated with a significant threat-to-life. using a linked data set, and to investigate the suitability of ACC data to do likewise.
Methods
For the linked data set of ACC to NMDS, a cross-sectional analysis will be carried out. Numbers, and where possible rates, of injury will be presented overall and for the following dimensions:
- age
- gender
- ethnicity
- industry
- diagnosis
- and by external cause of injury.
This descriptive picture will be compared with the picture for the analysis of ACC data on their own. Various thresholds for the length of time off work will be investigated: over 7 days, over 14 days, over 30 days, etc. Additionally, the concordance between NZHIS hospitalisation and ACC data, for common fields, will be considered.
Significance
Addressing the aims will provide a description of the epidemiology of serious non-fatal work-related injury for the first time in New Zealand using a case definition that minimises bias. It will also provide the opportunity to judge whether an epidemiological description based on ACC data alone is trustworthy. If it is, it will provide confidence to relevant government and non-government agencies to use ACC data to support priority setting, policy development and prevention practice.
Project Sponsor: Statistics New Zealand
Team:
- Injury Prevention Research Unit, University of Otago.
Back to top
3. Improving regional level estimates from national surveys using census and administrative data
– a case study using the 2002/03 New Zealand Health survey.
The key research question is: Can statistical models (in combination with census and administrative data) be used to produce regional estimates (of publishable quality) for a range of key variables from national level surveys in New Zealand? The project will focus on the NZ health survey and health related variables – but its results will be generally informative with regard to other national surveys run in NZ (for example the Labour Force Survey, the planned general social survey etc.). It will involve examining methodologies used in similar applications found in the international literature, applying these methodologies to the New Zealand data and examining the quality of the resulting outputs. The research has a ‘regional’ focus rather than a ‘small area’ focus. Although the methodologies used will be what are called ‘small area estimation’ techniques, it is important to note that the outcomes we will be focussing on are estimates for ‘large areas’ such as regional councils, district health boards etc. (and not territorial local authorities, census area units, nor meshblocks.)
Project Sponsor: Ministry of Health
Team:
- Ministry of Health
- University of Wollongong
- Statistics New Zealand.
Back to top
4. Identification and Integration of Child Abuse and Neglect Statistics for Otago and Southland
The identification and integration of child abuse and neglect statistics at a regional and local level will provide a tool for government agencies to measure the impact of abuse and neglect prevention programmes. The research questions focus on establishing the availability and quality of current data as well as assessing its usability for departmental decision-making. It is proposed in the first instance that a set of indicators and measures be developed for the Otago/Southland region; key components of the project method are:
- identification and integration of statistics from a variety of sources
- provision of a complete data set assessed for validity
- consultation with stakeholders to identify an agreed set of indicators
- identification of data gaps and possible solutions to fill these.
Outcomes for the project include establishing current child abuse and neglect statistics; a description of innovative methods of disaggregating these to a regional and local level; a report on the state of child abuse and neglect in Otago/Southland; an account of how the statistics can support local decision-making; and the challenges in implementing a similar system in other regions.
Project Sponsor: Ministry of Social Development
Team:
- Children's Issues Centre, University of Otago
- Injury Prevention Research Unit, University of Otago.
Back to top
5. Estimating Samoan, Cook Island Māori, Tongan and Niuean mortality rates
Research questions
Is it feasible to estimate reliable smoothed mortality rates using hierarchical Bayesian methods for Samoan, Cook Island Māori, Tongan and Niuean people? Including:
- differences in causes of death for major disease groupings (e.g. CVD, cancer)
- differences in mortality rates between sole and combination groupings of Pacific people (e.g. [Samoan] compared to [Samoan and European]).
Explore a methodological extensions to hierarchical Bayesian methods:
- allowing second level parameters that determine global fit of data to the prior to vary locally depending on adjacent data.
Background, methods and outcomes
Until recently there has been a dearth of reliable data on social outcomes, including health, for Pacific people. In the health sector, the publication of a Pacific health chartbook 7 and correct mortality rates (the latter using NZCMS data) has begun to rectify this data gap. However, a common plea from Pacific communities (e.g. at presentations and symposia) and policy makers is for a further breakdown by specific Pacific grouping (e.g. Samoan, Cook Island). But this is statistically challenging due to small numbers.
We proposed to leverage off NZCMS work being conducted under contract with the Ministry of Health, and explore the feasibility of publishing Pacific-specific mortality rates. To do this, we will incrementally conduct analyses using direct standardisation then Poisson regression before undertaking hierarchical Bayesian analyses to answer the above research questions. (More detail on methodology steps is provided below.)
As suggested in the background above, policy-makers and communities are keen for such information to further enhance prioritisation of Pacific policy, and understand in greater depth Pacific health disparities. The NZCMS already has a high public profile with respect to the presentation of ethnic mortality rates; we would anticipate that the output proposed in this application would also be well-received. The proposed smoothing methodology should also be transferable to other parts of the public sector, such as education, employment, welfare and justice.
Project Sponsor: Statistics NZ
Team:
- NZCMS, University of Otago
- Statistics New Zealand
- Ministry of Health
- Victoria University.
Back to top
6. Feasibility Study: Applications of Multivariate Mixture Models to Official Statistics
The project will investigate whether mixture models can be used to summarise existing datasets in more informative ways, whether they can help identify unusual or erroneous data, and whether the technology could make future information collecting more efficient. Benefits of applying multivariate mixture models to Official Statistics include increased usefulness of existing data and efficiency in future data collection as well as improved checking of data quality.
The aim of this feasibility study is to find the most appropriate official statistics dataset for the methodology and conduct some preliminary analysis and submit a proposal for the next OS Research funding round.
Project Sponsor: Statistics New Zealand
Team:
- Waikato University
- Statistics New Zealand.
Back to top
7. Case Study: Create Design-Standards for NZ Police Administrative Data that are Applicable for the Official Statistics System
By introducing standards and quality principles when undertaking business process changes, IT application development, and form design the project will take a valuable step towards making greater use of administrative data and enabling integration of data sources across the official statistics system. The research questions are:
- What process points provide the best opportunities to apply standards that ensure resultant statistical data in administrative systems is of good quality and able to be integrated with other data from across the official statistics system?
- What standards can and should be applied at these process points?
The methodology will take the form of a case study on New Zealand Police. It will involve interviews and workshops with subject matter experts, identification and collation of documentation on relevant business processes and standards.
The project aims to produce a standard that will be applied by NZ Police to ensure capture of quality statistical information in administrative systems and which could potentially be applied by other agencies.
Project Sponsor: New Zealand Police
Team:
- New Zealand Police
- Statistics New Zealand
- Ministry of Education.