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+
Discussion

This paper examined the quality of ethnicity information in administrative sources by comparing collection practices with the statistical standard, and by comparing the consistency of ethnic reporting in administrative sources available in the IDI with the ethnic groups that individuals reported in the 2013 Census.

Summary of main findings

Most administrative sources in the IDI use the official New Zealand statistical standard for ethnicity and aim to capture the same concept of ethnicity of cultural affiliation. The key aspects of this concept are for an individual’s ethnicity to be self-identified, and the ability to report multiple ethnicities.

In practice, the collection of ethnicity by government agencies is not entirely in line with the statistical standard for ethnicity. The most common discrepancy is that the question asked differs from the standard. Some agencies also have non-standard ways of treating multiple responses.

The census measures ethnicity at a single point in time (census day), and official population estimates also measure ethnicity at a given reference date. In contrast, many administrative sources reflect information collected from individuals at different points in time.

While the coverage of administrative sources varied, almost all people (99 percent) linked to the census had ethnicity information recorded in at least one data source in the IDI. Ministry of Health data had the highest coverage (98 percent), while coverage of other agencies is more limited because they come into contact with a limited part of the population (for example students, those on working age benefits, or ACC). For the New Zealand–born, ethnicity is available from birth registrations since 1995.

Consistency with census responses varies considerably by agency and by ethnic group. Birth registrations show the highest agreement with census, with ratios of total responses close to 1. Other agencies typically produce lower counts compared with the census for all the main ethnic groups.

A marked difference is seen between people who report single or multiple ethnic groups. For those reporting a single European, Māori, Pacific, or Asian ethnic group in the census, between 80 percent and 96 percent also have the same ethnic group in the administrative sources. However, apart from birth registrations, consistency is much lower for those reporting two or three ethnic groups in the census (less than half agree). The most likely cause of the lower numbers of total responses in the main ethnic groups appears to be fewer people with multiple ethnicities in the administrative sources.

Because of the differing coverage and quality of the data available, in practice responses from multiple sources must be combined in some way. The ‘ever-recorded’ method combines all ethnic groups reported in any source. This inflates the counts of the main ethnic groups, and in particular results in many people with Māori and Pacific ethnicity in administrative data who do not have those ethnicities in the census. Ranking sources on the basis of their agreement rates with census, and using the data from the highest-ranked source for each individual, brings the administrative data closer to the census. While the ranking method does not overcome all the limitations of the source data, results for multiple ethnic groups are better than for any of the sources used alone, apart from birth registrations. A key conclusion is that the method used for combining ethnicity data from multiple sources has a major impact on the results.

Further considerations

Measuring ethnicity is inherently challenging – partly because it is not a fixed concept. People may change how they identify themselves over time, or may identify themselves differently in different environments. Official population statistics on ethnicity are based on responses to the ethnicity question that people give in the census. The census is a single collection, in a relatively neutral context, and does not directly affect the individual. In contrast, the context in which ethnicity is collected in administrative sources varies considerably, and may influence how people respond.

The time reference period is a conceptual difference between official population statistics on ethnicity and that measured by administrative sources. The census measure refers to a single point in time, while administrative sources collect ethnicity at different times depending on the contact a person has with the agency.

The statistical standard for ethnicity encourages using a standardised concept, definition, collection, coding method, and output to promote data consistency and comparability in all official statistics. One of the main areas for improvement for administrative sources is in reducing the variety of questions used. Time-stamping of updates to the ethnicity field also needs to be made available with the administrative data. Efforts to improve ethnicity data already in place (for example by the Ministry of Health) are likely to improve data quality beyond what is evident in this comparison of 2013 data.

If administrative data is to be used to produce official ethnic population statistics in place of the census, then we face additional challenges that arise from using a combination of administrative sources. No single source covers the entire New Zealand–resident population, and an individual may have ethnicity recorded in one or up to six of the main sources in any combination. Linkage errors also introduce incorrect ethnicities.

Better methods of producing ethnicity from a combination of sources are needed. This may include rule-based approaches to remove older or spurious data, and using statistical models combined with independent sample surveys, which could be used to calibrate administrative responses.

Consideration could also be given to rationalising the collection of ethnicities to fewer agencies – thus reducing burden on the public, and allowing resources to be more focussed on quality.

  • 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+