Counting people

Julie Haggie
Chief Executive Officer
Transparency International New Zealand

A very important consultation is underway on the future of the New Zealand census. It seeks your social licence for Stats NZ to access, combine, use, and store your data from various programmes and sources. Effectively it also heralds the end of the five-year census as we know it. This is big.

We strongly encourage everyone to have their say before the consultation closes on 18 June, 2024.

There are a range of views; it’s not a simple issue.

The 2023 Census results and coverage

We are now starting to receive the population statistics gathered from the 2023 census. Very interesting, it shows demographic and geographic changes affected by COVID as well as general population trends.

Stats NZ asserts that 99% of NZ’s total population have been captured in this census and 97% of Māori. The higher Māori coverage was assisted by close collaboration with iwi leaders to encourage iwi led data gathering following a dismal performance in 2018.

A statutory independent review of the management of the 2023 census is available online. The reviewers suggested pushing out the next iteration to 2030 or 2031, as the national statistics agency grapples with a potential new approach.

Census completion

The high population coverage does not mean that 99% of the population completed a census form.

Stats NZ early info shows that 88.3% of the whole population completed the census form, with rates lower for Māori (77%) Pacific (80%) and 15-18 year olds (85%). Their target was 90%.

The 76 question census was a big ask – particularly in this era of five minute attention spans. Stats NZ is still analysing how well the forms were completed and we expect them to share it once it is available.

The completion and accuracy gap poses a problem for such an important count. As Stats NZ says: 

“The census is the only survey in New Zealand that covers the whole population. It provides the most complete picture of life in our cities, towns, suburbs, and rural areas. The data helps the government plan services. These include hospitals, kōhanga reo, schools, roads, and public transport. Councils, iwi, businesses, and other organisations also use the data to work out the needs in their area.” 

Filling the gap

How did Stats NZ ensure the missing 500,000+ were counted? It used ‘administrative data’ as well as data from previous censuses and ‘statistical imputation’ (whatever that means).

Admin data is collected by government agencies or other organisations while conducting their normal business. Examples of admin data are birth, death, and taxation records. For the 2023 Census, admin data was sourced through the Integrated Data Infrastructure (IDI).

For the 2018 census the admin data was retrofitted to fill in gaps. For the 2023 census Stats NZ used a ‘combined census model’ to address data quality issues from low census form completion responses. It only used admin data when the census form had not been completed (or not well enough completed).

View shift

The current consultation indicates a fundamental view shift on the part of Stats NZ reversing the traditional key focus on the census form first backed up by administrative data to an ‘admin-data-first’ approach.

This important view shift of Stats NZ towards the New Zealand census has developed over the last five years. This shift appears to be driven by five major factors:

  • the cost of running a census (for 2023 it was stated as being $316 million) 
  • the importance of having broader coverage which is undermined by falling census-form completion rates 
  • the increased utilisation and availability of large administrative data sets
  • a dangling carrot of more ‘real time data’ and 
  • the demand for better coverage to fill gaps for communities eg Māori, Pacific, LGBQIA+ communities and those living with disabilities. 

Effectively this could see the end of the five-year population census as we have known it since 1851.

Challenges from TINZ’s perspective

  • The census is a deeply embedded element of our ‘social licence’. Completing the census is a feature of citizenship. It is a legal requirement under the Data and Statistics Acts 2022, but that requirement reflects societal attitudes, of contributing, by choice, to our national knowledge base. It cannot be changed lightly.
  • As a legally contained element of social licence, the census has durability, consistency and transparency over time. Moving entirely to administrative data is a shift from the known to the unknown. Will administrative data have the same level of long-term sustainability and public scrutiny?
  • Where do privacy, privacy access and consent rights sit in the use of administrative data? NZCCL note the exclusion of public access to what is held in the Integrated Data Infrastructure (IDI).  We volunteer data in the census, but cannot access that held about us in the IDI.
  • Are we there yet with administrative data? How well can administrative data sources, either singly or linked together meet the needs and expectations of users? Administrative data may not be able to attribute socio-demographic information, it may be difficult to capture information on households, there may be biases in the information that is collected that results from the context in or purpose for in which it is collected.
  • No matter what data method gathering is prioritised, there will still be hard to enumerate groups, especially Māori, Pacific and younger adults. In the end which will be more effective?
  • With the weak balance between expert advice and political expediency, downstream implications are unlikely to be fully recognised or risks protected.
  • Civil society organisations and those who function as guardians of the public interest must have reason for confidence in those statistical practices they will not have the expertise to evaluate.
  • The unheralded termination of the Living in Aotearoa Survey in early 2024 left no opportunity for prior expert consideration of the implications. We don’t know what we don’t know, and the consequences could be costly.
  • Whatever course of action is chosen for the Census, it must be a scientifically valid solution, with collection and verification processes that are transparent.

Commentary about the proposed change to prioritising administrative data

Len Cook, former NZ Chief Statistician and former UK Statistician, articles in TINZ Transparency Times and Victoria University Policy Quarterly

‍New Zealand Council for Civil Liberties

Independent Statutory Review of the 2023 Census by Murray Jack and Geoff Bowlby

Blog Post written by:
No items found.