One in six New Zealanders did not complete the 2018 Census – a result which has already led to the resignation of the Government’s chief statistician.
Today the first set of data from the Census was released, including electoral population, population and dwelling counts. However, the low response rate meant Stats NZ has needed to use alternative data to fill the gaps.
An external data quality panel was set up in mid-2018 to provide independent advice. Its first report, released today, has endorsed some of the statistical methods used to plug the gaps. However, it has downgraded the quality of ethnicity to ‘moderate’ – rather than ‘high’ quality – and warned that in regard to iwi data Stats NZ has not met its Treaty oblications.
The panel’s second report, on data quality, will be published in December.
The SMC gathered expert comment on the report, feel free to use these comments in your reporting.
Associate Professor Polly Atatoa Carr, Public Health Physician, National Institute of Demographic and Economic Analysis, University of Waikato, comments:
“As independent advisors of the Government statistician, the data quality panel has considered the implications of the high levels of non-response for the 2018 New Zealand Census of Population and Dwellings. This non-response has resulted in significant use of previous Census data and administrative data to fill gaps and to improve data quality.
“The biggest gaps in 2018 Census are seen for Māori and Pacific peoples, and while the data quality panel has today endorsed the statistical approaches used to repair these holes, there are important questions raised, and important lessons to be learnt. High quality Census data, that we can be confident in, is critical in order to deliver equitable services and to monitor our responsibilities to our population – particularly to our Māori and Pacific communities.
“I highlight the challenges raised by the data quality panel, particularly in the same year we have seen the delivery of the first national ‘Wellbeing Budget’ and an unprecedented focus on the need to achieve health equity in New Zealand.
“While it is good that the administrative data has helped avoid a significant underestimation of important communities in Aotearoa New Zealand in Census 2018, the processes undertaken mean that the quality of data is not the same for all groups. Typically, the range of lost information is wider and of lower quality for the data reported of Māori and Pacific communities. For example, if we look at ethnicity data, there is lowest quality for Pacific populations – particularly when more detail is required, such as specific Pacific nationality, or Pacific peoples in specific geographic locations. These quality differences for our different populations create risk when that data is used.
“Of all ethnic groups, the highest proportion of Pacific people’s ethnicity data has been sourced from administrative data sources such as that found in the Department of Internal Affairs, the Ministry of Education and the Ministry of Health.
“Some data is described as even more limited, such as that collected about households and families where there is very little reliable administrative data to draw on. Not only does this result in important information gaps now for our families, but this also makes research on trends over time difficult within Census data.
“Looking at how things have changed for families over time is another very important mechanism that we have to see the impact our services, such as health, education, housing and social supports are having, and whether we are reaching where we need to in the way that we need to.
“I agree with the data quality panel regarding the need for comprehensive and open public consultation following these new Census data transformation processes. Pacific peoples have been most affected by the use of other data to improve the quality of Census 2018. Pacific peoples therefore need to be more involved in the decisions made about how these methods are applied, and what they mean. It is also important for us all to now have open community conversations about how acceptable and ethical it is to use administrative data in this way, and how Census can be improved in the future.”
No conflict of interest.