April 10, 2019

Census 2018 undercount

Statistics New Zealand has revealed to Parliament the scale of the undercount for Census 2018. In total around 700,000 people either did not participate or only partially answered the Census 2018 questionnaire.

Statistics New Zealand has revealed to Parliament the scale of the undercount for Census 2018.

In total around 700,000 people either did not participate or only partially answered the Census questionnaire. In total around 480,000 people did not participate in Census 2018. Given the estimated population at the time was 4.88 million, this equates to 10 percent of the population. To provide some further context, the 480,000 people is around the same size as the entire Waikato Region.

Added to the people who were missed entirely, were the 240,000 people who only partially completed the Census. This equates to five percent of the population, or about the same size as the entire population of Dunedin City and Lower Hutt City combined.

Looking back to the 2013 Census, two percent or around 90,000 people partially answered the Census, while seven percent did not respond or around 315,000 people. This means in 2013 we had a total of 405,000 people who either did not participate or only partially answered the Census questionnaire. This is around 60 percent of the total seen in 2018.

Given that some population groups are more likely to be missed, we can expect that the undercount rates for Māori, Pacific peoples, and young adults (aged 15–29 years), will be higher than the rate for the total population. This will call into doubt the ability of researchers and users of the data to undertake meaningful research with these groups and other small population groups. This is because, for the 2018 Census, Statistics New Zealand will be looking to impute the answers for 1 in 7 New Zealanders.

With the scale of the undercount now known there are increasing calls for Statistics New Zealand to abandon Census 2018 and hold a fresh Census in 2021. Having 15 percent of the population either missing from or only partially captured represents a failure of the Census to fulfil its purpose.

I believe Statistics New Zealand should release the 2018 Census data as soon as possible, in the best shape possible, as there is still some value in the data despite the large missing component. The data might still be useful - for Statistics New Zealand in its review to undertake Census 2021, to restore the public’s confidence, and the decisions made by Central and Local Government based on the data.