Young men hardest to count in census
Men, especially those aged 15–29 years, were more likely to be missed than women in the 2018 Census, according to the 2018 post-enumeration survey (PES), Stats NZ said today.
The male population overall had a higher net undercount (3.1 percent) than the female population (2.1 percent) in the 2018 Census.
The 2018 PES provides the official measures of census coverage, which includes people who completed census forms as well as people counted using administrative data. The PES shows the disparity was more evident for young adults than the wider population. People aged 15–29 years had an overall net undercount of 4.7 percent, with a higher rate for young adult men at 5.5 percent, compared with young adult women at 3.8 percent. Conversely, people at older ages were much less likely to be missed in the census. Men aged 65+ had a net undercount rate of 0.5 percent, and for women it was just 0.2 percent.
“Coverage in the 2018 Census was high, despite the lower than expected response to the census. This reflects the use of government administrative data sources to fill gaps when producing the 2018 Census usually resident population count,” population insights senior manager Brooke Theyers said.
Census coverage in 2018 varies by sex, age, ethnicity, and area of usual residence, and this is consistent with results in previous censuses.
Census Day was 6 March 2018, with the PES held shortly after this from 26 April to 2 July. The PES conducted interviews at 15,000 private dwellings throughout the country. Its purpose is to independently measure how many people didn't get counted by the census, and how many were counted more than once or counted in error. The PES has been carried out in New Zealand since 1996, and a post-enumeration or similar coverage survey is a core aspect of many censuses worldwide. Despite concerted efforts to count people correctly, censuses everywhere tend to miss people, and usually to a lesser extent, count them more than once.
“Nationally, the net census undercount rate was 2.6 percent (± 0.3 percent), or 124,800 people. This resulted from 171,800 people being missed in the 2018 Census, offset by 46,700 people who were counted more than once or counted in error.”
The net undercount in 2018 is similar to that of the two previous censuses: 2013 Census (2.4 ± 0.5 percent) and 2006 Census (2.0 ± 0.4 percent), and it compares well internationally.
PES results are a key input to the 2018-base estimated resident population New Zealand's population nears 5.1 million, also released today, which is the foundation dataset for population estimates and projections produced following the census.
Interim coverage rates
The PES provides the official measures of census coverage, and is specifically designed for this purpose. Prior to the availability of the PES results, Stats NZ produced an interim assessment of the coverage of the 2018 Census using the best information and methods that could be developed at the time. The PES results supersede these interim rates, which had indicated a net undercount of 1.4 percent nationally in the census.
Overall, the variations in coverage across sub-groups between the final PES results and interim rates are broadly similar. However, some differences are observed, particularly for children aged 0–14 years.
Coverage by broad ethnic group
People of Māori or Pacific ethnicity were more likely to be undercounted than other ethnic groups, with net undercounts of 4.4 and 4.9 percent respectively. The European ethnic group had the lowest net undercount of 1.8 percent. The younger age profile of the Māori and Pacific populations contributes to the higher undercount for these ethnic groups.
For territorial authorities and Auckland local board areas, net undercount was highest in the Māngere-Ōtāhuhu (4.0 percent), Ōtara-Papatoetoe (3.8 percent), and Manurewa (3.6 percent) local board areas, and lowest in the Thames-Coromandel (1.6 percent) and Buller and Carterton districts (both 1.7 percent). The areas with higher undercount generally have younger and more ethnically diverse populations. Similar coverage patterns were observed in previous censuses.
Census response rates
As well as measuring census coverage, the PES also provides the official measure of census response. While coverage was high, census response – the proportion of people completing their forms – was lower than in previous censuses. In the 2018 Census, the overall response rate was 82.0 percent, down from 92.2 percent in 2013. The lower rate in 2018 reflects the lower than expected response to the census, which led to the inclusion of admin enumerations in the census counts using data from government administrative sources.
Delayed release
The PES results were scheduled for release in March 2020, but were delayed until 23 September 2020 due to needing to resolve some methodology issues. This work has been completed, and further information on the delay is available in the PES report and first release content.
