Why I Take the Population Census Seriously

The low coverage of the 2018 Population Census and Dwellings has generated all sorts of difficulties. You are told that census results are vital for allocating electoral seats and  education and health funding. That is absolutely true but social researchers use its findings for many other purposes. For instance, it has a comprehensive survey of our housing stock and of people’s internal migration while it also used to guide the disability survey. Its detailed age and gender structure is vital for national and regional population projections.

(It may be especially useful for the reconstruction arising from Cyclone Gabrielle, especially for housing on the margins and isolated. Censuses are very good for dealing with such groups.)

Getting the count accurate is so important that there is a follow-up survey to avoid an undercount. One of the paradoxes of the census is that while every person and house should be covered, individual responses are only of interest when they are aggregated. Your response is not available to any other agency. Government Statisticians have told me they would resign if the statutory protection of the census was compromised.

The professional statisticians of Statistics New Zealand did a marvellous job patching the 2018 Census using administrative data but the Census collects information which is not in the various administrative data bases. Perhaps once a month I am working on something, consult the 2018 data base and say ‘damn’ because it is not there or I cannot trust its figures.

You might think that surveys fill the gap. For instance, once the only source we had on the state of the labour market was the quinquennial (five-yearly) census. In 1931 and 1941 it was not held; the first, in order to reduce government spending during the Great Depression, the second, because the country was at war. Uh, oh; those were two times when the level of employment and unemployment were exceptionally important. There is an argument among quantitative historians on just how much unemployment there was during the Depression with all sorts of estimates. The war problem is that with soldiers overseas and the retired and housewives replacing them, we have little idea what was happening in the domestic labour market.

(Even the 1945 census is problematic. It was held in September because they needed the data to allocate seats for the 1946 election. The traditional date for the census is the first Tuesday in March – in the next case, 7 March 2023. The labour market is seasonal, so one has to be cautious comparing a September figure with a March one.)

I found a surprising fact while looking at the unemployment data before the 1930s. The reported unemployment rates are low – below 3 percent. A 1931 census would have had a rate a number of times higher. Oh, 1991 is an exception too; the unemployment rate exceeded 10 percent.

We had a Household Labour Force Survey in 1991. However the HLFS is calibrated against the Population Census; you may find small retrospective changes to its estimates when the census data becomes available. In any case, a survey cannot go into the detail of a census. For example, how many people say they are economists?

There are items in the census which are not available anywhere else, as in the case of the question about religion. (My response gives me more trouble than any other census question. How do I provide a useful description of my complicated beliefs in a mere 39 characters?)

To show the complexity and importance of the census I focus on ethnicity questions. There is a ‘race’ question, where respondents are asked whether they are of Māori descent. That is required by law for the allocation of Māori electorates (a complicated exercise because while adults of Māori descent may or may not register for Māori electorates, children of Māori descent have to be allocated between Māori and General electorates.) Descent questions are a matter of fact, something necessary for legal purposes here.

But ethnicity is subjective. Statistics New Zealand guidance is as follows:

Ethnicity is the ethnic group or groups that people identify with or feel they belong to. Ethnicity is a measure of cultural affiliation, as opposed to race, ancestry, nationality, or citizenship. Ethnicity is self-perceived, and people can belong to more than one ethnic group.

So you may be of Māori ancestry but not judge yourself of Māori ethnicity. That is your right. (A racist regime would not allow that.) About half of those of Māori descent say they have a second ethnicity, usually ‘New Zealand European’ or Pākehā. We don’t know the exact proportion, not for 2018 anyway. What we do know is that is over the years that proportion has been rising. It may be more than half in the 2023 census.

One complication is that while one cannot change the facts of one’s descent (one may learn more about them), people change their subjective ethnicity. And not just between censuses. A study at the University of Otago’s Wellington School of Medicine found differences between what some people record on Census night and what is recorded on their death certificates. The change was enough to affect estimates of mortality rates and the associations with causes of death for Māori.

We tend to ignore the existence of those who describe themselves as ‘Māori-Pākehā’. The usual practice is to lump them in with sole-Māori, but on most socioeconomic measures they sit between sole-Māori and Pākehā. That means that sole-Māori are typically more deprived than all-Māori.

I have been following the convergence between Māori-Pākehā and non-Māori over time but the inadequacies of the 2018 census count meant I have been unable to since 2013. I would have thought that this evolution is one of the most important things happening on the cultural front in New Zealand. The population census is the only reasonably comprehensive source that a social scientist has.

So while, and rightly, we talk about the importance of the census to education, health and housing policy and to the electoral system, the data it collects has many other important uses. Let’s hope that the count on the night of 7 March 2023 goes a lot better than it did five years ago.

PS. One of my grumbles about the census is that while you are encouraged to complete your census form online – I can see the advantage to Statistics New Zealand of doing that – you can’t get a copy of the results for your own records. I have, after all these years, realised that I can fill in my form and transfer the data online from it, keeping the form as a record. I add that Statistics New Zealand and users of the census are very happy for those who are not confident with computers to fill in the form and post it or hand it to their census enumerator.