Why we need to save the 2016 Census

It has been 2 weeks since the ABS dropped the bombshell that the Australian Statistician is proposing to cancel the 2016 Census. In that time there has been a lot of support for the Census, but several major commentators (including Ross Gittins of the Sydney Morning Herald) have come out in favour of replacing the Census with an annual population survey. They claim that surveys can replace the Census. The Statistician himself, David Kalisch (who has only been in the job for 3 months), is now on record as saying that the Census “is not as useful as previously thought”. This is a disturbing statement from the head of the organisation for whom the Census is the flagship statistical collection, and it demonstrates how out of touch with the users of ABS statistics he is. These commentators are all completely missing the point about Census. Here is why a survey can’t replace a Census.

The value of local information

Here are just a few things which have come out of recent training sessions I’ve done with .id’s 250 council-strong client base, which can only come from the Census.

• The City of Parramatta’s Arabic speaking population has a higher Australian-born component than the total population of Parramatta. This changed people’s perceptions of the local population and challenged some myths.
• The City of Campbelltown has the most socially diverse set of suburbs of any Local Government Area in the country, with SEIFA index of disadvantage ranging from Claymore in the bottom 1% of the nation to Macquarie Links in the top 1%. This highlights the challenges of service provision and budget allocation for the City.
• NSW residents born in India had university degree qualifications at twice the rate of the general population, challenging perceptions about Australia’s migrant population..
• The City of Melville’s most affluent suburb, Winthrop, is also its most culturally diverse, with 35% of the population speaking a language other than English at home, and this is increasing rapidly since the 2006 Census.
• The indigenous population in the City of Armadale are predominantly buying their homes with a mortgage but in 2011, more indigenous households were moving from home ownership and public rental into the private rental market. This indicates to council that a key asset of the area – housing affordability – is being eroded.
• While 63% of all households in Victoria’s Mornington Peninsula held only one or two residents, and this proportion is increasing slowly over time, the increase in dwelling stock between 2006 and 2011 was mostly in larger households, with an extra 1,641 four bedroom homes over this period. This indicates a need for more diverse housing in the area, to allow older populations to downsize.
• The City of Maroondah has an emerging Burmese community, who all arrived in the area within the last 5 years (and some migration has continued since Census). The Burmese population are relatively young, and most have poor English proficiency, indicating a need for translation and support services from council to allow this community to integrate successfully.

These are just a handful of the stories which our local government clients uncover every day using data originally sourced from the Census. As you can see, many of them involve change over a 5 year period, which would be taken away from us if the 2016 Census is cancelled.

What these facts all have in common is that they relate to small areas or small population groups.

Yes, the Census could be replaced by a survey for the larger estimates of population and some characteristics – at the state and national level. This would still be quite accurate if you are interested in Australia or one of the larger states. The ABS has always been primarily focussed on producing national statistics, and the legislated requirement for the Census is to produce national and state population estimates for the distribution of electorates. The small area data produced by Census has been seen almost as a “by-product”, and this is what makes the proposal to cancel the Census so disturbing.

Because it is this small area data which is used every day to make important decisions for local governments and millions of businesses and non-profit organisations working in their local areas.

Why a sample survey can’t replace the Census

This is what we would lose if the Census moves to a survey. Even a large sample survey can’t provide small area estimates for all the population groups the Census does.

The reason is that even if you have a large enough sample size (and once you get to a large enough sample size, in the millions, you might as well run a Census), it has to be stratified correctly. ie. If you are trying to measure the characteristics of Arabic speaking mothers in Parramatta, you have to ensure that enough Arabic speaking mothers in Parramatta are included in the sample in the first place. So you have to predict the questions people will want to answer in advance and tailor your survey accordingly. This can’t happen for a national survey, which effectively means there would be a need to run a new survey every time you want to answer a local question. The Census has all this information readily available in one 5-yearly collection.

Since 2011, Australia has had rapid population growth, a mining boom and bust, a significant change in migration patterns due to government policy, and a shift in housing construction towards New South Wales. To say that we might have to wait until 2022 to get any information about these issues at the local level is asking decision makers to wait an eternity, and we have no doubt that poorer decisions will be made because of it.

There is still time to save the 2016 Census.

A final decision on the Statistician’s proposal has not been made. While apparently behind on planning, substantial testing for the 2016 Census has already been done by ABS. Radical changes were already proposed for 2016 to move the Census largely online, and to a mail-out model. If there isn’t time to implement these major changes, just fall back on the previous version, which has been tried and tested many times, most recently very successfully in 2011. Get those Census collectors out there and stop talking about cancelling the most important statistical collection we have.

As far as we can tell, the ABS has gone into hibernation and is not accepting public submissions on this (the official Census consultation process was completed in 2013 and the website still says that the results of this will be out in late 2014). But since the government will have to legislate to remove the requirement for 5 yearly Censuses, there is still time. Local Government can lobby their local federal members to vote against the plan and ensure that they hold the ABS to the current legislative requirements for 5-yearly Censuses.

Please send us your comments at the bottom of this page and tell us what losing the Census will mean for you and your organisation.

Read related blogs:

Axe the 2016 Census? You’ve got to be kidding!

Let’s join forces to Save the 2016 Census