I was recently forwarded an article published on the BBC website, which outlined calls by the Conservative Government in the UK to scrap the Census in its current format. The premise of their argument is that it has become too expensive and that the data can be obtained from other sources. Having worked with Census data for many years, I’ve certainly come across this argument many times. But how true is this? Can the data be obtained elsewhere? And can the argument come down to mere dollars? Let’s have a closer look at some of the issues.
I can really only address this argument from an Australian perspective, and though there are some similarities in the way the UK Census is conducted, the most obvious difference is timing. In the UK the Census is conducted every ten years, whereas in Australia it’s every five. The ABS estimates that the 2011 Census cost $440 million, or $19 per person (Read more about the cost of Census here). Economists love being able to measure things in monetary terms, but life isn’t that black and white and the benefits of the Census aren’t necessarily tangible.
A foundation for democracy – hmmm sounds important!
For example, one of the main reasons Australia holds a Census is to determine electoral boundaries based on the distribution of population. Cynicism aside, this is a foundation of our democracy and it requires a sound evidence base to support it. Furthermore, many billions of dollars are allocated on the basis of Census data. In fact, I think it’s great that Australia has a Census every five years rather than every ten, because we get such rich data that enables analysis of urban and regional change on a more frequent basis. I couldn’t imagine having to wait ten years for the data to be updated, like in the UK or US.
The ABS claims (and I have no reason to doubt them) that the Census is the largest logistical exercise undertaken in Australia. The field workforce is enormous, encompassing staff in the ABS offices across Australia, Area Supervisors, and of course, the Collectors at the coalface. This is clearly one of the major costs, but if the Census is going to be successful, some effort needs to be put into the human resource side of the equation. There are a variety of different roles but the most obvious one to most people is that of the Collector, whose job it is to deliver and collect a Census form from every dwelling in the workload. Anyone who has worked on the Census will tell you this is not necessarily an easy task, especially in a country with a settlement pattern ranging from dense urban to very remote. Minor differences in procedures aside, this approach has the advantage of a stable field methodology across the entire country, which is the basis of a sound dataset.
The “data from other sources” argument – how realistic is it?
Many times I’ve heard the argument “but you can get the data from other sources”. Where? What other dataset offers such rich variety on a national scale? Medicare? Not everyone goes to the doctor on a frequent basis and has to make a claim. Electoral role? Well that excludes people under 18, not to mention those who don’t enrol . Council rates databases? Years of dealing with local government has taught me there is a no consistent form of data collection for rates across all councils. Furthermore, it is a database of property owners, not residents, therefore excluding renters from the equation. In my view, it would take a very wide range of administrative data, as well as significant coordination effort, to cover the range of Census data readily available every five years. Not to mention the costs involved in making it useable, issues of comparability, timeliness, etc.
Administrative data can be sourced from both the public and private sectors, but by and large, it is not necessarily collected for statistical research purposes. I’ve had many years experience working with administrative data and I’m sure other researchers will nod their heads in agreement when I talk about the time, cost and even frustration involved in “cleaning” the data and making it useable. Do this on a large scale and suddenly the maintanence and processing costs increase. Not to mention comparability issues – many datasets held by government are administered on a state by state basis, and there can be differences in collection procedures, data fields – all things which can add to the cost of making the data useful and comparable. Yes, there are processing costs involved with the Census, but adoption and continual improvement of new technologies such as optical character recognition has reduced much of these.
So it’s not as simple as scrapping the Census in order to save some dollars. That’s a very economic based approach that does little to consider the less tangible benefits of Census data. Future changes to the Census in Australia will continue to take advantage of new and evolving technologies (eg. the e-Census), when adopted in great numbers, saves field staff a lot of time and the ABS a lot of money). While conducting a Census is an expensive exercise, there is no doubting its value as an evidence base for many purposes. Scrapping the Census and substituting the data through administrative sources simply does not have the same value, and requires a great deal of coordination effort. But what makes me laugh is the intrinsic ironies in the “scrap the Census” argument, because I have a feeling that those in a position to make such a decision, are the same ones who will scream the loudest when the data they require is no longer available.
If you would like to receive more Census updates , do follow us on twitter @dotid or subscribe to our blog (above). You may also like to visit us at id.com.au where you can access our demographic resource centre.