In the lead up to the 2015 Budget, there was much discussion about possible cuts in funding to the Australian Bureau of Statistics (ABS) that might lead to dramatic changes to the Census methodology. In particular, there appeared to be the possibility of changing from a once in five year methodology to a once in ten, with the intervening period supplemented by improved estimation based upon administrative and other data. Many people, members and others, raised their concerns about these changes to me, since they saw the Census data as a key input to their work, as well as to major government and business decisions.
As President of the Society, I wrote to the Australian Statistician, David W. Kalisch, acknowledging the long-standing relationship between the Society and the ABS, as well as raising these concerns regarding the Census, and expressing the hope that any changes would be based on good statistical principles, and not on the pressures of inadequate funding.
The reply and the subsequent news about funding were largely reassuring, but one issue remains a concern for me. The consultation concerning possible methodological changes appears to have been relatively narrow, focusing on government agencies, and I suspect that within many agencies themselves it was also narrow. This presumably reflects the reality that the purposes of the Census are defined by legislation, and the ABS must work with the primary uses as referenced in that legislation. However, the concerns I received clearly indicate that the use of the Census today is much wider than this.
The removal of most charges for accessing Census outputs, an achievement of Denis Trewin’s time as Australian Statistician, is a major cause of the wider use of the Census. One implication is that the consultation about future Census development should ideally be much wider. The second, more subtle, implication is that the diversity of uses of the Census make it much harder to implement sophisticated estimation to replace the hard work of collecting raw data – estimation typically makes assumptions which may be suitable for some purposes and not for others, and every day the Census is being used for new problems. Neither of these implications is good news for an agency with barely adequate funding, but they must be faced!
I believe that these issues are of central concern to all statisticians in Australia. If we don’t support the collection of good data and its use in making good decisions, who will? The Census is the highest profile statistical event in the Australian community. Many of our members are involved in designing it, managing it or using its results. And I would argue that a robust Census with a rich set of questions should result in better informed public debate, which is essential in a democracy.
We should all make our views and concerns about this known. In my role as SSAI President I will be trying to do that as much as I can.
Dr John Henstridge,
Please leave any comments in the box below. We look forward to hearing your views!