2016 Census Data Release - Old Parliament House

Independent Assurance Panel, 27 June 2017

Professor Sandra Harding, Chair, IAP



I’ll provide a few preliminary comments and then let you know the outcome of the Independent Assurance Panel’s work.

In August 2016, following Census night, the Australian Statistician made the decision to establish an Independent Assurance Panel to review the quality of 2016 Census data.

The specifics of the task set for the Panel are contained in the Panel’s Terms of Reference, located at Appendix C in our report.

The Panel was appointed in October 2016 and Panel members come with a wide range of experience and expertise.

Panel members are:
• Professor Lisa Jackson Pulver AM, Pro Vice Chancellor Engagement and Aboriginal and Torres Strait Islander Leadership, Western Sydney University and member of the Australian Statistics Advisory Council
• Professor Peter McDonald AM, Head of Demography, Centre for Health Policy, University of Melbourne.
• Peter Morrison, former Assistant Chief Canadian Statistician
• Dennis Trewin AO, former Australian Statistician (2000-2007)
• Anton Voss, Deputy Secretary, Tasmanian Department of Treasury and Finance and member of the Australian Statistics Advisory Council
• And myself, Vice Chancellor and President of James Cook University and former Chair of the Australian Statistics Advisory Council (2001-2006).

The Panel worked as a body independent of the Australian Bureau of Statistics to discharge our Terms of Reference, meeting on 8 occasions, in person, by video- and tele- conference and with a great deal of our work undertaken between meetings with discussions occurring via email and through a secure site provided for the purpose by the ABS.

The timing of the Panel’s work has meant that our analyses have been undertaken and conclusions drawn about the quality of the 2016 Census data at the national, state and territory level, but not for data at a more detailed level.

And now to it: I am pleased to let you know the outcome of our work.

Based on our analyses, the Panel has determined that the 2016 Census data is of comparable quality to 2011 and 2006 Census data and comparable to international benchmarks.

Therefore, the Panel has concluded that the 2016 Census data is fit-for-purpose, it is useful and useable, and will support the same variety of uses as has been the case for previous Censuses.

In making our assessment, the Panel used the results of the Post Enumeration Survey, a survey of 50,000 households undertaken independently of and immediately following the Census, to explore data quality. That Survey provides a quality check on 2016 Census data. Consistent with the 2011Census, the Post Enumeration Survey pointed to a net undercount of persons on Census forms. The 2016 result is a lower net undercount than the net undercount for 2011 and 2006. The net undercount is the difference between the undercount and the overcount. While the undercount is comparable to 2011, the Post Enumeration Survey revealed that some non-responding dwellings were incorrectly classified as occupied on Census night, resulting in over-imputation or an overcount.

Another important check on quality involves comparing the Census results with the Estimated Resident Population, which is Australia’s official population estimate. The Estimated Resident Population is rebased using the Census and then adjusted quarterly between censuses, taking account of births, deaths and migration. The expectation is that the new Census data should be broadly in line with the 2016 unadjusted Estimated Resident Population. The Panel found that 2016 Census data aligns well with expectations - and that these data can be used to rebase the Estimated Resident Population.

The Panel also examined a number of key topics in the Census, including population counts, sex, age, income, Indigenous status, country of birth, language, ancestry
and family structure. The Panel found that the levels and distribution of characteristics matched expectations well and were comparable to other data sources where applicable. However, one change goes to an increase in the number of people citing ‘no religion’ in the response to the religion question. Some of this change may have been caused by a change to the question format compared with previous censuses, or it may reflect a true, broader societal change.

The response rate for the 2016 Census is lower than, but comparable to, that from the last two Censuses, and is similar to response rates in other benchmark countries, specifically New Zealand, Canada and the UK. The lower response rate is partly due to the ABS over-estimating occupied private dwellings on Census night.

The Panel also made some observations on matters of interest. These are contained in the concluding section of our Report but today I want to comment on two issues that received a lot of media attention at the time of the Census.

On privacy concerns:
Prior to census night, public concerns were raised about privacy. Impacts are apparent with more persons reporting age rather than date of birth as well as a large decline in the number of people agreeing to have their Census form archived for 99 years. Further, while it is impossible to know whether some people provided valid names that were not their own, few names were withheld or clearly false names reported. While these outcomes have minimal impact on the accuracy of the Census, incorrect names and lack of date of birth for some respondents can hinder data matching activities.

With regard to the withdrawal of the online form on Census night:
We cannot be certain, but withdrawal of the online form for 43 hours from Census night may have resulted in more people using paper forms. This is unfortunate as online completions had somewhat higher response rates for individual Census items. Apart from this, the withdrawal does not seem to have had any particular impact on public co-operation with the Census.

To conclude where I substantively began, based on our analyses, the Panel has determined that the quality of 2016 Census data is broadly in line with expectations, and of comparable quality to the 2011 and 2006 Census data and international benchmarks.

Therefore, the Independent Assurance Panel charged with examining the quality of 2016 Census data has concluded that the 2016 Census data is fit-for-purpose, it is useful and useable, and will support the same variety of uses as was the case for previous Censuses.