Australia’s Census is our national self-portrait

Tuesday night was Census night. I like to think of the Census as Australia’s self-commissioned self-portrait. Australians participate as artist, sitter and patron all at once. What is revealing is not only the answers we give, but also the picture we paint with the questions we choose to ask.

Early versions of the Census revealed the challenges of colonial Australia. Governor Arthur Phillip’s reports to England consisted of counts of livestock and levels of sickness among the colony.

What has been included in the Census has also reflected the prejudices of the time. Until the 1967 Referendum, First Nations Peoples were excluded from official population statistics. The subsequent 1971 Census included 115,000 people who identified as First Nations.

The 2021 Census was the first to include questions on chronic illnesses, and the 2026 Census the first to include questions on gender and sexual orientation.

One picture among many, but still the only one that gets everyone in frame

For a long time, the Census was the only large-scale data available on the Australian population. But this is now beginning to change. Large administrative datasets like the Australian Bureau of Statistics’ (ABS) Person Level Integrated Data Asset (PLIDA) have become available to policy makers and some researchers. These datasets are extremely valuable, they link together people, businesses, and government services for statistical observation.

But to torture the art analogy further, their commissioned purpose is to support the operation of government. That means you won’t always get everyone in frame. We can’t say much about unemployed people from pay-as-you-go data that captures tax paid by employees, nor can we say much about the healthcare usage by migrants ineligible for Medicare.

The Census as we know it today is commissioned under the ethos of statistics for statistics sake. It is deliberately designed to pull together a large and diverse number of people for statistical observation and can ask questions that administrative datasets miss. On that dimension, it remains invaluable for researchers and policy makers.

Census data underpins much of what we do at e61. Most recently, when we wanted to understand the characteristics of migrants on regional visas, we turned to the 2016 Census. This provided a more detailed picture of the education levels, occupations and the English-speaking capability of migrants.

This information is often not included in visa applications and thus doesn’t reliably show up in the administrative data. Without it we would have not been able to identify that differences in occupation were the primary driver behind regional visa holders earning less than other migrants.

In other work, Census data helped us explore the drivers of the gender wage gap. By linking it to de-identified tax records, we were able to examine the role that income, occupation, education and caregiving responsibilities play. Administrative records of higher educational attainment are not accessible to researchers prior to 2005, nor is there a definitive indicator of caring responsibilities. The 2016 Census filled these gaps.

This analysis led to the somewhat surprising conclusion that differences in occupation only explain 20 per cent of the gender wage gap, with the unequal burden of caregiving instead the key driver. Having a large dataset such as the Census provided a huge advantage over using survey data in this setting, not least because it included a more granular set of occupations.

With only administrative data, we would have also missed the key insight that caregiving is the major driver of the gender pay gap. The Census is the only population wide dataset that captures this information.

These are only 2 examples of how the Census enables the work we do. But they demonstrate how we rely on the trust and cooperation of Australians to complete their Census forms accurately.

We should celebrate, but not take for granted the trust Australians place in the Census

I should probably declare an interest: e61 uses ABS data extensively, and Australian Statistician David Gruen sits on our board. I have had more opportunity than most to appreciate the value of ABS work. But the Census cannot depend on the enthusiasm of researchers alone – it also depends on the confidence of the Australian public.

The 2021 Census had a 96 per cent response rate. This should be one of the most celebrated aspects of the Census.

But this is not the case for everyone. A cautionary tale is the example of the UK where the response rate for the Office for National Statistics’ labour force survey has plummeted to just 25 per cent in early 2026. There is a question of how useful a national survey can be when most people now outright refuse to participate.

It’s worth noting that there is a legal obligation to participate in the Census and the UK labour force survey is voluntary. However, if the Census were to meet even small degradations in responsiveness, or the quality of responses were to drop, then good policymaking would suffer.

At the simplest level, it would be much harder to plan schools, hospitals, transport, aged care, local services or small-area policy with confidence if the accuracy of the Census decreases, or it captures a smaller and less representative slice of the Australian community.

The usefulness of the Census relies on the faith Australians place in the ABS as an institution to use the information carefully and for the purposes it was collected. The ABS is generally a trusted institution, but the Census asks for a particularly intimate kind of trust.

That faith has been tested before, most recently in 2016, when concerns about the retention of names and addresses were compounded by the failure of the online Census form.

The lesson is not that Australians are wrong to be wary. It is that the boundary around what Census data will be used for has to be clear. In other words, it will only be used for the purposes for which it was collected, not for enforcement against individuals.

This boundary has been defended with fervour by the ABS in the past. During WWII, Commonwealth Statistician Sir Roland Wilson pre-emptively set fire to Census of Wealth records to frustrate the Commissioner of Taxation’s enforcement efforts.

Hopefully it won’t require a bonfire again. But that fierce guarding of our privacy is exactly what gives Australians the confidence to answer honestly.

Those honest answers don’t only help develop policy today, they also provide a gift to the future, leaving a record for Australians we will never meet. The 2026 Census gave people the option of preserving their information in the Census Time Capsule, held by the National Archives until 2125.

While we leave this portrait for our descendants, it’s worth noting that portraits often reveal as much about their patrons as they do about their sitter. Bronzino’s portrait of Eleonora di Toledo is one famous example. At face value, it is simply a regal woman in her finest silks. But look deeper and it reveals the political ambitions of the Medici family she had married into, with the luxurious brocades and jewels she’s wearing intended to project the political legitimacy and economic power of a still-new dynasty.

Just as art historians have been able to look back at pieces of Renaissance art to understand the Medici family, our great-grandchildren may look back through these records and ask not just who we were, but also the portrait we sought to paint.