The leading nature of the Census religion question undercounts the non-religious population by more than 10 per cent, reveals a new survey commissioned by the Census – Not Religious? campaign.
An Essential Research poll has found that the question design that the Australian Bureau of Statistics (ABS) had originally proposed using for the 2026 Census would have delivered a ‘No religion’ result well above half of the population.
The survey tested the same question design that the ABS had proposed using for the 2026 Census (“Do you have a religion?”) versus the leading question (“What is your religion?”) that it ultimately opted to re-use.
When asked “Do you have a religion?”, 53.7 per cent of respondents picked ‘no’.
When asked the question being used for this year’s Census, 43.3 per cent of respondents selected ‘No religion’.
The Essential Research poll, conducted earlier this year, surveyed more than 2,000 people for each question design.
In October 2024, the ABS decided to abandon its proposed new question design after the Catholic Church asked the Albanese government to “reverse” the proposed changes to the question and after the government then failed to meet ABS deadlines ahead of a “critical” major test.
The Census – Not Religious? campaign has long argued that the leading Census questions inflates the religion result.
Previously, several other robust surveys from recent years — using various alternative questions — suggested that the non-religious population may be undercounted by as much as 11 percentage points.
The new Essential Research poll is the only survey among these that tests the ABS’ proposed — and now abandoned — new question design against the existing design.
The poll’s finding that 43.3 per cent of respondents answered ‘No religion’ to the leading question suggests that this year’s Census will witness a further rise in non-religious affiliation from the 38.9 per cent recorded at the 2021 Census.
Michael Dove, spokesperson for the Census – Not Religious? campaign, said the new polling provided the clearest insight into the extent of the impact of the leading question.
“We’ve long known that the leading nature of the Census religion question significantly inflates the data in favour of religion and undercounts the non-religious population,” he said.
“The Essential Research poll is the first to test the actual question that the ABS had proposed using versus the flawed question design that it decided to re-use. So it provides the best insight yet into the impact of the bias in the question being used in the 2026 Census.”
He said the ABS needed to fix the question for future Censuses.
“Clearly, the question being used for this year’s Census will fail to provide data users – including governments, agencies, elected representatives, academics and researchers, media, community organisations, and members of the public – with an accurate picture of religious and non-religious affiliation in Australia,” he said.
If you have media enquiries, please contact us by email at: admin@censusnotreligious.org.au