Census21 campaign responds to church leaders’ call for government interference

The Census21 – Not Religious? campaign has raised concerns with Prime Minister Anthony Albanese about the Catholic Church’s calls for the government to interfere in the work of reformulating the Census religion question.

In a letter to Mr Albanese and Assistant Minister Andrew Leigh (see below), our spokesperson Michael Dove said the Census21 campaign supported the Australian Bureau of Statistics’ (ABS) testing of an alternative that would “undoubtedly increase the accuracy and meaningfulness of Census data”.

In recent weeks, Catholic Church leaders have been lobbying the Prime Minister to reject the proposed change to the religion question that would introduce a filter question and remove the inherent bias in the text of the question.

In The Australian last month, former prime minister John Howard backed the church, accusing the public service of having an “anti-religious lobby” and wanting religion removed from the Census.

This week, Crikey published an article with comments from our campaign.

The ABS is testing a re-formulated question following a public feedback process that identified, as a major concern, the bias in the existing question.

The ABS is proposing to change the question from “What is the person’s religion?” to a two-part question – a) “Does the person have a religion?”, and, for those indicating ‘Yes’, b) a space where the name of the religion can be written.

In media reports, church leaders have claimed that changing the question would “weaken the accuracy” of the data.

The ABS has since issued a clarification about its proposed changes to the religion question and correct media reporting.

In the letter to Mr Albanese and Dr Leigh, the minister responsible for the Census, Mr Dove said the Catholic Church’s concerns were misplaced and would work against the interests of all users of Census results.

“This proposed change is designed to correct a fundamental flaw that has persisted for far too long and has artificially distorted the results of multiple censuses,” he said.

“The wording of the existing question fails to meet the most elementary standards of questionnaire design in social research. This is because the question presumes the respondent has a religion.

“This causes an error known as ‘acquiescence bias’, where the respondent is inclined to respond with a religion because that is what the question asks them to do. This has produced a bias in the results causing the number of people of religion to be overstated.”

Mr Dove urged Mr Albanese and Dr Leigh not to let political interference undermine the work of the ABS.

“…we are strongly supportive of the ABS testing an alternative that will undoubtedly increase the accuracy and meaningfulness of census data,” he said.

“Please do not undermine their expertise by allowing political interference in the census process.”

Letter to the Prime Minister Anthony Albanese and Assistant Minister Andrew Leigh, 24 May 2024

Dear Prime Minister and Mr Leigh,

I’m writing to you on behalf of the Census21 Group. The Census21 Group is a collaboration of secular and other organisations comprising Atheist Foundation of Australia, Sydney Atheists, Rationalist Society of Australia, Humanists Australia, National Secular Lobby and Humanists Victoria.

We have a common interest in accurate and meaningful government statistics to ensure fair distribution of scarce public resources, for the public good of religious and non-religious Australians.

We were troubled to see media reports in recent weeks about how Catholic Church leaders have been lobbying the government to reject the Australian Bureau of Statistics’ (ABS) proposed change to the religion question.

As is normal practice after each census, the ABS conducted a thorough review and consultation process in the period following the release of data in 2022. During that process, the ABS received almost 200 submissions from organisations and individuals who felt that the religion question produced a result that was biased in favour of religion.

The claims by Catholic Church leaders suggested that the proposed changes will “weaken the accuracy” of religion data from the census.

The concerns raised centre on two aspects.

1. Changing the question from “What is the person’s religion?” to a two-part question – a) “Does the person have a religion?”, and, for those indicating ‘Yes’, b) a space where the name of the religion can be written.
    
2. A move away from the listing of ten tick-boxes. The boxes were sequenced in order of the number of responses given in the preceding census, with a space for writing in the name of the religion if it was not included as a tick-box. The ABS is proposing to replace the boxes with a free-form space where people self-describe the identity of their religion, rather than being guided by a list.

We believe these concerns are misplaced and will work against the interests of all users of census results.

The first aspect of their concern is about an ABS proposal to make a small but important change to the wording of the question on religion. This proposed change is designed to correct a fundamental flaw that has persisted for far too long and has artificially distorted the results of multiple censuses.

The wording of the existing question fails to meet the most elementary standards of questionnaire design in social research. This is because the question presumes the respondent has a religion. This causes an error known as ‘acquiescence bias’, where the respondent is inclined to respond with a religion because that is what the question asks them to do.

This has produced a bias in the results causing the number of people of religion to be overstated. Given the importance of census data in informing policy, resource allocation and share of voice in the media, to not change the question would simply perpetuate the error and continue the distortion of results.

Having reliable and accurate data should outweigh any minor concern about comparability with previous census results. Best practice should dictate that changing the question is an imperative.

The second aspect of church leaders’ concerns relates to the ABS intent to test whether a free-form response, for those who indicate they do have a religion, will produce better quality data than produced by the tick boxes.

As a group, we do not have a strong view on this proposed change, but we do understand and respect the reasons why the ABS is choosing to test this alternative.

In the 2021 census, at the most detailed level, the ABS reported on 131 categories of response to the religion question. Due to a revision in the standard classification, there will be 139 categories in the 2026 census.

Clearly, there is a practical limit to the number of tick-boxes that can appear on a census form, and it is our understanding that the ABS is seeking to properly test the impact of moving to a completely free-form response.

In the public consultation phase of 2026 census planning, many organisations expressed concern that the tick-boxes created their own problems. For example, the tick-boxes of Catholic, Islam, Buddhism, and Greek Orthodox caused the identity of certain sub-groups of major religions to be invisible or under-counted.

In the case of ‘Catholic’, sub-groups such as Maronite, Chaldean and Syriac were probably under-counted. This was also true with sub-groups of Orthodox Christianity which caused considerable cultural tension because Greek was the only visible tick-box representative. Similarly with Islam and Buddhism, along with more than 80 sub-groups of other Christian religions.

It is easy to understand why organisations like the Catholic Church might feel worried about religious trends in contemporary Australia. But vested interests in perpetuating sub-standard data collection due to defective wording of the question should not be permitted to hold sway in this important area.

We note that the ABS has issued statements of clarification to correct some errors and incomplete information in recent reporting about the proposed testing and we support the content of those statements.

In summary, we are strongly supportive of the ABS testing an alternative that will undoubtedly increase the accuracy and meaningfulness of census data. Please do not undermine their expertise by allowing political interference in the census process.

Yours sincerely,

Michael Dove
Chair, Census21 Group

Media inquiries: [email protected]