Submission: Review of the Australian Standard for the Classification of Religious Groups and Religious Affiliation Standard

This is the Census21 – Not Religious? campaign’s submission to the Australian Bureau of Statistics’ review of the Australian Standard for the Classification of Religious Groups (ASCRG) and Religious Affiliation Standard (RAS).

Public submissions can be made by 18 November 2022 here on the Australian Bureau of Statistics website.

Census21 Group – Recommendations

Who we are

We are a group 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 in order to ensure fair distribution of scarce public resources, for the public good of religious and non-religious Australians.

Overall Context of ASCRG & RAS

We appreciate that a revised ASCRG & RAS will be used as the coding frame for all future ABS surveys where religion is the subject of research, including the 2026 census. We are grateful for this opportunity to provide feedback on the current classification and suggest ways it might be improved.

We are gravely concerned that the ASCRG has become a cumbersome classification and the RAS is conceptually weak. Consequently, we believe that the framework for collecting religious data is in danger of losing sight of its meaning and purpose. The ASCRG has evolved into a classification that serves multiple interests and lobby groups but, in so doing, compromises the presentation of meaningful, coherent data to support the fundamental purpose, i.e., to provide an evidence-base for policy, allocation of public resources, and to deliver a useful data set for social researchers.

Our five major recommendations, with supporting rationale, provide suggestions of how the context and form of the classification could be improved. In addition, we highlight four minor inconsistencies that confuse the interpretation of data drawn from the 2021 census.

Summary of Five Major Recommendations

  1. Re-word the question and insert ‘if any’ as a qualifying suffix
  2. Include the underlying concept of ‘Affiliation’ within the scope of this review
  3. Continue the listing of options ordered by size of response from the previous census
  4. Clarify the definition of “Religion” in the context of the Census (we recommend High Court judgement 1983)
  5. Abandon the ‘operational definition’ of religion

Five Major Recommendations – with supporting rationale

1. Recommendation 1: Question Wording

At a minimum, the question should be modified to, “What religion does the person belong to, if any?”

A preferred approach is to use a two-part question – a) Does the person have a religion? b) What is the person’s religion?


The current question presumes that the person has a religion. This results in a serious form of research bias known as acquiescence bias. While not addressing all deficiencies with the question, adding “if any” is the simplest way to remove some inherent bias in the current question, while maximising the continuity of question wording in relation to longitudinal data and analysis.

We note that many respected social surveys (e.g., the Australian Survey of Social Attitudes) already include the rider “if any”, or a similar conditional clause, as part of the published questions.

We also note that the Irish census, conducted in March 2022, has also adopted this qualifier, in line with best contemporary research practice.

2. Recommendation 2: Religious ‘Affiliation’

The underlying concept of ‘Affiliation’ should be considered ‘in-scope’ for the review of the ASCRG and RAS and replaced with a concept that is less ambiguous and more usefully aligned to the purpose of collecting data on religion – i.e., to provide an evidence-base to support policy, funding, and representation decisions.


We note that the censuses conducted in Scotland and Northern Ireland in 2021 used the question, “What religion, religious denomination or body do you belong to?”

The wording in this and other jurisdictions indicates a rejection of the weak and poorly understood concept of ‘Affiliation’.

Also, please see Recommendation 5.

3. Recommendation 3: Categories on Census Form

Continue to list the top ten categories in declining order of response from the previous census.


We understand that the 2021 results and a revised classification may influence the identity and sequencing of pick-boxes listed on the census form. We support the concept of listing the top ten categories in declining order of the numbers of responses from the 2021 data, with ‘No Religion’ as the first option.

Such a listing is also consistent with most internationally comparable census organisations.

4. Recommendation 4: Definition of Religion

The classification of religions should start and end with the definition of what comprises a religion. Any belief system not conforming to the definition of religion should not be represented in the classification.


The best guidance for a definition of religion, as the ABS itself states in several documents, is the reference to the 1983 High Court decision where it set out such a definition:

The criteria of religion are twofold:

  • Belief in a Supernatural Being, Thing or Principle;
  • The acceptance of canons of conduct which give effect to that belief.

The classification seems to depart from this clear definition by including many categories that are clearly inconsistent.

Apart from ‘No Religion’, which we accept as a necessary category, all other categories of broad group 7 do not qualify. All categories in this group should be subsumed into the single code, ‘No Religion’.

It is an egregious misclassification to suggest that Atheism (2021 question) or Humanism (2016 question) should be considered as examples of a religion or a religious ‘affiliation’. Atheism is the denial of supernatural beliefs and, therefore, religion.

Atheism, Humanism and Rationalism are philosophies that have nothing to do with religion and should not be suggested as examples of a valid response. Similarly, “Secular Beliefs” has no place in a classification of religions. It is primarily a term that is used to describe a view about the role of religion in government decision-making. It makes no logical sense as a religious category because it is a term that could equally apply to religious people as well as to non-religious people.

As it stands, significant proportions of those who are atheist, agnostic, humanist, rationalist or of other non-religious philosophy will simply pick the “No Religion” box. This renders the count of those who wrote in a specific non-religious answer as, at best invalid, and at worst, a misleading representation of those sub-groups.

The trajectory the classification has taken since its inception in 1996, means that a clear definition of a religion has become conflated and confused with vague notions of ‘spiritual belief’ or ‘world view’. In most cases, these “beliefs” or “world views” are not, in fact, religion and should not be counted as such.

Data collection on world views, philosophies and spirituality may be interesting and useful in the domain of social research but has no meaningful place in a definition of religion in the census, given that the results are a key justification for the allocation of constrained public resources.

5. Recommendation 5: Operational Definition of Religion

The ‘Operational Definition’ of a religion should be abandoned.


The ABS attempts to differentiate between a ‘Nominal Definition’ and an ‘Operational Definition’. We see this differentiation as an arbitrary construct that fails the test of intellectual rigour required for measurement integrity.

We see ‘Affiliation’, as used in the Operational Definition, as a weak and poorly defined concept that makes it impossible to control how respondents may interpret the word – and hence the quality of the data produced.

In interpreting the question, respondents will most likely be prompted to recall any event from their childhood or background – e.g., baptism, parental guidance, ancestry, school experience – which probably has no bearing on their recent past, their present, or their future connection to a religion that meets the High Court definition.

Further, many respondents may think that they have a loose ‘affiliation’ with multiple religions/beliefs/world views. With the benefit of an informed study, there is much to admire about the positive values from, for example, Sikhism, Islam, Buddhism, Christianity, and non-religious philosophies. Many informed people could claim an affiliation with more than one of these, but the ASCRG does not currently accommodate the broad and ill-defined concept of affiliation with multiple religions. And, because of the complexity in collecting the data and interpreting the results, it is doubtful whether it should. This serves to illustrate the weakness of ‘affiliation’.

But the key question the ABS should ask itself is, “Why is the data being collected?”. A weakly defined ‘Affiliation’ is an insufficient justification to base entitlement to substantial automatic tax concessions and the ‘by default’, continued allocation of multiple billions of dollars of public resources to religious organisations across health, education, caring and media services. Similarly, less ambiguity will produce more meaningful social research.

Four Minor Recommendations with supporting rationale

6. Recommendation 6

In relation to the Detailed and Narrow Group (4-digit and 2/3-digit) level of reporting the suffix ‘so described’ should be removed from ‘No Religion, so described’ category.


The pick-box on the form is unambiguously ‘No Religion’. No other categories (e.g., Catholic, Anglican or Baptist) carry a suffix.

7. Recommendation 7

In relation to the Detailed (4-digit) level of reporting the category ‘Secular Beliefs and Other Spiritual Beliefs and No Religious Affiliation, nfd’, even with its small numbers, is inappropriate.


The category labelled ‘No Religious Affiliation, nfd’ does not exist, so it should not be included in the category label. If an obscure or unclear response does not conform to the definition of a religion, it should be counted as ‘No Religion’.

8. Recommendation 8

In relation to the Narrow Group (2 or 3-digit) level of reporting, remove ‘Comprises ‘No Religious Affiliation, nfd’ from the footnote (c) used for the ‘Other Spiritual Beliefs’ category.


The category labelled ‘No Religious Affiliation, nfd’ does not exist, so it should not be included in the footnote. For accuracy, the label should be ‘Other Secular or Spiritual Beliefs, nfd’. However, please also refer to our comments on Secular Beliefs in Recommendation 4.

9. Recommendation 9

In relation to the Broad Group (1-digit) level of reporting, the label for broad group 7, “Secular Beliefs and Other Spiritual Beliefs and No Religious Affiliation”, is unwieldy and inappropriate. The label should simply be ‘No Religion’.

At the very least, the group should be re-labelled “No Religion, Secular Beliefs and Other Spiritual Beliefs”, reflecting the weight of responses from past censuses. But also refer to our comments on “Secular Beliefs” in Recommendation 4.


Including ‘No Religious Affiliation’ as an add-on to a long and cumbersome label diminishes the significance of the 98.8% of 2021 responses within this group who chose ‘No Religion’.

Before 2016, the ‘No Religion’ label was used for the broad group 7, including the secular and spiritual categories. While appreciating the need for that to change, the current assemblage is equally unsatisfactory. ‘No Religion’ is, by a long way, the largest single detailed response category of the whole classification. It should be unambiguously identified at all levels of reporting, rather than buried within a large number of categories which may or may not be broadly related.