The problem with the census question

17 June 2021

In parliaments across Australia and in the media, we often hear the claim that Australians remain largely religious people.

“Look at the 2016 census,” the argument goes. But therein lies the problem.

Many point to the census figure of 60% of Australians having a religious affiliation as a reliable measure for religiosity, but it is not. This figure is fundamentally flawed – and for a number of reasons.

To begin with, the question asked in the census is biased in favour of producing a response in favour of religious affiliation.

As pointed out by Neil Francis, Fellow of the Rationalist Society of Australia and author of the new-published Religiosity in Australia report, the current phrasing of the question – ‘What is the person’s religion?‘ – presumes all people have a religion.

“The question really ought to say: ‘What is the person’s religion, if any?’ That is actually a balanced survey question,” he says.

“So people who really don’t have religious beliefs but they have some family history – for example, their family has been Anglican, Catholic or whatever – will tend to tick the box because they’ve been asked what their religion is.”

In his report, Francis argues that adding ‘if any’ would relieve the pressure on people to mark a religious denomination and also avoid the confusion of listing ‘No religion’ as a religion to choose from.

Another problem relates to the context in which the question is asked, following previous questions that ask about the person’s culture.

These built-in biases lead to the overstatement of meaningful religious commitment, as many people answer the question based on weak family historical grounds rather than actual religious belief and practice.

Francis highlights the results from other surveys to illustrate how the difference in wording can have a major impact.

The Australian Values Survey of 2018, for example, asked, ‘Do you belong to a religion or religious denomination? If yes, which one?’ This study returned a ‘No religion’ result of at least 11 percentage points higher than other studies which had presumptive wording, including the Census and the Australian Election Study.

In the United Kingdom, the British Social Attitudes Survey of 2018 asked, ‘Do you regard yourself as belonging to any particular religion?’ The ‘No religion’ result there was 13% higher than what was indicated in the Office of National Statistics study, which used the biased wording of, ‘What is your religion?’

Although the use of religious denominations in the census remains a “crude and unsophisticated” measure, Francis notes that, at a minimum, it provides a simple headline figure that can be tracked over time.

Since ‘No religion’ became a possible response in 1971, the proportion of Australians reporting no religious affiliation has continued to grow while both major Christian denominations (Anglican, Catholic) and minor Christian faiths have lost a significant proportion of their flocks.