Why does the Census undercount the nonreligious?

The discourse over the use of and problems with nonreligious as a catchall term is extensive. However, it is the term most widely used in surveys, population studies and equality monitoring, so has a lot of uses despite its limitations.

The limits of nonreligious as a unit of analysis can be summarised as: (1) it counts to many people including the ‘spiritual but not religious’, religiously unconcerned and religiously undefined, or (2) it counts too few, because it doesn’t capture people who identify as part of a religious cultural group while holding a largely atheist, humanist, or secularist outlook.

Criticisms of the term’s negative connotations are more wide-ranging. Personally, I have few problems being counted under a catchall term which applies, just because it’s not the main way I’d describe my identity.

Accepting for the moment the use of nonreligious as a category, there are still problems with how we count this. This article takes the British Census as a case study and considers the implications for how society measures and serves the needs of the nonreligious. It will do this by considering six potential ways that the Census dramatically undercount the nonreligious compared to other surveys.

There are two major estimates of Britain’s nonreligious population. Population figures for Northern Ireland are treated quite differently. Since 2001 the British Census, run by the Office for National Statistics (ONS) every decade, has had a voluntary question on “What is your religion”. The British Social Attitudes survey (BSA), run annually by NatCen, asks “Do you regard yourself as belonging to any particular religion?”

Between the 2001 and 2011 Census, those answering “no religion” rose from 14.8% to 25.1% and this trend is expected to continue when the 2021 results are published. Between 1983 and 2019 the BSA showed a steady rise in the nonreligious from 31% to 53%. There are plenty of surveys with different formulation of the question which show a lot of variances, but the BSA figure is a lot closer to the preponderance of these.

Why does the Census matter?

The Census 2021 website says that the question on religion “…helps your community by allowing local and central government to better plan services and set aside resources for your area…” The ONS also say that “To enable them to meet requirements of the Equality Act 2010 users stated a need for population data about religion to inform equality monitoring.”

The Census is a vital data source for public bodies, policy makers and businesses across the country. It also shapes a lot of our discourse around demographics and national identity. The Census is used to prop up the narrative that we are a ‘Christian nation’, justify faith based public schooling, and pastoral support being organised around religion.

This is an area where atheist and humanist groups in many countries have been actively campaigning. The language of the UK Humanist campaign “If you’re not religious, say so” has been particularly influential.

1. Framing

“What is your religion?” frames an assumption of religiosity. It prompts a religious answer, and given the social context of the Census, many recipients may assume they should answer with their ‘official’ or family religion. This is probably the biggest problem, because this framing is replicated in so many equality monitoring procedures. It adds to the perception that the nonreligious should not be a group of concern to equality policies, or that they are a subgroup of religion, rather than the actual protected characteristic of “religion or belief”.

The ONS have tried to address this problem by making “No Religion” the first option – potentially giving it a primacy bias boost. They also tested the question “What is your religion or belief?” but felt that this “changed how respondents thought about the religion aspect of the question leading them to provide answers related to religious belief, rather than affiliation.” And that “Religious affiliation was the concept that most closely aligned to the definition of religion in the equalities legislation.” This privileges only one dimension of (non)religion.

2. Voluntary and different importance

The religion question is one of only three voluntary questions on the Census, the others being sexual orientation and gender identity. When faced with a long survey, the option to skip a question would be appealing. One might reasonably assume that religious people would be more motivated to express a positive answer to the question. Further research however casts significant doubt on this hypothesis. In 2011, 94% of people completed the Census, of these 93% completed the voluntary question on religion. ONS testing ahead of the 2011 Census showed that the response rate was lower than the average for compulsory question, but comparable to questions such as those on employment and ethnicity. In most equality monitoring situations questions will be voluntary, but questions about religion or belief should not be treated as more or less important than any other aspect.

3. A singular question

The Census must be available in paper form, and the ONS say that space restraints mean they cannot include, and so have never tested, a filter or two stage question. The BSA by contrast does not have a paper version and makes use of a filter question “Do you regard yourself as belonging to any particular religion?” before providing additional options for those identifying as belonging to a religion. This could have an affect of encouraging respondents to answer as not religious to get through the survey quicker, but this is likely minimal.

Belief has multiple dimensions. One may believe that a particular worldview is true, or that it is good to believe regardless of its truth. The same is true for identity; a person may have an internal or personal worldview identity, which differs from the external identity they perform. They may culturally affiliate to varying degrees with different (non)religious cultures. Different people place various levels of importance on their (non)religion, it may be central or peripheral to someone’s worldview. A (non)religious worldview may be consciously or subconsciously held.

Some sort of two-part question would probably be the best approach for surveys to adopt, providing the right balance between detail and ease of use. The two parts could either ask respondents about (1) their belief and (2) their identity or affiliation. Or (1) could differentiate between religious and not, and (2) provide more detailed options.

4. Families and children

When thinking of their ‘family religion’ people may be more likely to give a cultural or historic answer. Nonreligious people may not be comfortable or safe self-identifying, where one family member or ‘head of the household’ fills in the form. Younger people are more likely to be nonreligious. The BSA only covers adults 18+, but all respondents are answering for themselves rather than any parents on children’s behalf.

I would be very interested in comparing the (non)religious data from Census responses filled in online or through the corrections and mistakes process, with paper responses.

Any equality monitoring form should only ask about the identity of the individual responding. Where data needs to be gathered on the (non)religious identity of children too young to express a meaningful preference, it might be better to record their (non)religious background.

5. Timing and past responses

The Census is a big cultural event with huge news coverage and advertising. It is likely that people remember how they last answered a question. They may also be thinking about the question as it applies to the last ten years of their life. The nonreligious may be less likely to see this as a settled identity. I’m not sure how much of an impact this could all have, but it is good practice in any monitoring to stress that the question is about the respondent at the time of the question.

6. Ethnicity and culture

There is an anomaly created by some religious categories also being seen by many to be ethnicity categories, though how to resolve this conundrum is a bit beyond this article’s scope and the impact on counting the nonreligious unclear. It certainly could be that nonreligious people who identify as culturally or ethnically Jewish, or Sikh could be under counted. Implication of this include the need to respect people’s self-described ethnic or cultural identity while avoiding conflating this with religion, and the need to differentiate between aspects of religion as discussed above.

Conclusion

These are all issues that the ONS have engaged with, and this article has drawn heavily on their question development reports for 2011 and 2021, as well as contemporary debate in AHS and religious focused media.

Britain’s emerging nonreligious majority is one of the biggest social and demographic changes in our history. We need to consider the impact of this change on almost every area of public policy. To do that, we need to better measure the size and diversity of this group, and those it overlaps with.

Thanks for reading

Let me know what you thought of this article and if you want to hear more. If you value this content and are able to financially support it, that would be great. Just £2 a month would help with hosting and other costs, help improve our content.

Please join our community on Twitter and Facebook to help share the blog, and pass it along to anyone who may be interested.

Photo information: Girl in Yellow Dress Playing With Abacus, Yan Krukov

One thought on “Why does the Census undercount the nonreligious?

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: