Data collection and publication - religion/belief: guidance

Analysis and presentation of results

Because there are several dimensions to religion, talking and reading about religion or specific religions without qualification can be confusing and lead to misinterpretation by data users. Therefore, if presenting data from the harmonised religion question it is important to be explicit and refer to the specific 'concept' being measured i.e. religious belonging. It is also recommended that when presenting data on religious belonging it should be accompanied by a short note, such as:

Respondents were asked the question 'what religion, religious denomination or body do you belong to?' which measures belonging – that is both loose self-identification and active or formal belonging to a religious group

Data should be analysed and presented in a way that is most useful to users of the data, with consideration of what kind of analysis would be most likely to reveal any inequalities that require action.

Small numbers can cause problems with statistical reliability when analysing data and hinder publication of figures because of the need to avoid identification of individuals. It is important that the results should be presented in as much detail as possible. It is better to list all of the categories and where possible the figures. If the numbers are too small to publish then the figures for that category should be suppressed and an explanation provided of why this has been done. Care should be taken to ensure that if only one figure is suppressed, that the figure can then not be calculated by simple subtraction from the totals displayed. Alternatively, consideration should be given to increase sample size by aggregating several years of data where appropriate.

Another way to overcome the problem of small numbers is to aggregate data together under the headings 'Christian' and 'All Other religions' but this may not be an ideal solution as it can hide inequalities that are occurring between the groups under each heading. For example the 2011 Census showed that there were differences in the unemployment rates between Buddhists, Hindus and Muslims^[9] and the latest figures from the Scottish Surveys Core Questions (SSQC 2019) shows differences in employment rates between Roman Catholics and members of the Church of Scotland.^[10] However such aggregations may sometimes be necessary in order to present any data. It is recommended that the term 'non-Christian' is not to be used as it is offensive to some people.

Comparing the 2022 census with the 2011 and 2001 censuses

It is possible to compare data from the new 2022 question with data from the 2011 and the 2001 Censuses, but it must be noted that Pagans will have recorded their religion in the 'Other' write in box for the Census. When comparing data from the new question with data from the 2011 and 2001 Census questions this should be noted as numbers under the 'Other' heading may be reduced as a result.

Alternative questions on religion

A religion question on affiliation was asked in the 2021, 2011 and 2001 Censuses of England and Wales. This is the recommended religion question and layout for use on a survey in Scotland when wanting to harmonise with the rest of GB and the UK, or if information on religious affiliation is required, the following question should be asked:

Question: What is your religion?

• 1. None
• 2. Church of Scotland
• 3. Roman Catholic
• 4. Other Christian, please write in
• 5. Muslim, write in denomination or school
• 6. Hindu
• 7. Buddhist
• 8. Sikh
• 9. Jewish
• 10. Pagan
• 11. Another religion or body, please write in
• 12 Refusal (spontaneous only)
• 13. Prefer not to say (non-interviewer led questionnaires only)

The recommended breakdown for Christian denominations for Scotland (like Northern Ireland) is different to that of England and Wales. This is in order to provide data on which to examine the differences between the main Christian groups. Consultation by the General Register Office for Scotland revealed limited user demand for questions on active practise, belief (including non-religious belief) and religion of upbringing for the 2022 and 2011 Census.^[11]

Intersectionality

An intersectional analysis gives an insight into how a combination of socio-demographic characteristics might relate to specific forms of disadvantage. For example, disadvantage for a minority ethnic, Christian woman may be different to a white, Christian man.

There is an increasing awareness that taking an intersectional approach to research, policy making and operational decisions is important. This is because intersectionality can give insight into the experiences of different groups in society, and how particular characteristics can combine to impact on an individual's experiences. However, there is not always disaggregated data available to support such an approach. This is likely to mean that the information on which important decisions are made is not fully representative of the population it is intending to measure.

For example, data on demographic characteristics may be collected by a public body, but not disaggregated in an intersectional way due to issues around sample size and risk of disclosing an individual's identity. In these cases, organisations should not risk disclosing information about an individual's age.

Where sample size and quality allows, data should be disaggregated, including by sex, gender reassignment, race, religion or belief, age, disability, and sexual orientation, where combinations of these factors can result in discrimination, disadvantage and inequality. Being able to identify cases where combinations of factors are resulting in disadvantage enables policies to be developed and action taken to address these issues.