1. Other pages in this release

Statistician's comments

“This is part of a programme of work we are doing to explore inequalities in our society. The reasons for inequalities are complex, as today’s findings show, with a range of factors to be taken into account. These findings are not intended to provide definitive answers but to add to the growing evidence base on equalities.”

Paola Serafino, Centre for Equalities and Inclusion, Office for National Statistics

Back to table of contents

2. Background

In 2017, the Office for National Statistics’s (ONS’s) Centre for Equalities and Inclusion began an audit of equalities data to identify the sources of data available to understand the experiences of people in the UK across the nine protected characteristics covered by the Equality Act 2010.1 The audit aimed to highlight where gaps exist in the quality and coverage of equalities statistics and was a starting point to take forward work with others to prioritise and fill the gaps.

The audit identified approximately 60 sources of data from official surveys, other government-funded surveys and administrative data that include information on religion. As part of planned work following on from this, the Centre convened a group of representatives from across government to explore these data sources and establish the extent to which they could be used to describe the experiences of people of different religious groups in England and Wales.

When convening the group to explore the data on religion, all the devolved administrations were invited to participate and the Welsh Government accepted this invitation. The Centre is grateful to the analysts from a range of government departments and agencies, Welsh Government and the Equality and Human Rights Commission, who have worked with us on this.

This exploration of the data was organised around the domains defined in the Equality and Human Rights Commission measurement framework (PDF, 15.66MB), including areas of life that are important to people and enable them to flourish. The remaining articles of this release explore outcomes for people of different religious identities across the domains of justice and personal security, work, education, health and participation.2

One of the Centre’s aims is to improve the evidence base particularly for groups that may be invisible in routine reporting of statistics, for example, because they are present in insufficient numbers in sample surveys for reliable estimates to be provided. As a result, the focus of this work was to capture the full range of religious groups contained within the Government Statistical Service (GSS) harmonised principle on religion, not just those that have the largest numbers. For England and Wales, the religious groups are:

  • No religion
  • Christian
  • Buddhist
  • Hindu
  • Jewish
  • Muslim
  • Sikh
  • Any other religion

Only statistics that can be presented across most or all of these religious groups are included in this release.

This is the first phase of a longer programme of work in which we will work with others to explore options for improving the data available on religion. As a starting point, we have considered the quality of the data in detail, as well as where we have information and where it is lacking. We also provide indications of possible differences between groups in the other sections of this release and plans for the next phases of work.

Notes for: Background

  1. The Equality Act applies in England, Wales and Scotland and defines the following as protected characteristics: age; disability; gender reassignment; marriage and civil partnership; pregnancy and maternity; race; religion or belief; sex; and sexual orientation. Although there is some overlap with the protected characteristics in the Equality Act, separate legislation applies in Northern Ireland.

  2. The articles within this release do not coincide fully with the domains in the measurement framework, reflecting the statistics that it has been possible to present.

Back to table of contents

3. Definition of religion

The Government Statistical Service (GSS) harmonised principle on religion recommends that, where a single question is used in data collection, the concept that should be measured is religious affiliation. The 2021 Census: Assessment of initial user requirements on content for England and Wales: Religion topic report (PDF, 780KB) defines religious affiliation as “how respondents connect or identify with a religion, irrespective of whether they actively practise it”.

In line with this, estimates presented in this release capture the concept of religious affiliation. However, although the 2021 Census topic consultation identified strong user need for data on religious affiliation, there was also evidence of demand for data covering religious beliefs and practices. The GSS Harmonisation Team plan to conduct an implementation review of the religion principle to identify how it is being used across government. The outcome of this review will inform future work in this area, which may include additional questions to measure concepts such as belief and practice.

Back to table of contents

4. Things you need to know about this release

The statistics presented are estimates and as with all estimates, there is a level of uncertainty associated with them. Where available, 95% confidence intervals have been shown. These indicate the range within which we would expect the true value to lie for 95 out of every 100 samples drawn at random from the population. Wide confidence intervals, often associated with small sample sizes or large sample variance, indicate a wider range of values within which we would expect the true value to lie.

Throughout this release we have assessed statistical significance using non-overlapping confidence intervals. This method has the limitation that some estimates with overlapping confidence intervals may be significantly different but will not be identified as such (that is, the false-negative rate will be inflated). In addition, no adjustments have been made for multiple comparisons.

In many cases, sample sizes for specific religious groups are small and confidence intervals are large and overlap with one another. This makes it difficult to make robust comparisons between groups. Only statistically significant differences (as defined in each part of the release) are commented on in this release. Caution should therefore be exercised when making other comparisons between religious groupings as observed differences may not be statistically significant.

The 2011 Census question on religion was voluntary and just over 7% of the population of England and Wales opted not to answer it, equivalent to just over 4 million people in total. If some religious groups were more likely than others not to respond to this question, then the census data may not present a true picture of these groups, though the voluntary nature of the question is an important principle in taking a human rights-based approach to data collection (PDF, 292KB), allowing individuals to choose whether to disclose their identity.

In line with the 2011 Census, questions in all surveys relating to religion are voluntary and respondents can opt not to reveal their religious affiliation. Throughout this release, we have assumed that the distribution of outcomes of non-respondents in the different religious groups is similar to that of those who did respond. However, if this assumption does not hold, this could affect the results presented.

Back to table of contents

5. Size of the religious populations

The most up-to-date official estimates of the population identifying with the different religions in England and Wales are available from the census, which was last carried out in 2011. Estimates are also available for Wales from the Annual Population Survey (APS), though these do not capture the full range of religious groups.

The ONS has been exploring a method for providing more up-to-date estimates using the APS, but these are currently just illustrative estimates and we are actively seeking feedback on both the method and the usefulness of these estimates. The method adjusts the APS estimates (which exclude most people living in communal establishments) so that they cover the entire population and are consistent with the mid-year population estimates.

This research has shown that at the national level for England, applying the method provides a distribution of religious affiliation similar to the census. There is a decline for the “Christian” group, counteracted by higher proportions for all the other groups, with the largest increases seen for the “Muslim”, “None plus Not stated” and “Other” groups. The quality of estimates produced by this method for local and unitary authorities (LAs and UAs) is less clear. The completeness at LA and UA level does not currently appear good enough for us to recommend its use.

The advantage of the census is that estimates are available at a greater level of granularity than is achievable with household surveys, allowing statistics to be presented for England and Wales separately, disaggregated by a variety of characteristics. The census also has more complete coverage of the population, including individuals living in communal establishments, and children, although it is likely to be adults in the household who report on behalf of younger children. Both groups are generally out of scope for surveys of private households, on which many official statistics are based, although there are some surveys that are specifically targeted at children, including several cohort studies and schools-based surveys like the Longitudinal Study of Young People in England.

Balanced against these advantages is the limitation that census data are only updated every 10 years, providing a snapshot at a particular moment in time, and the population may change considerably between censuses.

Religious affiliation

Across England and Wales in 2011, the profile of religious affiliation was skewed, meaning there were a few large groups and several much smaller ones (Figure 1). The religion that the largest proportion of the populations in both England and Wales identified with was Christianity (59% and 58% respectively). Almost a third of the population of Wales (32%) and a quarter of the population of England (25%) did not identify with any religion.

Figure 2 shows the proportion of the populations of England and Wales who identified with minority religions (that is, not identifying as Christian or with no religion) in England and Wales in 2011. While 1 in 20 (almost 2.7 million) people in England identified as Muslim (5.0%), only 1.5% of people in Wales (just under 46,000), identified in this way.

Those who identified as “any other religion” made up the smallest part of the population of England (0.4% or almost 228,000), while in Wales, this was the case for those who identified as Jewish (0.1% or just over 2,000).

Religion affiliation by age group

The age structure of the population of England and Wales in the different religious groupings in 2011 is shown in Figures 3 and 4, respectively. For both countries, a much greater percentage of those who identified as Muslim were in the younger age groups compared with other religions, and a much greater percentage of those who identified as Jewish or Christian were in the older age groups (50 years and over) compared with other religions.

In England, a third of those who identified as Muslim were under 16 years old (33%) and a similar proportion were also in this age group in Wales (32%). Around 4 in 10 of those who identified as Christian (43%) or Jewish (40%) were aged 50 years and over in England. In Wales, around half of those who identified as Christian or Jewish were aged 50 years or older (48% and 50% respectively).

Back to table of contents

6. Attendance at religious services or meetings

As religious affiliation is the concept that the Government Statistical Service (GSS) harmonised principles recommend be captured in routine data collection, there is a breadth of information available in relation to this. In addition to this, it is also interesting to consider religious practice, to explore the extent to which identity and behaviour align.

Figure 5 shows the percentage of adults in England and Wales who reported that they regularly attended religious services or meetings (once a month or more) in 2016 to 2018. The highest rate of regular attendance was among those who identified as Sikh (75%). Those who identified as Christian were less likely than average to regularly attend a religious service or meeting (29%).

Analysis of the breakdown of the England and Wales population by religious affiliation in the Understanding Society, UK Household Longitudinal Study (UKHLS) revealed a higher proportion of the population reporting that they do not identify with any religion compared with the Census. This may have an effect on the figures presented in relation to religious practice.

Back to table of contents

7. Data strengths and limitations

Our exploration of the existing data sources has shown that statistics exist to describe the experiences of people of different religious affiliations across a range of areas of life. These come mainly from surveys, though some administrative data are also available (see the Equalities data audit for details of sources including information on religion). However, in none of these areas is there a comprehensive picture of outcomes and experiences across all religious groups.

While some of the limitations relate to a lack of any data on certain outcomes, the most obvious limitation relates to the sample sizes for the religious minority groups, when considering most sources other than the census or administrative data.

Reflecting the size of these populations in England and Wales as a whole, in many cases sample sizes for specific religious groups in surveys are small. Therefore, we have been unable to provide estimates that are robust enough to compare all the different religious groups for England and Wales separately. Given that many existing sources have insufficient sample sizes to provide robust comparisons at the country level, further geographical disaggregation is not possible for any but the largest groups.

Similarly, our ability to explore intersectionality is also limited. Intersectionality refers to the differing experiences of people based on their status in relation to multiple characteristics, for example, a woman with a disability and a specific religious affiliation in a particular socio-economic group.

However, despite these limitations, the data do provide the opportunity to undertake other analytical work. In particular they offer the opportunity to look at religious affiliation alongside other characteristics that may affect outcomes. We have included examples of these in other pages of this release, but there is scope for this work to be extended to consider a wider range of outcomes and to take a wider range of characteristics into account.

Back to table of contents

8. Improving the data

We are increasingly turning to administrative data to address some of the limitations of our survey data sources. As we do so, it becomes more important that consideration is given to including a greater breadth of information about the people to which it relates, while always recognising that this aim must not distract from its primary purpose in administering services.

Key to its use in this way is embedding the human rights-based approach to data collection (PDF, 292KB), which stipulates that people self-identify in relation to their characteristics, including religious affiliation. While this is an issue for all data collection, it needs to be explored carefully in relation to administrative data sources, gathered originally for non-research purposes, where other practices may apply.

A number of initiatives are planned that have the potential to address a specific limitation or gap in the existing data in the areas of life where data are most lacking. Some of these rely on linking data sources to provide larger samples of data on relatively small populations, often linking census to administrative data.


Currently, the availability of data exploring the educational outcomes of people of different religious identities is limited. This is, in part, because some of the main sources of data on educational attainment do not currently capture information on religious affiliation. Even where data are available, they are often not sufficiently detailed to allow for detailed geographical or intersectional analysis.

To address this gap, the Office for National Statistics’s (ONS’s) Centre for Equalities and Inclusion is currently exploring the potential for a new linked dataset called “Data for Children”, to be used to deliver fresh insights into the relationship between individual characteristics, family background, geography and educational attainment in England. This work is being informed by a working group consisting of representatives from across government, academia and the third sector.

The Data for Children proof of concept dataset links Census 2011 to an extract of the English National Pupil Database. The ongoing development of these linked data is being led by a partnership between the Office of the Children’s Commissioner and Admin Data Research (ADR) UK. The ONS will publish the results of the feasibility analysis during 2020.

The requirements for future iterations of the dataset centre around the ability to identify different types of vulnerability and interaction between characteristics. Improvements to the data during 2020 will focus on the linkage methodology and expanding the information available to include the following:

  • Higher Education Statistics Agency (HESA) student record information from the All Education Dataset for England (AEDE)
  • Individualised Learner Record (ILR) records from AEDE
  • School-level information from Ofsted
  • Higher education intentions information from Universities and Colleges Admissions Service (UCAS)
  • Children in Need data from Department for Education (DfE)
  • Children Looked After data from DfE

This range of data will allow for a more complete longitudinal picture of educational experience, keeping the population recorded in Census 2011 as the base population to which information from other sources will be linked. This could potentially support analysis of religious identity from the census across the domains of school attainment, further education, higher education and requirements for state support.

Subject to the results of the feasibility analysis and the availability of these future iterations of the dataset, the Centre for Equalities and Inclusion will work with others to conduct analysis and publish relevant findings.

A similar project is currently being explored by ADR Wales.


Timely and robust objective and subjective health measures by religious affiliation are also currently lacking. The ONS is undertaking a feasibility study to model health state prevalence estimates for use in improving the estimates of health state life expectancies in England by using the relationships found in linked Hospital Episode Statistics (HES), mortality and 2011 Census records. The aim is to improve the accuracy and granularity of health state life expectancy statistics, allowing improved estimates at the local authority level, and in turn improve local public health decision-making.

While the current research aims of this project are specific to improving estimates of health state prevalence, initiatives such as this offer the opportunity to investigate how gaps in the evidence on health by religious affiliation could be addressed.


There are also gaps in the data coverage relating to religion and crime.

The ONS’s Centre for Crime and Justice are considering the creation of a combined three-year dataset using the latest Crime Survey for England and Wales (CSEW) data based on a new methodology (for more information see Improving estimates of repeat victimisation derived from the Crime Survey for England and Wales). The project will investigate the feasibility of providing more information on religion in the future, aiming to cover all religious groups and enable more granular analysis of issues such as:

  • type of crime experienced by victims of different religions (for example, violence with injury, violence without injury, robbery and theft offences, and fraud)
  • experience of types of domestic abuse experienced by victims of different religions
  • religiously-motivated hate crime experienced by different religious groups
  • all hate crime strands experienced by each religious group, potentially exploring the relationship between crime, religious belief and other characteristics such as age, ethnicity and where a person lives

Living standards

The Department for Work and Pensions is currently exploring the potential for religious breakdowns with the view to publish this information by Universal Credit claimants, as part of their regular official statistics. They hope this information will be made available to users by summer 2020.

Back to table of contents

9. Next steps

This work has focused on the extent to which we can compare the life experiences of people across different religious groups as a starting point for a broader programme of work to address the limitations and gaps in the evidence base.

The Centre for Equalities and Inclusion will continue to work with others to improve the data available to explore the outcomes of people with different religious affiliations. Updates on progress will be published on our website and shared with interested stakeholders via our newsletter.

Back to table of contents

Contact details for this Article

Paola Serafino
Telephone: +44 (0)1633 651538