The Religion Counts team looks at the relationship between religious identity and the likelihood of voting. 15/05/2024

It has been over 10 years since Theos published its Voting and Values in Britain: Does religion count? report, examining the relationship between religious and political commitments in Britain. As in 2014 so today, religion remains an understudied variable in British political life, and important questions – especially in this General Election year – on the existence and significance of the religious vote remain unanswered. As we have seen in the 2024 local elections, religious voting patterns still play a real part in British democracy.

Theos is exploring in a series of regular articles how different aspects of religion – affiliation, attendance, religiosity – affect the vote. Our research covers voting intentions, likelihood of turnout, attitudes on welfarism vs individualism, attitudes to migration, British identity, amongst other things.

Our main source of data is the British Election Study (BES), a national representative survey which has been conducted at every General Election since 1964, and in particular its 23rd (May 2022) and 25th (May 2023) waves. The BES is “a non–partisan, objective independent study providing… data and research into British general elections”. It is also the UK’s longest–running social science survey.

This first article of our religious voting patterns research series will look at turnout according to religious affiliation, attendance, and self–reported religiosity.

According to the BES, those with no religious affiliation are now in the majority. However, near half the population retain some religious affiliation, and these religious affiliations may influence how they vote in upcoming elections. The table below breaks down wave 25 of the BES by religious affiliation. The BES is consistent with other surveys in terms of its overall breakdown of non–religious versus religious believers.

However, we note that the sample underrepresents religious believers, and in particular religious minorities. Even when adjusted for weight, Muslim voters account for only 1.9% of the total sample in the 25th wave of the BES, despite making up around 6.5% of the total UK population according to the latest 2021 Census. This is a recurring issue in many national surveys of the UK population, which tend to be representative in most aspects, save religious affiliation.

Christians are more likely to vote

It is well–established that religion, “by nurturing civic skills, inculcating moral values, encouraging altruism, and fostering civic recruitment” is a primary source of social capital. [i] Previous studies in the UK have found greater political engagement among religious individuals, but especially among the Christian population, which also affects their turnout in elections. [ii]

In the first instance, we will be looking at turnout likelihood by religious affiliation.

In May 2023, 64% of all respondents said that they would be very likely to vote in a general election tomorrow. This would be roughly in line with recent General Elections. (Turnout was 66.2% in 2015 and 67.3% in 2019.)

Religious voters in general, and Christians in particular, are more likely to vote on polling day than non–religious voters: 82% of religious voters say they are “quite likely” or “very likely” to go compared to 73% of non–religious voters. Controlling for age and education closes the gap between the two groups somewhat, but the same trends remain.

Anglicans (84% very or fairly likely) are the most likely to vote, followed by Other Christians (82%). Despite strong theological differences within the Other Christian (Church of Scotland/Presbyterians, Pentecostal and Evangelical denominations as well as Methodists, Baptists, United Reformed Church, Free Presbyterians, Brethren, and Orthodox Christians) group, most have very similar voting preferences – with the exception of Orthodox Christians. Roman Catholics (78%) come out very close to the national average.

Muslims and other religious voters are the least likely to vote

As of May 2023, Muslim (65%) and other religious voters (70%) are significantly less likely than Christians to turn out on polling day, and fall even below the non–religious voters.

Over a fifth of Muslims (21%) were uncertain about voting, answering “hard to say”, and one fifth of other religious voters (20%) – which includes Jewish, Hindu, Sikh, Buddhist, and other religions – were either “very” or “fairly unlike” to vote, the highest of any group.

The Muslim population is disproportionately younger and from a lower socio–economic class. It also has a large proportion of first–generation immigrants. All of these aspects might contribute to this. That said, it is also important to note that the turnout gap between Muslims and other groups of the population does not close significantly even when controlling for age, gender, and level of education received.

Practicing Christians are even more likely to vote

We have broken down affiliation into an objective measure of attendance at religious services or meetings (communal practice) and into a subjective measure of the significance of spiritual beliefs in one’s life (religiosity).

Christians who are likely to vote in the next election practise their faith more frequently than those who are uncertain or unlikely to vote.

While the difference is more modest for some religious groups than for others, the likelihood of voting among Christians appears to be positively correlated with the frequency of communal practice: 86% of practicing Christians were likely to vote on polling day versus 83% of non–practicing Christians.

For Roman Catholic voters especially, but also for Anglicans, there is an important difference in the average frequency of communal practice between those who are likely to vote and those who are unlikely.

The ‘Other Christian’ group is the exception to this rule. This is mostly due to the small Orthodox Christian population, for whom the relationship is reversed. For most other Christians in that group, the difference in frequency of practice between those likely or not to vote is very small, but average frequency of practice for both groups is much higher than for other Christians.

When asked the more subjective question “How much difference would you say spiritual beliefs make to your life”, the same dynamics can be observed. Of the Christians surveyed, 38% felt that their spiritual beliefs made “some” or “a great difference” to their life (v. 22% frequently practising communally), of the Muslim population, 72% (v. 62%), and of the Other religious, 57% (v. 41%).

The ‘Other Christian’ group are again an interesting exception: both the more and less likely to vote are similarly religious in their self–perception. They are, in addition, also the most devoutly religious of all Christian groups.

Frequently worshipping or more religious Muslims and other religious voters are less likely to vote

These trends are reversed for Muslim and other religious voters, for whom those less likely to vote are on average more frequently practising. Muslims and the other religious combined, 60% of those practising their faith were likely to vote on polling day compared to 88% of the non–practicing population.

This observation should be nuanced by the fact that frequency of communal practice varies widely between Christians and Muslims. Indeed, whereas 62% of Muslims practised frequently (i.e., monthly or weekly), only 22% of Christians did so. It should also be nuanced by the fact that in analysis controlling for age and education, the trend broke down for the other religious, with only a marginally small difference in likelihood to turn out between the weekly practising and never practising. However, the multivariate analysis did not affect the observations on the Muslim population – it is still the case that more frequently practising were less likely to say that they would vote.

Looking at self–reported religiosity for both Muslims and other religious voters, those who are likely to vote are on average less religious than those who are uncertain or unlikely to vote. This mirrors the previous graph looking at frequency of religious practice. However, it should also be noted that the difference in average religiosity between Muslims that are likely to vote and those unlikely to vote is small.

Controlling for age and education has the effect of reversing the trend for both Muslims and followers of other religions: the very religious Muslims appear to be much more likely to vote than the less religious. This is the opposite effect to that we observed in the raw data. However, sample sizes are small. We see that, when controlling for age and education, likelihood to vote in the Muslim sample is still lower than that in other groups.

Conclusions

There is no single general ‘religion effect’ here, but rather multiple different effects. Different religious identities will affect turnout in different ways. Yet there is also no doubt that religion counts – even when demographic variables are taken into account – and should be factored into any discussion on the role of religion in the general election.

The data also makes it clear that religiosity and religious practice, not just affiliation, matters. But even for frequency of practice there is no single rule: while for Christians the likelihood of voting increases with practice, for minority religions, particularly the Muslim community, this relationship is reversed. 

Note about the data

Both the 23rd and 25th waves of the BES data were collected by YouGov via a representative internet panel using an online sample of YouGov panel members. Figures are weighted and percentages are rounded to the nearest unit. Further information on our statistics can be given upon request.

For data looking at affiliation alone, only the 25th wave of the BES is used. When looking at communal practice and self–reported religiosity, data from the 23rd and 25th waves has been combined, keeping only participants who were common to both waves.

Variables were coded so as to be consistent with previous Theos research on religious voting patterns. Predicted and standardised values with control variables (age, gender, education, and religiosity) are available for consultation upon request.