The post-2015 British Election Study ran from May to September last year surveying a random sample of the British public regarding their vote choice in May 2015. It also asked broader questions including attitudes towards the EU and whether respondents would vote Leave or Remain in a future referendum. Demographic data, including ethnicity and religious affiliation and attendance, were also gathered in the study. Here Siobhan McAndrew analyses this data, finding a notable difference in voting between Anglicans and members of other faith groups, as well as how an English identity (or its manifestation as distinct from Britishness) has a strong association with referendum voting intentions.

The United Kingdom EU referendum result is of particular interest for those working in the area of religion, ethnicity and values. Early analyses from both academics and public commentators alike have noted the importance of identity and values for attitudes towards the EU, and indeed towards politics in general. I have written elsewhere on how the result can be understood both in terms of value differences between social generations, and in terms of national identity, particularly Englishness.

Older age groups tended to vote Leave, and a majority in England and Wales voted Leave while majorities in Scotland and Northern Ireland voted Remain. Early polling evidence also suggests that Christians were more likely to vote Leave than members of other faith groups and none.

My colleague Ben Clements has analysed European Social Survey data for the British Religion in Numbers website, and noted differences across religious tradition in attitudes to the EU. The patterns are suggestive that Roman Catholics, non-Christians and Nones have warmer attitudes than Anglicans and other Protestants. This may be because Catholics in Britain often have Irish, Polish, Italian or other heritage, and are also more accustomed to thinking in terms of religious commonality across Europe. Religious Nones tend to be younger, and potentially therefore to think of themselves in more cosmopolitan terms, although this is arguably a generational effect rather than a secularity effect. The extent to which lower Anglican affinity for the EU is down to age or generational effects (since there has been a strong generational drift away from Anglicanism since the Second World War or longer) or down to something inherent in the religious identity itself needs further probing.


We can look for answers from various sources. The British Election Study (BES) team is currently running a post-referendum survey to ask people how they actually voted. However, we do have the post-2015 British Election Study available for analysis now. This ran over May-September last year, surveying a random sample of the British public regarding their vote choice in May 2015, as well as many other questions on broader political attitudes and behaviours. 2987 people were surveyed via a face-to-face interview. Questions included items on their attitudes towards the EU and whether they would vote Leave or Remain in a future referendum. Demographic data, including ethnicity and religious affiliation and attendance, were also gathered.

Respondents were asked,

If there was a referendum on Britain’s membership of the European Union, how do you think you would vote? Would you vote to leave the EU or to stay in?

Response options were to leave the EU; stay in the EU; not to vote; and ‘don’t know’.

Of the BES sample, 30 per cent said they would vote to leave, 47 to remain, 7 per cent would not vote and 15 per cent did not know. This looks out of step with the result last month and perhaps a little stronger for Remain than opinion polls at the time. However, this may be largely down to spontaneous ‘don’t knows’ having been recorded by interviewers rather than a choice being forced. In any case, we are more interested in differences across social groups than predicting the overall outcome.

Let’s first break it down by religious group, by social generation, and by strength of Englishness. For the first, I recoded the religious affiliation variable so that we have a six-fold measure of religious tradition: Nones (43 per cent), non-denominational Christians (13 per cent), Roman Catholics (9 per cent), Anglicans (21 per cent), other Protestant (including the Free Churches and Presbyterians – 6 per cent), and non-Christians (8 per cent). Non-denominational Christians are an interesting mixture of those who are only nominally-Christian – often simply to distinguish themselves from non-Christians, and effectively as a cultural marker  – and those who are affiliated to New Churches, often evangelical. Because they tend to be younger, and the latter more counter-cultural, it’s useful to separate them out. The non-Christian groups, admittedly an awkward amalgam, were individually few in number; and while it would obviously be preferable to examine differences between Hindus, Sikhs, Muslims, Jews and Buddhists, this requires a larger sample or booster sample of ethno-religious minorities.

Table 1: Support for Leave by Religious Tradition.

Percentage indicating Leave versus Remain, don’t know or would not vote (%) Unweighted base
None 28 1252
Non-denominational Christian 28 385
Roman Catholic 29 273
Anglican 43 668
Other Protestant 27 193
Non-Christian 20 195
Total 30 2966

Source: British Election Study Post-Election Face-to-Face Survey.

What is notable is the difference between Anglicans and members of other faith groups, including Nones. Non-Christians are more often of immigrant background and therefore less averse to the EU in general. Why Anglicans are more supportive of Leave than other Protestants, such as Baptists and Methodists, is less self-evident.

We now turn to differences across the social generations. We use Ipsos MORI’s definition: those born before 1945 are considered pre-war (14 per cent); those born between 1945 and 1965 are Baby Boomers (32 per cent); those between 1966 and 1979 Generation X (23 per cent) and those born from 1980 Generation Y (30 per cent). The differences in support for Leave are given below.

Table 2: Support for Leave by Social Generation.

Percentage indicating Leave versus Remain, don’t know or would not vote (%) Unweighted base
Pre-war 41 510
Baby Boomers 38 1091
Generation X 27 695
Generation Y 19 691


Here we see a clear generational gradient: each successive social generation is less in favour of Leave.

We can also examine how strength of Englishness versus Britishness affects the tendency to support Leave. Because of the particular interest in English national identity, respondents in Wales and Scotland have been screened out for this calculation. The measure we use here is from the Moreno question, first designed to assess dual identities in Spain. This is not perfect, since people can feel that they have both a very strong English and British identity, for example, but nevertheless it has become a standard and tells us something about how respondents rank these two identities. Respondents were asked,

‘Which, if any, of the following best describes how you see yourself?’

Response options included English not British (15 per cent); more English than British (11 per cent); equally English and British (52 per cent); more British than English (11 per cent); and British not English (12 per cent). Coders also captured whether people responded don’t know, none, or other, but we set those aside for now.

If we separate out those who feel more strongly English from those who feel equally English and British, or more British, then we see the following.

Table 3: Support for Leave by Relative Strength of Englishness vs Britishness.

Percentage indicating Leave versus Remain, don’t know or would not vote (%) Unweighted base
More English 48 648
Equally English/British or more British 27 1729


The difference here is quite large, and suggests that English identity, or at least its manifestation as distinct from Britishness, has a strong association with referendum voting intentions.

A final aspect which has been much-discussed is the association between voting Leave and being working class, or even having a working-class identity, which may be a cultural identity separate from objective labour market status, involving having grown up in a working class family and maintaining a working class lifestyle to some extent. The BES includes both subjective and objective measures of class: people were asked the nature of their occupation, and were also probed as to whether they considered themselves to be working class, middle class or other. The results break down as follows. For the occupational measure we use the NS-SEC typology as to whether an occupation is professional, intermediate or routine.

Table 4: Support for Leave by Occupational Status and Self-Identified Social Class.

Percentage indicating Leave versus Remain, don’t know or would not vote (%) Unweighted base
Professional 23 1032
Intermediate 33 655
Routine 36 1123
Self-identified working class 34 1947
Self-identified middle class/other/don’t know 24 1040


What is interesting here is both the occupational gradient and also the high percentage of those identifying as working class – one that is actually higher than that in Routine occupations. This is a phenomenon noted elsewhere and again suggestive that traditional identities remain important, perhaps as a reaction to economic and social change.

But how far is social generation fundamentally driving many of these differences? For this we turn to multiple regression so that we can assess the effect of each of a set of variables. We wish to test the effect of social generation, religious affiliation, national identity and class. We should also control for other potential sources of variation: being male, which is often associated with being more likely to take political risks; education (distinguishing those with at least some higher education from those with secondary or vocational educational qualifications, and those with none); a control for whether the respondent was resident in London; and a measure of whether the respondent reports whether they attend a place of worship or not. This last term is designed to capture the effect of ‘religious social capital’ and active religiosity on attitudes compared with religious identity; attendance both varies by religious tradition, of which we should take account, and moreover is often thought to promote openness or at least attenuate nativist attitudes. To the extent that we can distinguish visible ethnic minority status and religious background, a term is included for whether the respondent is White, or other than White.

We use a logit specification, with the dependent variable taking a value of 1 if the respondent indicated support for Leave, and 0 if they supported Remain, or that they would not vote or did not know. The results are as follows.

Table 5: Results of logistic regression models predicting support for Leave.

Base category: respondent is female, member of the Baby Boom generation, has secondary-level qualifications or vocational qualifications, has no religious affiliation, does not attend a place of worship at least monthly, is other than White, is not resident in London, has professional/managerial occupational status, is not self-identifying working class. The Moreno scale measure is treated as continuous. Variables highlighted in blue are significantly associated with a lower likelihood of supporting Leave; those in pink with a significantly higher likelihood. * p < 0.1; **p < 0.05; *** p < 0.01.

We expected that male respondents might be more likely to support Leave, but that does not show up in these results. Generational effects, however, do show up: while the Pre-War Generation seems no more likely than the Baby Boom generation to support Leave, members of Generation X and Y are significantly less likely to do so. The education effects noted by pollsters and others are also evident here: those with at least some higher education (degree-level education but also higher-level NVQs or teaching and nursing diplomas) are significantly less likely to support Leave.

Turning to the religious affiliation measure, we see one interesting result. The ‘Religious Nones’ form the base category. We see that Catholics and other Protestants are not significantly more or less likely to support Leave; neither are non-Christians, which could be interpreted as their sharing a worldview with Religious Nones in this regard. What is especially notable is the result for Anglicans, who are significantly more likely to support Leave. This is interesting because the effect holds after controlling for social generation and education.

“What is especially notable is the result for Anglicans, who are significantly more likely to support Leave. This is interesting because the effect holds after controlling for social generation and education.”

A term was added for religious practice because we often see that affiliation and practice have countervailing effects, for example on attitudes to immigration. We might well imagine that those who attend church regularly are different to those who are only nominally religious. In this case, however, we find no effect. The term for being White rather than a member of a visible ethnic minority or of dual heritage also had no effect, at least for this sample. The expected effect of being a resident of London was also insignificant. As it happened, Leavers tended to be more likely to be resident in outer London but as a whole London was strongly Remain. Perhaps this was a false negative; perhaps attitudes changed in the year or so between the survey and the referendum; or perhaps much of the London effect is primarily down to education, occupation and civic-national identity.

Finally, we look at national identity and class identity variables. We treat the measure derived from the Moreno scale as continuous, and a higher score on this measure means ‘more English’. We see that the effect is strong and positive. Having an intermediate occupation rather than a professional or managerial occupation has no significant effect, but those with a routine rather than professional occupation are more likely to support Leave, albeit at the 10 per cent level of significance. The effect of having a subjective view of being working class, however, is positive even after controlling for occupational status. This is of considerable interest. It suggests that at least at this point (May-September 2015) support for Leave was related to identity and ‘mentalities’ rather more than to objective structural position, at least to the extent that these can be distinguished using regression analysis.

Should we take this Anglican effect at face value? We could think of a number of mechanisms through which it could work – or alternatively of a few different factors for which this is acting as a proxy. Anglicanism itself was defined as a national church and moreover as a medias res, pragmatic, distinct from continental forms of Catholicism and Protestantism. Affiliation to the national, established church also corresponds to nativist instincts. Secondly, there is possibly a rural/market town effect. The Church of England arguably provides a social and symbolic function in areas outside urban England, in areas with stronger national feeling, which both shores up affiliation and then serves as a proxy for such feeling. Thirdly, maintaining a religious affiliation is traditional and socially-conservative given a wider environment of public and private secularity. It signifies deference to the past, to long-established institutions and to the values of older generations. By contrast, the EU is something of a modernist project, and cosmopolitanism associated with liberal values. Fourthly, the choice to maintain affiliation may be associated with a particular view of the social contract: norms of duty and reciprocity together with what over the years has become a cultural defence against outsiders and the ‘undeserving’.

There is much more to be done here. First, the post-referendum survey is eagerly awaited: how far did these effects show up in behaviour as well as in what people said they would do? Further, the mechanisms linking religious affiliation, class identity, generational status and educational experience to referendum vote preference need elaboration. In this case, we have not explored basic values (authoritarianism, liberalism) as either prior orientations or as mediators of these effects. Extending the analysis to Scotland, Wales and Northern Ireland, and using sources with larger ethno-religious minority samples would also allow further probing of the sense of Englishness, Britishness and EU support by religious tradition: how does a sense of local versus British nationality vary by religious group? Do different religious minorities show significantly different levels of support for the EU? And if there is variation, why is this? Answers will inform us of the nature of the imagined communities which we inhabit in the UK.

About the author

siobhanSiobhan McAndrew is Lecturer in Sociology with Quantitative Research Methods at the University of Bristol. She specialises in the social science of religion and culture. Stata syntax to reproduce the results above is available from the author on request.