EXPLAINING BREXIT NORTH AND SOUTH OF THE BORDER
David McCrone
The pall of Brexit 2016 hangs like a dark cloud over Scottish and British politics, and yet, arguably, we do not properly understand how it came about or what its effects have been. We are told, for example, that the 2017 British General Election is ‘about Brexit’ in that the SNP saw it as the basis for a second ScIndyRef once the process was complete, and the Scottish Tories opposed it, largely on the same basis.
Before becoming enveloped in post-Brexit politics, we should clarify what actually went on, and in particular, who voted for it, and why. The broad outlines are clear, and have become part of the political furniture. Across the UK as a whole, 51.9% voted Leave, and 48.1% Remain. England as a whole voted Leave (53.4%); Scotland voted Remain (62%). Furthermore, all council areas in Scotland voted Remain. Thus did Brexit provide a change in the material basis of Scottish-English relations, and make a second ScIndyRef seem more feasible.
Whydid Scotland and England vote differently? More specifically, did similar kinds of people – in socio-demographic terms - vote differently in the two countries? Using data from the Scottish Social Attitudes survey (SSA) carried out post-Brexit, and for England, the British Social Attitudes survey (BSA) from which we can extract people living in England[1], we can compare the two samples. We have comparable data for respondents on the following: sex, age, social class, education, religion, national identity, party identification, as well as two sets of attitudinal scales: libertarian-authoritarian[2], and left-right[3]. We also have specifically Scottish questions asked in SSA: on preferred constitutional preference, how people voted in BGE2015, how they voted in ScIndyRef2014, and how they were likely to vote in ScIndyRef2 should that arise.
Who voted for Brexit?
Our first task is to compare responses on a number of social and demographic characteristics, which in summary form are as follows. For simplicity, we will focus on the Remain option, the Leave response being the obverse in percentage terms[4]:
Sex
% Remain / Scotland / England / d S-EWomen / 68 / 52 / +16
Men / 61 / 49 / +12
dW-M / +7 / +3
Age
% Remain / Scotland / England / d S-E18-24 / 85 / 70 / +15
65+ / 59 / 35 / +24
dyng-old / +26 / +35
Social class
% Remain / Scotland / England / d S-EEmployers, Managers etc / 73 / 62 / +11
Manual workers / 59 / 37 / +22
dmc-wc / +14 / +25
Education
% Remain / Scotland / England / d S-EWith degrees / 78 / 77 / +1
No qualifications / 51 / 23 / +28
ddegree-none / +26 / +54
Religion
% Remain / Scotland / England / d S-ERC / 76 / 56 / +20
C of S/ C of E / 59 / 42 / +17
none / 64 / 52 / +12
dRC-CofS/E / +17 / +14
Party support
% Remain / Scotland / England / d S-ELabour / 79 / 66 / +13
Cons / 55 / 46 / +9
Lib-Dem / 80 / 74 / +6
dLab-Cons / +24 / +20
Note: SNP 63%
National Identity
% Remain / Scotland / England / d S-ENational not British / 61 / 27 / +34
British > or not national / 61 / 64 / -3
dnat-Brit / 0 / -37
Note: ‘national’ = Scottish or English
Libertarian-Authoritarian scale
% Remain / Scotland / England / d S-ELibertarian / 93 / 89 / +4
Authoritarian / 51 / 33 / +18
dlib-auth / +42 / +56
Left-Right scale
% Remain / Scotland / England / d S-ELeft / 61 / 47 / +14
Right / 71 / 51 / +20
dL-R / -10 / -4
We can summarise these tables as follows:
- By and large, Scots were more likely to vote Remain than the English regardless of the variable in question. Even among people in Scotland and England with similar characteristics, Scots were much more likely to vote Remain.
- That the differential within each variable is broadly similar, with regard to sex, religion, party support, left-right. Differences occur with regard to age (9 percentage points), social class (11 ppt), education (18 ppt), national identity (37 ppt), libertarian-authoritarian (14 ppt).
- In Scotland, Remain voters tended to be female, young, of higher social class, more educated, Catholics, Labour or Lib-Dem supporters, libertarian, and marginally to the Right. In England, Remain voters tended to be female (just), young, of higher social class, more educated, Catholic, Lib-Dem or Labour, and libertarian. In those respects, Scotland and England are quite similar in support for Remain. Similar kinds of people voted in roughly similar ways.
- In one key respect, there is difference. National identity mattered far more in England than in Scotland[5]. The more ‘English’ you are, the more likely you were to vote for Brexit. Being ‘Scottish’, on the other hand, made no difference at all. This might seem odd, given the conventional wisdom that national identity is much stronger in Scotland, but the key point is that being Scottish is so ubiquitous north of the border, that it has no independent, that is, discriminating, effect. In England, on the other hand, thinking of yourself as English was a strong influence on the vote for Brexit. ‘English’ nationalism matters. Put another way, if you thought of yourself as only or predominantly English, you were far more likely to vote for Brexit (66% voted Leave); and if you thought of yourself as only or predominantly British, you were more likely to vote Remain (64%). In Scotland, ‘predominantly Scots’ voted 35% to Leave, (and 65% Remain), comparable Brits in Scotland voted 39% to Leave (and 61% Remain).
- These findings are obtained simply by examining the crosstabs, but they are helpful in trying to work out why Scotland and England came up with different Brexit results. We cannotsimply assume that national identity is what made the difference, because crosstabs do not in themselves tell us anything about interactive effects. A difference might simply be the result of the operation of some other variable, that, to take a putative example, ‘religious’ difference such that being ‘Protestant’ (at least, adhering to the national or established churches, Church of Scotland or Church of England) may simply be an artefact of age, or social class. If this were so, we would expect any religious difference to disappear if we put the key variables into the same statistical pot.
The Scottish data gives us further refinements in that we have further questions on politics including past and future independence voting intentions. Let us, however, start with the ‘English’ data analysis to provide a benchmark for the Scottish data.
Explaining Brexit in England
Modelling all nine variables in order to account for people’s Brexit vote, we find that three lose their significance: left-right, sex, and religion. Religion, for example, seems to be a function of age (being older), and insofar as older people were more likely to vote Leave, any independent effect of religion weakens considerably when age is included in the explanatory model using binary logistic regression[6]. Focusing on the remaining six variables, social class, age, national identity, party identification, education and position on the libertarian-authoritarian scale, we find that all have independent effects on Brexit vote. This indicates that the most likely person in England to vote for Brexit tended to be: an elderly working class person with no educational qualifications, thinking of themselves as English, identifying with the Tory party, and having authoritarian views. These characteristics seem to be autonomous such that each has an independent effect[7]. Above all, having authoritarian views, which is independent of one’s educational level, seems especially associated with voting for Brexit. Thus, the odds ratio (Exp(B) in logistic regression) is more than twice that of other variables (> 2.5) indicating its power as regards Brexit vote[8]. To sum up the Brexit vote in England: the significant factors were lib-authn values, education, age, party identification, and social class.
Explaining Brexit in Scotland
The factors accounting for Brexit vote in Scotland are similar, but there are differences. Religion, age and national identity are relatively weak, but sex and education are strong. Social class only operates if education is omitted, suggesting that the latter has a stronger effect than the former. As in England, having authoritarian values helps to account for voting for Brexit; set against sex and education, such values have an odds ratio of 2.7, which means that people were more than two and a half times more likely to vote for Brexit, even if you are male and with no educational qualifications, the other most significant factors.
What of ‘political’ variables? SSA 2016 included questions on party identification, BGE2015 vote, constitutional preferences, and intended vote in ScIndyRef2. Perhaps surprisingly, none of these have a strong effect on Brexit vote, and none show significance in the models. It is the case that Left-Right values do figure in the Scottish models (s=.041 when education, sex and lib-authn are included, all at more significant levels), bearing in mind that being on the Right is associated with a vote to Remain. In other words, being on the Right in Scotland makes people less likely to vote Leave, and in the Scottish models ‘Left-Right’ remains in the frame, when it drops out of significance in England, possibly because of the power of national identity in England, and not in Scotland. Tentatively, we might explain that in terms of the ubiquity of ‘being Scottish’ across the political divide such that voting Remain was not seen primarily as an expression of national identity.
What is common as regards Brexit vote in Scotland and England is the strong association with authoritarian values. In both countries, there are odds ratios of more than 2.5, far in excess of other variables. Why that should be so is something of a puzzle. One way to try and solve it might be to break down the scale into its different components. Doing this for the Scottish data, and including sex and education in the model, we find that the key items on the scale are[9]: respect for traditional British values, the death penalty as appropriate, and censorship as necessary. The three ‘law’ elements – stiffer sentences, children taught to obey the law, and people should always obey the law – are not significant.
Disaggregating the components of the lib-authn scale for Scotland (excluding sex and education), we find most significant are ‘traditional values’ and ‘death penalty’, with ‘censorship’ significant at .030 level. ‘Stiffer sentences’, ‘obey the law’, and ‘obey wrong law’ are not significant. For England, we find a similar distribution, with ‘traditional values’, and ‘death penalty’ the most significant; ‘stiffer sentences’ more significant than in Scotland; ‘censorship’ at a comparable level; and ‘obey’ and ‘wrong law’ not significant. We can only conclude, conjecturally and without properly understanding the mechanisms, that Brexit appealed, north and south of the border, to those of ‘conservative’ dispositions, especially as these related to ‘traditional values’, support for the death penalty, censorship. While we might have expected factors like age and (lack of) education to explain these associations, it did not seem to be the case.
ScIndyRef2
The Scottish survey asked about likely vote in a future Independence referendum. Given the ways in which political parties have linked this to Brexit, it is instructive to exam the relationship more closely. Focusing on three variables – lib-authn, ScIndyRef2 and education, vis-à-vis Brexit vote, and using logit in loglinear analysis – we find that those most likely to have voted Remain in Scotland (as many as 96%) were the most highly educated, libertarian and intending YES voters. Indeed, having libertarian values gave a major boost to voting Remain, ScIndyRef2 and education being equal – by as much as 30 percentage points over ‘authoritarian’ voters of a similar ilk. This is not to conclude that, for example, education did not matter, for among intending YES voters with degrees, there was premium to vote Remain both for libertarians (+25) and for authoritarians (+24).
Conclusions
- We can conclude that the Brexit vote in Scotland was not driven by a different set of influences compared with those in England, with the exception of the role of national identity south of the border. Simply put, the more ‘English’ you felt, the more likely you were to vote Leave. In both countries, we can see the influence of age, education, and above all, lib-authn values. Overall, however, support for Remain was much stronger in Scotland than in England, even among people less likely to vote for it, such as older people, those with little education, and so on. Even where there was differential support for Remain/Leave, we find stronger support for the former north of the border.
- Furthermore, the links between ScIndyRef 2014, Brexit and ScIndyRef2 are not at all straightforward, which is possibly why drawing lines straightforwardly from YES1 to Remain to YES2, and conversely from NO1 to Leave to NO2 is not an easy task, as both Nationalist and Unionist parties have found out.
7th June 2017
1
[1]For both SSA and BSA, the samples are based on people resident in the two countries, hereafter referred to, for shorthand, as Scots and English.
[2] BSA and SSA use measures derived from attitudinal scales as follows: ‘Young people today do not have enough respect for traditional British values’; ‘People who break the law should be given stiffer sentences’; ‘For some crimes, the death penalty is the most appropriate sentence’; ‘Schools should teach children to obey authority’; ‘The law should always be obeyed, even if a particular law is wrong’; ‘Censorship of films and magazines is necessary to uphold moral standards’. Respondents are labelled ‘libertarians’ or ‘authoritarians’ (or in between) according to how they respond.
[3] Similarly, respondents are placed on a Right-Left scale according to how they respond to the following: ‘Government should redistribute income from the better-off to those who are less well-off’; ‘Big business benefits owners at the expense of workers’; ‘Ordinary working people do not get their fair share of the nation’s wealth’; There is one law for the rich and one for the poor;’ ‘Management will always try to get the better of employees if it gets the chance’.
[4]Our focus is on internal variation, excluding don’t knows and missing data, and hence, the marginals do not match the overall result. For example, the BSA England survey gives Remain at 51% and Leave at 49%, in contrast to actual result of 47/53.
[5]In the English sample, 28% said they were only or mainly English, 25% that they were only or mainly British, with 47% equally English and British. In the Scottish sample, 55% said they were only or mainly Scottish, 10% that they were only or mainly British, and 34% equally Scottish and British.
[6]Modelling social class and religion in England together gives the latter a significance level of .011, but this lengthens considerably to .761 when modelled against age rather than social class.
[7]For example, having authoritarian values is not a function of low levels of education, but seemingly independent of educational achievement. Just why having ‘conservative’ views on law and order, child discipline etc relates to support for Brexit is not clear.
[8]For example, the odds ratios for education, age and national identity are, respectively, 1.29, 1.28, and 1.21.
[9]see footnote 2 for question details