One of the common features of political polling is that while a plethora appear in advance of an election or a referendum, nobody bothers to conduct a poll afterwards. For newspapers, the most common commissioners, polls are of interest beforehand because they provide a basis for speculating about what will happen when polling day does arrive. Once the contest is over, however, their attention moves on to considering what is going to happen as a result of the vote and little effort and resource is expended on explaining why the vote turned out the way that it did. Instead we usually have to wait for one or more academic surveys to appear before we obtain robust evidence on who voted a particular way and why.
However, in the case of the independence referendum a couple of polls have been conducted of how people voted. Together with the polls taken in the final hours of the campaign they enable us paint a relatively clear picture of the demography of the Yes and No vote. At the same time, we can also examine the pattern of the results in the 32 local authorities to see whether the places where Yes and No did relatively well were indeed the kinds of places that we would have expected them to do well, given the patterns uncovered by the polls.
The first poll of how people actually voted was conducted by YouGov on 18 September. In this exercise they recontacted as many as possible of the people whom they had interviewed (via the internet) for their final poll published in The Times and The Sun on the morning of polling day. It comprised 1,756 people who told people on the day that they had voted, and 783 who in previous polls had advised YouGov that they had already voted, an overall total of just over 2,500 respondents. The poll was used by the company to predict at 2230 on referendum night that the No side would win with 54% of the vote.
The second exercise was undertaken by Lord Ashcroft. The poll was conducted on 18 and 19 September, partly by phone (1,216 respondents) and partly online (831 respondents), giving a grand total of 2,047 interviewees. The balance of Yes (45%) and No (55%) votes in this poll matches the actual result, but it is not possible to tell from the published tables whether or not this is because the poll was weighted to ensure that it matched the actual result.
Between them the two exercises confirm the presence of four patterns that were evident in the polls throughout most of the campaign: women, older people, those in more affluent circumstances and those who were born elsewhere in the UK were all relatively reluctant to vote Yes.
According to the Ashcroft poll, 47% of men voted Yes, compared with 44% of women. That is a relatively small gender gap as compared with that identified in the final polls – on average the Yes vote amongst men in each company’s final poll was 51%, while it was only 43% amongst women. YouGov’s figures are closer to those averages, with 51% of men voting Yes and only 42% of women, so it seems quite likely that the smaller gender gap in the Ashcroft poll is simply a product of the random variation to which all polls are subject.
The two polls break down their respondents by age rather differently, but both identify a big difference between the voting preferences of older voters and the remainder of the population. According to Ashcroft just 27% of those aged 65 and over voted Yes, while only 43% of those aged 55-64 did so. In contrast 52% of Ashcroft’s 45-54 year olds voted Yes, 53% of the 35-44 year olds, 59% of 25-34 year olds and 52% of those in the 16-24 age group. These figures have led some to conclude rather indelicately that older voters thwarted the will of the under 55s. However, the YouGov poll suggests that the age gap may not have been quite that stark. It has 34% of those aged 65 and over voting Yes, compared with 45% of those in the 60-64 age bracket, 47% of those aged 40-59, 55% of those 25-39 and 49% of those 16-24. Neither poll contains enough 16 and 17 year olds for us to be able to make any definitive statement about how this newly enfranchised group voted. What we can note is that in both polls the highest level of record support for Yes was amongst those in their late twenties and thirties rather than amongst those aged between 16 and 24.
Both polls report that those classified as ‘C2DEs’ , that is those in working class occupations, were rather more likely to vote Yes than those in more middle class ABC1 jobs. According to Ashcroft, 44% of ABC1 voters voted Yes compared with 47% of C2DE ones. The gap in YouGov’s data is a little larger, with 41% of ABC1 and 50% of C2DE respondents saying the voted Yes. The average gap in those final polls for which the information is available was somewhere inbetween these two estimates, with 44% of ABC1 and 50% of C2DE voters saying they intended to vote Yes.
However, these figures (not least because they are simply based on a binary division of the population) may understate the extent to which those in different social circumstances voted differently. According to Ipsos MORI (from whose last two polls the relevant information is uniquely available), no less than 65% of those living in one of the 20% most deprived neighbourhoods in Scotland voted Yes, compared with just 36% of those in the one-fifth most affluent.
Finally, according to YouGov only 26% of those who were born in England, Wales or Northern Ireland voted Yes, little more than half the equivalent figure (49%) amongst those born in Scotland. In contrast, the division of the vote amongst those who were born outside the UK (many of whom would have been Irish, Commonwealth or EU citizens) was much less distinctive, with 41% saying they voted Yes. Meanwhile, people’s sense of national identity was also reflected in how they voted. In Ipsos MORI’s two final polls, no less than 88% of those who said they were Scottish and not British voted Yes, compared with 65% of those who said they were ‘More Scottish than British’ and 26% of those who felt ‘Equally British and Scottish’. Amongst the two remaining small groups, only 9% of those who said they were ‘More British than Scottish’ identified themselves as Yes voters, and just 13% of those who said they were British and not Scottish.
These patterns were reflected in the pattern of the kinds of places that recorded a relatively high Yes vote and those in which the No side did relatively well (though as the ratio of female to male voters does not vary much by local authority we cannot expect to find any evidence of the gender gap in the actual results). In those local council areas with a relatively high proportion of people aged 65 or more (i.e. those in the top third of all council areas on this criterion), the Yes side on average won just 39% of the vote, while in those with a relatively low proportion the average Yes tally was 47%.
Nothing was more strongly correlated with the level of Yes support in each council area than the level of unemployment. In those areas with relatively high unemployment, Yes support averaged 51%; in those with low levels of unemployment it was just 39%. There were similar differences between places with a high (51% Yes) and low (40%) proportion living in one of Scotland’s most deprived neighbourhoods and between areas with a relatively large (40%) and small (51%) proportion of people engaged in professional and managerial occupations.
Finally, the Yes vote stood at 48% amongst those areas in which a relatively low proportion of the population was born elsewhere in the UK, while it averaged 39% in those places where relatively higher numbers of people were born elsewhere. Equally, the Yes vote averaged 49% where a relatively low proportion (in the 2011 Census) said that they were (exclusively) British, and just 38% in those where a relatively high proportion gave that response.
These general patterns were reflected in the places with the highest and lowest Yes votes. A majority voted Yes in just four areas, Dundee (57.3%), West Dunbartonshire (54.0%), Glasgow (53.5%) and North Lanarkshire (51.1%). Between them these four areas are amongst the top six in terms of the unemployment claimant count and the proportion living in an area of multiple deprivation, and amongst the bottom six in their proportions of people in professional and managerial occupations. They are also all in the bottom third in respect of the proportion of their populations that are aged 65 or over and the proportion born in the rest of the UK.
The four highest No votes were in the two Northern Isles councils of Orkney (67.2%) and Shetland (63.7), together with the two councils adjacent to the border with England, Scottish Borders (66.6) and Dumfries & Galloway (65.7%). These are all areas in the bottom half of the league table of local authorities in respect of their levels of unemployment and percentage living in deprived neighbourhoods. All bar Shetland also have a relatively large population of older people What most distinguishes them, however, is that they are all amongst the top seven in terms of the proportion born in the rest of the UK and in the top six for the proportion who describe themselves as British.
Of course, describing the patterns of the kinds of people who were more or less likely to vote Yes or No does no more than give us clues as to why people voted they way they did. What we can note at this stage is that women, older people, those in ABC1 occupations and those born elsewhere in the UK were all, according to YouGov’s final poll for The Times and The Sun, relatively pessimistic about the economic consequences of independence. And as we have repeatedly noted on this site, nothing seemed to matter more to voters in deciding whether to vote Yes or No than their perceptions of the economic consequences of leaving the UK. Doubtless those who were less well off were more easily persuaded that independence might hold out the prospect of a better tomorrow, and that helps explain why the Yes vote acquired its relatively working class, less affluent character. But why women proved persistently reluctant to accept the economic case for independence will probably provide the basis for analysis and commentary for quite a while to come.