Since the general election in December 2019 there has been much talk about how Labour should responsd, trying to recapture voters who are more ‘socially Conservative’, more ‘English’, who voted Leave in the referendum now that Brexit is ‘done’; there are no shortage of groups who did not vote Labour in 2019 to aim at.
This short series of charts highlights the way that values and identity interact in voting decisions. Data are taken from wave 16 of the British Election Internet Panel Study, collected after the EU parliament elections in June 2019, and are for England only. Values are measured on two scales an economic ‘left-right’ scale and a non-economic scale measuring ‘liberal-authoritarian’ values. The items used to create these scales are:
In each case people are asked whether they agree or disagree with the statement on a 5 point scale. The average of their responses is taken as their position on the latent scale variable. Taking these positions on the two scales each scale is broken down into three groups (these are not equally sized but ratehr defined relative to the theoretical mid-point of the scale). For the left-right scale we have a ‘left’, ‘centre’ and ‘right’ group. For the liberal-authoritarian scale a ‘liberal’, ‘centre’ and ‘authoritarian’ group. Combining these two creates nine groups in the ‘value space’ defined by the two scales. (Note at the aggregate level the two scales are uncorrelated, you cannot predict someone’s ‘liberal-authoritarian’ values by knowing their position on the ‘left-right’ scale).
This is combined with a scale of ‘Englishness’ which asks people how strong they feel on a scale from 1–7 with 1 being ‘Not at all English’ and 7 being ‘Very strongly English’. In the charts the ‘not very English’, ‘very strongly English’ and mid-point (4/7) are shown. More than half of the reponses fall into one of these three categories with the very strongly English the largest (a third of responses).
Turning first to the EU referendum vote. As shown in many analyses of the vote, the economic left-right position has only a marginal effect on the proportion on the Leave vote. For example, comparing those who are Not at all English (the Blue bars) and who are ‘authoritarian’ the proportion voting leave varies from 63% for the ‘left’ group’ to 71% for the ‘right’ group. This is dwarfed by the difference between the ‘liberal’ and ‘authoritarian’ groups. Comparing those on the ‘left’ economically the difference is from 5% voting leave among the liberal group and 63% among the ‘authoritarian’ group.
Englishness does make a difference to this, though a smaller one than the ‘liberal-authoritarian’ value divide. Comparing across the colours within each groups shows this is almostly consistently a 20 percentage point difference between the ‘not at all’ English and the ‘very strong’ English, but it is not an entirely linear relationship. For those not on the left economically, those in the centre of the Englishness scale are less likely to be Leave voters than those at either end of it.
The same analysis for Labour voting in 2017 shows a quite different picture. Here there remains a clear divide according to the left-right scale. This is modified by Englishness, for example among the Liberal left group there is an 11 percentage point difference between the most and least English groups. It is also modified by the liberal-authoritarian scale, on the ‘left’ there is a 23 percentage point difference between the liberal and authoritarian groups. But in this instance it is the left-right difference that is the largest. Though notably in 2017 Labour won more than half the votes of the ‘liberal-centre’ group who did not feel strong English, indicating the importance of the combinations of values and identities.
The data for 2019 voting is not yet available and the data collected in June 2019 for vote intention is not very useful for explaining the 2019 result as it captures a high point for both the Liberal Democrats and the Brexit party and pre-dates the Conservative leadership change. But there is a measure which asks people how likely they are to ‘ever vote’ for a party on a 0–10 scale. Taking those who rated Labour as 6/10 or higher as ‘likely Labour’ voters we can see how this looked 6 months before the election.
The proportions in each group are lower here than for 2017 voting. Perhaps reflecting the resurgence of smaller parties, or perhaps indicative of a wider disconnection with the Labour party that was an early warning sign of the loss of support across all groups that occurred in the general election 6 months later. However, the pattern of the voting is also slightly changed. The overall shape according to the ‘value space’ is similar, Labour does best among those on the left economically, and this is modified by liberal-authoritarian values so that the more liberal groups are more positive about Labour within each economic value group. But the role of Englishness has changed here. With Labour now doing best among the ‘middle’ group of the English and for those not on the economic left doing equally badly among those who feel ‘not at all English’ and those who feel ‘very strongly’ English. It is possible that this is especially influenced by the growth in popularity of the smaller parties, and will be something to look out for the 2019 data when it is released.
Are the differences in 2019 a reflected of a politics that has been restructured by the EU Referendum? To look at this we can look back at past voting using the same framework. A note of caution, the figures below show 2015 and 2010 voting patterns according to current values and identities. This is important as we can’t read off from it exactly how these groups as they were constituted in 2015 and 2010 voted, voters have left the electorate since then. But it does show how voters still in the electorate in June 2019 voted in the past — if these voters have *never* voted Labour they will be much harder to convince perhaps than those who have more recently changed.
Starting with 2015, the differences between groups are similar in profile, and it is worth noting more similar to the pattern of the likely Labour voters in 2019 than the actual Labour voters in 2017 (in this sense when we have data for 2019 it is possible we will see strong similarities with 2015 voting patterns and 2017 as the outlier in recent elections). The differences are smaller than in 2017 or in terms of EU referendum voting but they are broadly similar. Left-right economics is a strong predictor of Labour voting but the liberal-authoritarian divide also matters. In fact to such an extent that the ‘Centre-Liberal’ group were more likely to have voted Labour than the ‘Left-Authoritarian’ group. Again there is a clear difference outside of the ‘Left’ group in terms of Englishness. Within the economic left higher levels of English identity are associated with fewer Labour voters. But for the centre and right groups it is those with ‘moderate’ Englishness that are more likely to support Labour.
The divides driven by liberal-authoritarian values interacting with left-right values and Englishness interacting with left-right values are clear even in 2015. They were emerging before the 2016 referendum and everything that followed in terms of Labour’s position on a 2nd referendum. They were also emerging before Cobryn took over as leader. They were more muted in 2015 than at later elections but they were present. The events since have supercharged a division that was already becoming less dormant.
If we compare this with 2010 voting (again with the caveat that this is not a profile of 2010 voters but how those in these positions in the electorate voted in 2010) the picture is quite different. [Further comparison with 2005 is available her
For those on the left economically, there is no difference in how they voted in 2010 according to the ‘liberal’ authoritarian scale. This is largely because of the appeal of the Liberal Democrats among the liberal-left group at this election. Among those who did not feel English and were in the ‘liberal-left’ group the Liberal Democrats were the first choice party in 2010. Because these are not data for the whole electorate in 2010 we should be wary of over interpretation but analysis of data collected in 2010 which is representative of the electorate shows a similar pattern across the two-dimensional value space.
Comparing across the last 4 elections and the EU referendum vote shows the complexity of the way that values and identities combine. It also shows that these divides pre-dated the 2016 referendum but that they have been accentuated by it (though whether the trends would have continued regardless is unclear). Understanding voters priorities and connections with politics cannot be reduced to a single dimension or identity; they prioritise different aspects of these when asked different questions. Understanding the how these combine is essential for any party seeking to build a winning coalition in the 2020s.
Paula Surridge is a Senior Lecturer at the University of Bristol. She tweets at @p_surridge. This article was first published on the blog platform Medium and has been reposted with the permission of the author. Image credit: CC by UK Parliament/Flickr.
