Values are often seen as the mechanism through which social positions are connected to political attitudes and behaviours; more enduring than individual policy and party preferences. Whilst it has been widely demonstrated that ‘education liberalises’, and that economic self-interest is connected to attitudes to economics, other values are less well understood; notably values relating to ‘populism’ and ‘ethnocentrism’.
Using data from wave 10 of the British Election Panel study allows us to look at the associations between social positions and a range of value scales. Four ‘value’ scales are derived from the wave 10 data (collected late in 2016). These measure the ‘economic’ (or ‘old’) left-right; liberal-authoritarian values; populism and ethnocentrism. The scales are created by coding the items such that low values represent the ‘left, ‘liberal’, least populist and least ethno-centric position respectively; each set of items are then averaged to create a position on each scale between 1 and 5.
The items making up each scale are:
Social positions are measured using age, gender, ethnicity, household income quartile, education level and housing tenure; to capture key dimensions of income, wealth and ‘cultural’ capital. As all of these measures are ‘categorical’ to include them in the models below a ‘reference’ category, against which the effect of the other categories is compared, must be selected. These are age 18–25, lowest income quartile, male, white British, no education qualifications and having a mortgage. The same linear regression model with these social demographic variables is estimated for each of the value scales and the standardised coefficients of these models compared. This allows us to assess the relative size of the effect of, for example, age on populism and authoritarianism or age and education on populism.
For each scale a linear regression model is fitted including these socio-demographic predictors.
Left-right (economic) values
As we would expect the largest effect on ‘left-right’ values is linked to economic position, specifically the effect of being in the highest income quartile vs the lowest. However, it is this contrast with the richest that seems to be of significance here, contrasts between those in income quartiles 2 and 3 are not significantly different from the lowest income quartile while the effect of the 4th income quartile is much smaller. There appears to be little income ‘gradient’ here and more a key contrast between the top incomes and everyone else.
Effects of other factors are relatively weak, but it is worth noting that once income and education have been taken into account young people are more right-wing than those in the 46 -55 and 56–65 age groups. Young Britain may be a hot-bed of socialism but if it is this is more likely driven by lower incomes rather than generational change.
Looking at liberal-authoritarian values there are two key elements with relatively large effect sizes. Age effects here are larger than they are for the left-right value scale, older people are less liberal than younger people (though the differences between those age 46–55, 56–65 and 65+ are not large). Second, as well documented elsewhere, education is the largest effect on liberal-authoritarian values, with those with an undergraduate degree (or higher) considerably more liberal than those without (note that the A level category here includes those who are currently students in HE but who have not yet obtained a degree level qualification).
Ethnocentric values also show stronger relationships with age and education than with income or housing tenure. However, here the age effects are more pronounced, not only in the size of the effects but also as a gradient across the age groups, each age group is more ethnocentric than those younger than them. This effect is not explained by their educational qualifications or relative economic status. But perhaps reflects the changes in understanding of the nation-state and the more globalised into which successive generations have been socialised.
The roots of populist sentiment are hotly contested, some arguing these are primarily driven by those who feel a cultural threat to their local (or national) way of life, while others suggest they are driven by the economic anxiety of those ‘left-behind’ by globalisation. Most commonly, populism is linked with authoritarian values but also with a ‘new’ nationalism. It is not, therefore, surprising to find that the influences here are complex. The largest effect is that of education, again the key divide is between those with a degree level qualification and those without; those with degrees are less populist than those without. Economic effects are almost all non-significant, except for the effect of being in the highest income quartile. The most well-off are significantly less populist than the least well off. Finally, each of the age groups are more populist than the youngest age group but we see little evidence here of an age gradient (unlike ethnocentrism) and effects are smaller than for the liberal-authoritarian scale.
The table below summarises these effects across the four value scales. Effect sizes that are 0.05–0.1 are shown with a single +/-, those between 0.1 and 0.1.5 by ++/ — and those over 0.2 by +++/ — -
The table highlights the complexity of the relationship between the different values. Age and education have opposite effects on the left-right value scale to each of the others, but these effects also vary. The liberal authoritarian scale is strongly related to both age and education, while populism is more strongly related to education but has a residual age effect and ethnocentrism shows the greatest age gradient.
In seeking to explain how value divides connect to contemporary political behaviour, understanding how they are rooted in socio-demographic positions is an important first step. The models here identify the complexity behind this, using a small selection of socio-demographic variables. They identify that the factors underpinning a ‘left-wing’ political outlook are often in tension with those underpinning a ‘liberal’, less ethnocentric and less populist outlook and it would be a mistake to think of these as a single ‘dimension’ of political values (even where it has sometimes historically been the case that values have been packaged together). For three of the four sets of values analysed it suggests that divides related to education (and specifically the attainment of a degree level qualification) is the most important single contrast. There is a substantial difference between the average positions of those with degrees and those without on what are increasingly seen as the ‘new’ divides in our politics. However, this alone does not tell the whole story, age itself remains important for these values in addition to the effects of education, most notably on values connected to ‘ethnocentrism’ (the ‘new’ nationalism).
Do values influence vote choice?
Much has been written already about how these values connect with voting in the EU Referendum, less attention has been paid to how they connect with vote choice in the 2017 general election. To what extent have these ‘new’ values replaced the old ‘left-right’ divide as key to understanding British electoral outcomes?
A simple model illustrates the impact these scales have in distinguishing Conservative and Labour voters in 2017. There is no equivalent here of the standardised coefficients to enable easy comparison of the size of the effects of each scale. However, we can compare the global model fit for the model which contains all four value scales against the model fit if each one were excluded to see which makes the greatest contribution to the model. This is done using the change in AIC in the models. The larger the change in the AIC the more ‘information’ is being lost from the model by removing that value scale.
As the figure shows, there remains a very strong relationship between vote choice in the 2017 general election and the ‘old’ economic values relating to ‘left’ and ‘right’. Conversely, there is little to be gained, in terms of predicting vote choice between Labour and Conservative, in knowing how ‘populist’ a voter is. The other value scales do add useful information to the model — knowing the liberal-authoritarian or ethnocentrism position further helps to separate Labour and Conservative voters but these effects are substantially smaller than the traditional left-right dimension.
We should probably not find this entirely surprising. I suggested prior to the election that Labour would do best where the focus of the campaign was on economic issues and that was indeed the focus of much of the campaign. The two parties were seen as further apart on this left-right dimension than they had been for some time. On other levels it appears more surprising. The impact of the EU referendum was supposed to be to give new impetus to cultural and identity divides, pushing the nationalist and authoritarian ‘leave’ voter to the Conservatives. Moreover, the unexpected wins for Labour in Canterbury and Kensington seemed to back up this perception, that young liberal voters flocked to Labour. This narrative may well still be true, these models are aimed at understanding the votes of individuals not of blocks of voters or constituencies. They serve as a reminder that the electorate is much bigger than those in marginal seats, that the young voters supposedly behind the ‘youthquake’ that surged Corbyn to…well not terrible defeat…are a small minority of all voters and that even when change is occurring many of the old patterns still hold. Voting decisions are complex, rooted in social values which themselves are complex and cross-cutting. It would be unwise to ignore old or new values in understanding contemporary British politics and political behaviour.