By Dr Ben Clements, University of Leicester (reposted with kind permission from British Religion In Numbers)
As the weekend round-up of religious news on BRIN flagged up, the British Election Study (BES) 2015 has released the first version of the 2015 general election results dataset. This dataset (and the accompanying documentation) can be obtained here. Across parliamentary constituencies, the dataset includes the vote share for each party at the 2015 general election. It also includes religious affiliation data from the most recent English and Welsh census and and Scottish census (2011). The religious affiliation data are available in separate variables measuring the proportion in each of the following categories: Christian, Buddhist, Hindu, Jewish, Muslim, Sikh, other religion, no religion, not stated. Using this dataset we can perform some basic analysis of this aggregate level data – that is, across constituencies – to look at the association between religious affiliation and party vote share at the general election.
The results are reported in the table below. This reports the correlation coefficients for the bivariate associations between three of the religious affiliation indicators (% Christian, % Muslim, % No religion) and four party vote share categories (% Con, % Lab, % Lib Dem and % UKIP). These coefficients indicate the direction and strength of the association between two variables. They can range in value from -1 to +1. A positive value indicates that as one variable increases in value, the other variable also increases in value. A negative value indicates that as the value of one variable goes up, the value of the other variable goes down. The larger the (positive or negative) value, the stronger is the association between the two variables.
When undertaking and reporting this sort of analysis, two points are particularly important to bear in mind. First, as Denver et al (2012), pp. 36-37) make clear, correlation coefficients cannot tell us whether variation in one variable (here, level of religious affiliation) causes the variation in another variable (here, party vote share). They can only show the extent to which two variables are associated – that is, whether they increase or decrease in value together (they are positively correlated); or whether as one increases in value, the other decreases in value (they are negatively correlated)). Secondly, as Denver et al (2012) also caution, given that we are looking at aggregate-level data (based on information pertaining to the constituency-level) we cannot conclude from these data that the same association is present amongst individuals within constituencies (to presume this to be this case would be an ‘ecological fallacy’).
Religious affiliation and party vote share at the 2015 general election: Bivariate correlations
Source: British Election Study 2015 Constituency Results.
Note: ‘(n/s)’ indicates a coefficient that is not statistically significant. All other coefficients are statistically significant.
All of the correlation coefficients are statistically significant, with the exception of the association between Christian affiliation and Lib Dem vote share, but their magnitudes clear vary. The correlation coefficients for Christian affiliation indicate that it is positively-associated with Conservative and UKIP vote share and negatively-associated with Labour vote share. The correlation coefficients for Muslim affiliation show negative associations with the vote shares of Conservative, Lib Dem and UKIP, but a positive association with Labour vote share. For no religion, at the constituency level there are negative associations with the vote shares of Conservative, Labour and UKIP and a positive association with Lib Dem vote share. As the magnitudes of the coefficients show, the strongest association is between Muslim affiliation and Labour vote share, and – interestingly – followed by that between Christian affiliation and UKIP vote share (and thus higher than that obtained for Christian affiliation and Conservative vote share).
Finally, it should be noted that variation in other constituency-level indicators – such as socio-economic circumstances or ethnic group composition – is often associated with variation in party vote share.
An analysis of religious affiliation and vote choice at the general election at the individual-level will be posted when suitable BES post-election data become available.
British Election Study 2015 Constituency Results.
Denver, D., Carman, C. and Johns, R (2012), Elections and Voters in Britain. 3rd edition. Basingstoke: Palgrave, pp. 36-37.