Two Notes on the Psephology of the Euro-Elections

by John Curtice, Patrick English, Stephen Fisher and Eilidh Macfarlane.

Even though the votes have yet to be counted, the Conservatives and Labour already seem to be undertaking their own post-mortems of what promises to be poor results for both of them in the European Parliamentary elections. Within the Conservative party there is a lively debate taking place about whether or not in the wake of the anticipated success of the Brexit Party, Tories could and should embrace at least the possibility of leaving the EU without a deal. Meanwhile, Labour, anticipating perhaps that the party has lost votes to others whom have adopted a clearer position on Brexit, looks as though it is about to consider the possibility of coming out more firmly in favour of some kind of public ballot.

Doubtless, all sides in this debate will look to the pattern of the results on Sunday for evidence to support their view of what their party should do. There has already been some speculation that the pattern of reported turnout in this election (information about which has already been released by some councils) suggests that voters in more Remain-inclined areas were more likely to turn out to vote, indicating perhaps a new determination among anti-Brexit voters to express their views at the ballot box. Meanwhile, we can anticipate that the results themselves will be poured over for evidence that the Conservatives or Labour have lost ground more in Remain-voting areas than Leave-inclined ones or vice-versa. However, in both instances caution will be required in interpreting the evidence when the full panoply of results is unveiled on Sunday night.


First of all, we consider the evidence on turnout. Two patterns can be observed in the data available so far. (Our evidence comes from the figures for 143 councils where the data have been collected by Matt Singh and/or Patrick Heneghan.) The first is that turnout appears to be up by a couple of points or so on 2014, and thus may well be a little above the average for previous Euro-elections. The second is that turnout appears to have increased more in those areas where a majority of voters backed Remain in 2016 than it has in those places where Leave were most popular.

Two notes of caution about the data on turnout are in order. The first is that the figures on turnout released so far are based on the total votes cast, including votes that might eventually be deemed invalid; turnout is conventionally calculated in the UK on the basis of the total number of valid votes cast, that is, excluding those votes which are deemed to be invalid. These amounted to 0.5% of all votes cast in 2014, enough to reduce the turnout as conventionally calculated by 0.2%. So even if the figures released to date prove to be representative, any increase in turnout is likely to prove a little less than has been reported so far.

The second note of caution arises from the fact that in 162 councils in England the European elections were held on the same day as local elections. It was clear from the pattern of results in 2014 that, other things being equal, this had the effect of increasing the turnout in those places where there were local elections – for some voters the local elections proved a stronger inducement to vote than did the prospect of electing MEPs. This time around, however, local elections have not been held on the same day anywhere (other than a few local by-elections), and so have not potentially boosted the turnout anywhere. Rather, much of England had local elections three weeks ago – and that might have had the opposite effect of reducing turnout in the Euros because some voters felt disinclined to vote twice in three weeks.

Whether or not there was a local election last time does seem to have had an effect on the pattern of the change in turnout since 2014. As we might anticipate, the figures released so far indicate that it has typically increased more in places where there were notany local elections in 2014 (on average by three points) than in places where there were(half a point) – the effect of which is to eliminate what had been a three-point difference in turnout between such areasin 2014. Equally, however, the increase in turnout seems to have been much less (one point) where there were local elections earlier this month than where it was not (five points).

In so far as we are interested in a measure of the extent to which these European elections engaged the interest and participation of voters more than the contest five years ago, attention should be focused on the change in turnout in those places where there was neither an local election in 2014 nor in 2019. That is as much as six points. The figure for Britain as a whole will underestimate the extent to which voters have been more engaged.

The impact of the incidence of local elections on turnout should clearly be taken into account in undertaking any analysis of whether it has increased more in those places that backed Remain in 2016 than it has in those that supported Leave.  As it happens, however, doing so confirms rather than undermines the overall story.

Overall, turnout appears to be up on average by four points in those areas where less than 55% voted Leave in 2016, compared with just half a point in those where fewer voted that way. The effect is to double what was already a three-point difference between the level of support in the two types area in 2014 to a six-point one now.

This contrast is replicated if we look separately at those places that held local elections in 2014 and those that did not. In the latter, turnout is up by two points in places where Leave won over 55% of the vote in 2016, but by five points where they won less than 55%. The equivalent figures in those areas where there were local elections are an increase of three points and fall of between one and two points respectively. Taking into account whether or not there were local elections this year does not eliminate the difference between predominantly Remain and mostly Leave voting areas either.

Such a pattern is, of course, no more than suggestive. It does not necessarily prove that turnout has increased more among Remain voters than their Leave counterparts. It may be that both Remain and Leave voters have been more likely to turn out in Remain-inclined areas than they were in 2014. But even if that is all that has happened it is still likely to be beneficial at the margin at least to those parties that are more popular among Remain voters.

Changes in the share of the vote

But why is there a need for caution in interpreting the election results when they are unveiled on Sunday night? If the polls are to be believed the Conservatives have lost ground much more heavily since 2017 among those voters that voted Leave in 2016 than they have among those who voted Remain. However, we also know that at the 2017 election the Conservatives increased their share of the vote among those that voted Leave, while losing ground among those who voted Remain – not least because the party won over many a voter who had previously been backing UKIP.

As in 2014, the results of the Euro-election will be declared council by council rather than parliamentary constituency by parliamentary constituency. This means that, perforce, the only easily available estimate of change in party support will be how the result this time compares with what happened in 2014, when UKIP were riding high and the Conservatives much less so than in 2017. This figure will reflect any changes in the pattern of party support between 2014 and 2017 (during which period UKIP and Leave voters more generally switched to the Conservatives) as well as any reversal of that pattern that may have occurred since 2017 (as indicated by the polls).

As the table below shows, the geography of the Conservative performance was indeed rather different in 2014 from that which pertained in 2017. In 2014, the Conservatives performed least well in council areas where the Leave vote was lowest in 2016. However, the party also performed relatively poorly where the Leave vote was highest (not least because this was where UKIP performed most strongly). In 2017 in contrast, the Conservatives performed badly in parliamentary constituencies where the Leave vote had been relatively low, while there was no sign (against the backdrop of a weak UKIP performance) of the Conservatives doing less well in the areas that voted most strongly for Leave. Moreover, the difference between the Conservative share of the vote in the most strongly Leave areas and those where Leave was relatively weak was much bigger in 2017 than it had been in 2014.

Table: Mean Conservative and Labour Share of the Vote in 2014 and 2017 by % Leave Vote 2016

  2014   2017  
% Leave 2016 Con Lab Con Lab
0-42 20.9 29.6 27.8 45.2
42-47 27.2 19.6 39.4 40.9
47-52 28.9 20.2 46.2 39.3
52-57 27.1 20.9 47.8 39.0
57-62 26.4 20.5 47.1 41.9
62-100 23.1 24.0 47.4 43.3

Note: Figures for 2014 are for local authority area. Figures for 2017 are for parliamentary constituency; categorisation of each constituency by % Leave vote in 2016 is based on estimates produced by Chris Hanretty.

This means that if, as the polls suggest is the case, support for the Conservatives falls much more heavily as compared with 2017 among Leave voters than it does among Remain voters, this will not necessarily be reflected in a sharper drop in Conservative support since 2014 in the most pro-Leave areas than elsewhere. Rather, all such a pattern might do is to restore the position in 2014 whereby the Conservatives were not performing well in the most pro-Leave parts of Britain. In short, a pattern whereby the Conservative vote falls by much the same in both Remain voting and Leave voting areas would, in fact, be consistent with the claim that the party has lost ground more heavily among Leave voters since 2017.

In Labour’s case there is, fortunately, much less of a complication. As the table shows, Labour performed best of all on average in 2014 both in those places where Leave performed least well in 2016 and in those where Leave scored best. A not dissimilar pattern was in evidence in 2017. If Labour does lose ground particularly heavily in either very pro-Remain or very pro-Leave areas, it probably will be reasonable to argue that it is a pointer to the possibility that the party has lost ground among Remainers or Leavers (as appropriate) in particular since 2017. Equally, if the party’s vote falls – or simply holds steady – irrespective of the Remain/Leave divide locally in 2016 it will be reasonable to suggest that the party has had difficulty retaining the support it had in 2017 among both Remainers and Leavers. We will have to wait and see on Sunday night which proves to be the case.

Thanks to the BBC, Chris Hanretty, Patrick Heneghan and Matt Singh for help with the data.


5 thoughts on “Two Notes on the Psephology of the Euro-Elections”

  1. Is there a typo in this sentence?

    – Overall, turnout appears to be up on average by four points in those areas where less than 55% voted Leave in 2016, compared with just half a point in those where fewer voted that way –

    I think fewer should read more.

  2. Reblogged this on RogersLongHairBlog and commented:
    An Excellent Thread analysing the ins and outs here By a Young Political Scientist at Oxford, definitely a name to watch.

    also some collaborations on turnout declarations ahead of the vote.

    To put it crudely, Bull Shit is Like Farting everybody likes the smell of their own.

  3. “Overall, turnout appears to be up on average by four points in those areas where less than 55% voted Leave in 2016, compared with just half a point in those where fewer voted that way” – surely the words ‘less’ and ‘fewer’ there are contradictory? Is there a typo and one of those words should be replaced with ‘more’?

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s