If the internet polls are correct, England will vote narrowly to leave the EU but be held in by, principally, Scotland and Northern Ireland. Charlie Jeffery crunches the numbers to assess the territorial implications of the EU referendum vote.
Most analysis of voting intention in the upcoming referendum focuses on the Britain-wide headline figures that most opinion polls produce. These headline figures can disguise important variations in voting intention by UK nation. Britain-wide polls exclude Northern Ireland completely. They also level down what we have now known for some time as quite stark national differences within Britain, especially between a more Eurosceptical England and a more Europhile Scotland.
Are these differences likely to be significant enough so that one or more parts of the UK votes to leave the EU, while one or more parts votes to stay? Most attention on this question has been given to whether Scotland will be pulled out of the EU against its will be sheer weight of numbers in the eventuality of an English leave vote and the associated speculation that this might trigger a second Scottish independence referendum.
But when we last reviewed the available data we raised an alternative question: might we see an outcome in which England is kept in the EU against its will. We return to these questions here, using polling data from the start of 2016 to 31 May 2016.
We are now blessed with good and easily available data from opinion polls of reliable sample sizes helpfully collated on specialist websites in Scotland, Northern Ireland and Wales. John Curtice’s What Scotland Thinks website records 20 polls which asked the referendum question between January and the end of May and Roger Scully’s Elections in Wales website six Welsh polls in that period. Since January a newcomer, LucidTalk has been delivering monthly tracker polls in Northern Ireland, with four now published, filling what had been a notable gap. The averages of these sets of polls are set out in Table 1 (with those saying don’t know or won’t vote left aside to produce simple remain/leave figures). All of the polls in Wales and Northern Ireland and most of them in Scotland were internet polls.
Table 1: Remain and Leave in Scotland, Wales and Northern Ireland, January-May 2016
The figures in Table 1 are a little rough and ready, especially in Wales and Northern Ireland where there have been fewer polls. But the pattern is clear: Wales is pretty evenly split on the referendum question, while Scotland and Northern Ireland show clear majorities for remaining in the EU. Significantly the headline figures hide a polarisation within Northern Ireland, with 80%+ of nationalists/republicans in favour of Remain and 60% or more of unionists in favour of Leave.
It is rare for polls to be conducted in England only. However because of England’s size – 84% of the UK’s population – reliable data can be found in standard Britain-wide polls. A number of polling companies – ICM, Ipsos MORI and Survation – have been disaggregating a figure for England in their EU referendum polls, typically based on 800+ respondents for telephone polls, rather more for online polls. What is interesting is how results from phone and internet polls differ. Table 2 sets this out by giving the averages of seven telephone polls and nine internet polls for which disaggregated results for England were reported from 1 January to 31 May 2016.
Table 2: Remain and Leave by Phone and Internet Polls in England, January-May 2016
The difference is striking. In telephone polls there is a clear 55:45 margin for Remain. In internet polls Leave edges it. This difference in findings of phone and internet polls has been subject of much debate on Britain-wide polling, with proponents of the different methods each claiming they have it right, and expert adjudictors unable to come to a clear conclusion on who is right or wrong.
The stakes are rather high if we project the different England averages in Table 2 alongside the data in Table 1 into a UK-wide calculation of voting intentions. We do this for phone polls in Table 3, where we weight the respective averages by the share of the UK’s population each nation has. This again is rough and ready: the respective averages are generated across different numbers of polls in each place; and the calculations assume a uniform turnout in each part of the UK. But in this case the outcome is straightforward: all parts of the UK (Wales only just) are in favour of remaining in the EU.
Table 3: A UK-wide Calculation Using the Phone Poll Average from England
Table 4 does the same using the average for internet polls in England. In this case the UK as a whole sees a wafer-thin margin in favour of Remain, but with a (slim) English majority for Leave. If this pattern were replicated on 23 June England would indeed be held in the EU against its will.
Table 4: A UK-wide Calculation Using the Internet Poll Average from England
But – if internet polls are a good guide – the margin is tiny, and if the Remain vote in England fell slightly to 48.0% (a 0.6% change), then there would be an overall Remain vote of 49.94%, in other words an extraordinarily tight decision to leave the EU in which Wales (just), Scotland and Northern Ireland would be pulled out of the EU against their will.
Either outcome could bring significant consequences for the internal stability of the UK. If England were in against its will, would we see a mobilisation of English nationalism by defeated Brexiteers, pushing further on the lines of the English campaign the Conservative Party ran in the last UK election? If England swung an overall Leave vote what consequences might there be in a Northern Ireland where 80% or more of one community is in favour of EU membership? And would we see the second Scottish independence referendum that former SNP leader and Scottish First Minister Alex Salmond insists would be a consequence of an English Leave trumping a Scottish remain?
One thing is clear: if the internet pollsters have it right, the UK’s relationship with the EU may not be the only thing in the in-tray of the UK Government on 24 June.