The EU Referendum: the Welsh Verdict

Well, as Margaret Thatcher once said, “it’s a funny old world”. The implications of the UK’s vote to leave the EU will be many, and at least some of them may be profound. There will be plenty of time to assess the broader implications, as they unfold. For now, I’m just going to look at some of the information we have regarding how people voted.

The performance of the pollsters in this referendum will be subjected to detailed scrutiny and, in at least some cases, criticism. But on one matter at least, they appear to have been absolutely correct. With regard to the different nations of the UK there had been a consistent pattern in the polls over the last few months: Scotland was clearly pro-Remain; Northern Ireland was also pro-Remain, although somewhat less so; and things looked much closer in both England and Wales, with Wales tending to be marginally more pro-Remain than England. And that is exactly how it turned out. These are the figures for voting (and turnout) by nation:


Nation % Remain % Leave % Turnout
Scotland 62.0 38.0 67.2
Northern Ireland 55.8 44.2 62.9
Wales 47.5 52.5 71.7
England 46.6 53.4 73.0


Strikingly, the two nations that voted for Remain, Scotland and Northern Ireland, have both actually become significantly more pro-EU since 1975. Then, only 52.1% of people in Northern Ireland and 58.4% in Scotland voted for Remain. Wales and England have both travelled a long way in the other direction: 41 years ago, 64.8% voted Remain in Wales, and 68.7% in England.

It is also interesting that turnout was lower in the two pro-Remain nations. This didn’t change the result – even if both Scotland and Northern Ireland had had the same turnout rate as England, then Leave would still have won. It may be simply that the referendum engaged more people in England and Wales. But there may also have been some element of voter fatigue – and also activist fatigue – in Scotland, in particular, which has had a lot of campaigning and voting over the last three years.

What about the vote in Wales? How did Wales come to vote Leave? Answering that question in detail will take plenty of further analysis. But we can see which areas voted Remain and Leave. I’ve listed the 22 Welsh local authorities here in order of how they voted, from the most pro-Remain to the most pro-Leave:


Local Authority % Remain % Leave % Turnout
Cardiff 60.0 40.0 69.6
Gwynedd 58.1 41.9 72.3
Ceredigion 54.6 45.4 74.4
Vale of Glamorgan 50.7 49.3 76.1
Monmouthshire 50.4 49.6 77.7
Ynys Môn 49.1 50.9 73.8
Swansea 48.5 51.5 69.5
Carmarthenshire 46.3 53.7 74.0
Powys 46.3 53.7 77.0
Rhondda Cynon Taf 46.3 53.7 67.4
Conwy 46.0 54.0 71.7
Denbighshire 46.0 54.0 69.1
Bridgend 45.4 54.6 71.1
Newport 44.0 56.0 70.2
Flintshire 43.6 54.6 74.8
Merthyr Tydfil 43.6 56.4 67.4
Neath Port Talbot 43.2 56.8 71.5
Pembrokeshire 42.9 57.1 74.4
Caerphilly 42.4 57.6 70.7
Wrexham 41.0 59.0 71.5
Torfaen 40.2 59.8 69.8
Blaenau Gwent 38.0 62.0 68.1


So only 5 of the 22 Welsh local authority areas voted for Remain. One of them most striking things about this result is that every single one of the valleys authorities voted for Leave – many of them in very large numbers. It must be of great concern to Labour, the dominant political party in the valleys for decades, that these communities so strongly rejected the party’s message to support the UK’s continuing membership of the EU.

But then a notable feature of the results is just how many prominent Welsh politicians, of various parties, found themselves on the wrong side of their own local communities. Among those who supported Remain, First Minister Carwyn Jones saw his Bridgend area vote to Leave; Plaid Cymru leader Leanne Wood, and Labour Shadow Cabinet member Chris Bryant, saw their advice rejected by the voters of the Rhondda (and the rest of Rhondda Cynon Taf); and Cabinet member Stephen Crabb took a different side from the majority of people in his native Pembrokeshire.

Meanwhile, on the pro-Leave side Conservative leader Andrew RT Davies also found himself on the opposite side of the majority of his fellow voters in the Vale of Glamorgan, as did David Davies in Monmouthshire. But that was not the case for all politicians. Former Welsh Secretary David Jones would have been delighted that the majority of voters in Clwyd followed his advice in supporting leave, while current Welsh Secretary Alun Cairns was (just) in the pro-Remain majority in the Vale of Glamorgan. And UKIP Assembly leader Neil Hamilton was also supported by the majority of voters in his home base – as 52.5% of voters supported Leave in Wiltshire.

How does voting behaviour in this referendum compare with previous ones? Well, the 1975 European referendum results were counted in the old counties of Wales. It may take some time to assemble the results for those – when I have them I will share them with you. Just as a bit of fun, though, I investigated the statistical correlation between voting in this referendum and in the 1997 and 2011 devolution referendums. And, put simply, there is no relationship! The correlation coefficient between 2016 voting and 1997 is 0.012 – in other words virtually no relationship at all. And the coefficient for 2016 compared with 2011 is 0.000 – absolutely no relationship whatsoever! So whatever else it was related to, voting in this referendum was not related – at the aggregate level at least – to how people voted on Welsh devolution either in 1997 or 2011.


Update, 27/06/16:

A brief addendum to last week’s post on the voting patterns in the EU Referendum.

In 1975, votes were also counted by local authority area. However, in those days there were only 8 main Welsh local authorities, rather than the 22 that we have today. So comparing the results of today with those of 1975 could not immediately be done.

However, my friend Harry Hayfield has come to our psephological rescue. (Thank you, Harry!). He has computed the 2016 results into the local authority boundaries of 1975. That process generates the following figures:


Old Local Authority Remain Leave % Turnout % Swing to Leave, 75-16
Clwyd 120,791 (44%) 153,948 (56%) 72.1 25%
Dyfed 98,732 (47%) 112,567 (53%) 74.1 21%
Gwent 132,230 (43%) 172,468 (57%) 71.3 19%
Gwynedd 54,135 (55%) 44,998 (45%) 72.9 16%
Mid Glamorgan 100,270 (46% 119,503 (54%) 68.6 11%
Powys 36,762 (46%) 42,707 (54%) 77.0 28%
South Glamorgan 138,469 (57%) 103,444 (43%) 71.4 12%
West Glamorgan 90,958 (46%) 104,937 (54%) 70.3 16%


When one compares voting patterns in 1975 and 2016 on these boundaries, the correlation (for the technically minded, the Pearson’s r coefficient) between the % Remain/Leave votes in 1975 and in 2016 is 0.39. In other words, there is something of a relationship between voting patterns in the two EU membership referendums. But it is far from an overwhelmingly strong relationship.


  • Robert Tyler

    The only smile I’ve cracked so far today came from your comment on Hamilton.

    • nunu

      They are comparing one year. Stupid. Boston has seen huge rise in immigrants.

  • Nigel Marriott

    Hi Roger,

    Well it’s been an interesting time for pollsters and elections!

    Prior to the vote, I had wanted to build a model based on demographics to predict the council by council declarations. I ran out of time and didn’t have all the data I wanted either so gave up. After the first 5 declarations, it suddenly dawned on me that the declarations reflected the 2014 EU election results and after 40 or so declarations, I built a quick and simple model to predict all remaining councils and it worked quite well with nearly all councils leave within 6% of the actual vote.

    Today with the full results, I’ve rebuilt the model correlating the Leave vote with the “insurgent votes” of the 2014 EU elections i.e. the shares for UKIP, BNP, Green, SNP & PC for Great Britain i.e. ignoring established parties. R-squared is 92% and standard error is +/-3%. Higher r-squared are possible but this has the virtue of simplicity and is regionally independent for the most part.

    I am happy for you to publicise this thought (especially if you can credit me!) as I haven’t seen anyone else say this but basically the Brexit vote could have been predicted back in 2014. To my mind this tells me that the vote yesterday was the continuation of a trend rather than a one-off.

    • Andrew Draper

      Hi Nigel

      If this is a continuation of an existing trend, where does it go next, do you think? Any predictions?

      • Nigel Marriott


        It’s very hard to say what the future will be as I think we need to understand the trend first before trying to extrapolate.

        For myself, I have been worried about how an increasing section of the population have had their views ignored and trashed by the political class over the last 20 odd years, a trend I am now calling Project Sneer. This is exacerbated by the increasing trend for our representatives to be ex-political researchers rather than “real” people. The tragic murder of Jo Cox has the potential to allow us to start seeing our politicians as “real” people again but in turn it requires politicians to abandon Project Sneer. I am very much in the camp that the voters are always right and it is the first duty of a politician to accept the people’s verdict rather than ignoring it or belittling it or asking them to vote again like the EU. I wish we had more first class politicians like the late Tony Benn, whose views might have been miles away from mine, but he always had my respect in the way that he accepted the people’s verdict. Look up his reaction to the 1975 referendum when he was on the losing side.

        The book “Revolt on the Right” is well worth a read as the authors are academics who rely on a lot of data to support their inferences. They clearly highlighted the danger of the trend I am describing but it is also clear that politicians, especially the Labour party and the EU, have chosen to ignore this danger. My hope now is that they will no longer do so.

  • Paul Rowlinson

    The 1975 results were:
    Clwyd 69.1 30.9
    Dyfed 67.6 32.4
    Gwent 62.1 37.9
    Gwynedd 70.6 29.4
    Mid Glam 56.9 43.1
    S Glam 69.5 30.5
    W Glam 61.6 38.4
    Powys 74.3 25.7
    Source:Etholiadau’r Ganrif, Beti Jones (1999) p.156

  • J.Jones

    Here’s a correlation that explains the “leave” voting by LA:-

    Ceredigion 7 10 19 84
    Powys 7 10 21 83
    The V. of Glamorgan 7 11 19 82
    Gwynedd 8 11 25 81
    Monmouthshire 6 13 18 81
    Conwy 7 13 20 80
    Cardiff 8 11 13 80
    Isle of Anglesey 8 12 20 79

    Can you guess what it is?

    • J.Jones

      Give up? It’s the graded list of the percentage of adults who hold educational qualifications. The final percentage; 84% in Ceredigion, refers to the percentage holding GCSE level 2 Qualifications or higher. Of course a similar list from the Welsh index of multiple deprivation would have pretty well the same candidates with low level’s of deprivation, with Cardiff as the only exception. Cardiff however has the highest level of adults educated to degree level.

      So for those who would say that “Leave” is a triumph for ignorance the evidence is persuasive. Other factors are obviously in play of course; Gwynedd and Ceredigion bastions of Nationalist and Liberal politics respectively but also with a high proportion of students. Powys with little poverty but sparsely populated and rural. Anglesey voted “leave” but very marginal.

  • Jonathon Harrington

    I am fascinated that the turnout in Wales is far higher than ANY elections held for the Welsh Asembly which has never attracted more than 50% of the electorate. What does this indicate? Do the Welsh regard membership of the EU as being more significant than any influence the WA has on their lives?
    How does Carwyn Jones threat to speak directly to Brussels if the vote went as it did now sit with voters?
    Personally I think we could do with keeping things simple by reverting to the pre 1997 arrangements with Westminsterbut would welcome any other ideas. JAH

Leave a Reply

Your email address will not be published. Required fields are marked *