So How Did We Do?

As most of you will be aware, the team behind the Welsh Political Barometer polls ran the first-ever ‘on-the-day’ poll at a National Assembly election this year. It was something of an experiment – so how successful was it? Here are some relevant numbers (plus, to help give everything some context, the vote share figures for 2011):


Constituency Vote

Party 2011 Result On The Day Poll 2016 Result Poll ‘Error’
Labour 42.3 33 34.7 1.7
Plaid 19.3 19 20.5 1.5
Conservative 25.0 21 21.1 0.1
UKIP 16 12.5 3.5
Lib-Dems 10.6 8 7.7 0.3
Others 2.8 3 3.5 0.5



Regional Vote


Party 2011 Result On The Day Poll 2016 Result Poll ‘Error’
Labour 36.9 30 31.5 1.5
Plaid 17.9 21 20.8 0.2
Conservative 22.5 19 18.8 0.2
UKIP 4.6 16 13.0 3.0
Lib-Dems 8.0 6 6.5 0.5
Greens 3.4 4 3.0 1.0
Others 6.7 4 6.5 2.5


In short, I think we did pretty well. UKIP support was over-stated, and that is something that will require some further investigation. All the other parties were estimated within two percentage points of the true figure on both ballots. And the on-the-day poll got the main narrative issues right:

  • Significantly declining Labour support since 2011
  • A large rise in UKIP support since 2011
  • A close fight between Plaid and the Conservatives for second, with Plaid probably coming out on top
  • The Liberal Democrats declining further from 2011
  • The Greens wholly failing to make a serious breakthrough


I’ll evaluate the UNS seat projections – and how they would have fared against other methods – later on. But my initial, and tentative, conclusion is that our experiment was at least a qualified success. And I think my colleagues at ITV Cymru-Wales (who put up the vast majority of the money for the On-the-Day poll) and the very hard-working team at YouGov who have worked on all our Welsh polling (led by Adam McDonnell and Laurence Janta-Lipinski) both deserve lots of credit for this.


  • gwil williams

    Would Neil MCEvoy have won the regional list seat if Leanne Wood had not won the Rhondda?

  • Dylan Foster Evans

    So had Leanne Wood not won the Rhondda, would that have meant David Melding not getting a regional list seat?

  • John R Walker

    ‘UKIP support was over-stated, and that is something that will require some further investigation.’

    Not really – a lot of UKIP list votes (including mine…) went to Abolish the Welsh Assembly Party and that’s UKIP’s fault for abolishing their abolish the Assembly policy from earlier years. Entirely predictable!

    • Roger Scully

      Quite possibly, John – but that’s something we will need to investigate. I don’t think we can just assume that is generally the case.

      • George Lancett

        Well I would certainly have voted for UKIP if I had not voted for Abolish’

  • J McKenzie

    Very impressive Mr Scully, Well done.

  • J.Jones

    I would argue, with John Walker above, that to leave “Abolish the Welsh Assembly” out of your list and include the Green party is perverse. UKIP lost a significant number of voters when it “went native” and decided to be just another party of Devocreep and soft nationalism. As a result they will fade into obscurity aided no doubt by the contempt that they engendered by bringing Reckless and Hamilton back from obscurity. Surely no one in Wales will forgive them for that.

    ATWAP out polled the LibDems as well as the Green party in the North Wales regional list as well as South Wales East but out polled the Green party everywhere. If there is one thing that the polling showed it is that applying a generally accurate Wales-wide poll and extrapolating to predict seats is the weakness. Llanelli didn’t change hands and I didn’t see any poll that predicted that it wouldn’t fall to Plaid. RCT did change hands and I didn’t see any poll that predicted that.

    If there’s one area that should be concentrated on its not national opinion it’s prediction of national political representation. I would also argue that polling itself leads to bland, conviction-less consensus in politics in Wales. Find out what the majority want and then adopt that policy or fail to oppose it is a receipt for uniformity and stagnation.

  • Lyn Thomas

    I don’t think that polling can be “blamed” for the results, and yes I suspect that the higher vote for the Abolish the Welsh Assembly Party in the Mid and West Wales region was a reaction to the UKIP list choices. What is still shows is that despite some noises about using the list vote tactically Labour (and other party associates) largely still vote the same way on the list as they do in the constituencies. I would love to see a breakdown of the list vote by constituencies if that is available?

    My comments are that with a resurgent UKIP Plaid did actually better than other parties, with a modest increase in its total vote percentage and an actual gain in constituency seats. The electoral system once again gave a bonus to the largest party and I suspect the 40:20 split will become untenable when the electoral system is devolved and the Westminster seats are reduced to 29.

  • Christian Schmidt

    1) I struggle to see how you could possibly poll well for something like the ‘Abolish the Welsh Assembly Party’ – how could you measure in a poll whether the voters that wish to abolish the assembly will bother to vote?
    2) I don’t think many UKIP voters will be that engaged with the nuances of UKIP’s position towards the assembly, but as they are generally seen as having a rather negative view on devolution I think it is fair to assume that the difference between UKIP polling and result is best explained by the rather good result for the Abolish the Welsh Assembly Party
    3) I think the Rhondda (and Llanelli and Bleanau Gwent) result(s) are probably best explained by the lack of strong trends / strong issues in the campaign. In such a case individual candidates matter more. As Labour’s win was not really in doubt, generally Labour-leaning voters in the Rhondda could safely vote for Leanne Wood – who as been undeniably impressive in attacking Tory-policies (and UKIP) over the last year or so.

    • Ddirpytnop

      The ATWA Party did not have candidates standing for the constituency seats but the differences between the polling figures for UKIP and it’s actual vote percentage were similar for the constituency and the regional votes. Therefore, there has to be something more behind the polling error. It would be interesting to know who the ATWA voters were supporting in the constituency contests.

      • Sian Caiach

        One unusual point I noted in Llanelli was that we had well over a hundred “spoiled” ballot papers which were just left blank rather than defaced or voting for more than one candidate etc. I’ve never seen that many blank ballot papers before. Did anyone else at a count notice the same?

  • Richard Owen

    Diolch yn fawr iawn am yr holl boliau a’r dadansoddi medrus dros y misoedd diwethaf.

    Meanwhile, I’m sure you’ll be giving a stern telling off to that undependable character, UNS. He really can’t be trusted, can he!

  • Sian Caiach

    I was very impressed by the vote of the Abolish the Welsh Assembly Party, known locally here in Carmarthenshire as the”Kipper Splitters” who appear to be former UKIP supporters outraged at the 2014 u turn by UKIP on the matter of supporting and standing for the Welsh Assembly. Assuming that the majority of their votes come from UKIP sympathisers, on the lists, could they have prevented, as they surely intended, UKIP members from getting list seats by splitting the UKIP, list vote?

Leave a Reply

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