Skip to main content
Coronavirus (COVID-19): Latest updates.


What the Welsh Polls Say (and have said)

27 April 2015

Well, it’s getting close now. There are just ten days to go to the general election. (And yes, I know what you are all thinking: if only campaigns could last longer).

While there will continue to be a plethora of GB-wide opinion polls right up until the eve of poll, here in Wales we are unlikely to have any further ones until the next Welsh Political Barometer poll, which should be published on the evening of 6th May.

You can find substantial detail about the various polls that have been conducted in Wales in recent times in the Opinion Polls section of the Blog; I have also discussed them in numerous posts. But I thought that it might be useful to some blog readers – never say I don’t think of you – to provide some summary information on what the Welsh polls have been saying in relatively accessible and concise form.

So here, first of all, are the average general election poll ratings of the main five parties in Wales between 2012-2014. As we can see, the only substantial trend during those years was the decline in Labour support and the rise of UKIP:


Average Poll Ratings for Westminster General Election, Wales

Party 2012 2013 2014
Labour 51.0 48.3 40.5
Conservative 23.5 22.0 23.5
LibDems 6.0 8.3 6.1
Plaid Cymru 10.5 10.3 12.0
UKIP 5.7 8.3 13.4


These are the polls conducted before the start of the campaign in 2015:


Pre-campaign Welsh Opinion Polls During 2015

Party January January Early March Mid-Late March
Labour 38 37 39 40
Conservative 21 23 25 25
LibDems 7 6 5 5
Plaid Cymru 12 10 10 11
UKIP 13 16 14 14
Greens 6 8 6 5
Pollster ICM YouGov YouGov YouGov


And these are the polls that have been conducted during the official campaign itself:


Welsh Opinion Polls During the Campaign*

Party 31/03/15 15/04/15 30/04/15
Labour 40% 40%  39%
Conservative 27% 26%  26%
LibDems 6% 6%  6%
Plaid Cymru 9% 12%  13%
UKIP 13% 13%  12%
Greens 5% 4%  3%

* All polls conducted by YouGov. Dates listed for polls are dates when fieldwork was completed.


Finally, here is a summary of the main findings from the Ashcroft constituency polls for the five Welsh constituencies that he has looked at. (In the case of Cardiff North he actually polled it twice; I’ve reported the more recent set of figures). Numbers here are for the question asked in these polls that asked people how they would vote in their specific constituency:


Ashcroft Constituency Polls (changes on 2010 result in each constituency)

Constituency (Fieldwork dates) Labour Cons LibDems Plaid UKIP Greens
Cardiff North (July 14) 41 (+4) 30 (-8) 6 (-12) 8 (+5) 12 (+10) 3 (+2)
Cardiff Central (Sept 14) 36 (+7) 17 (-5) 24 (-17) 9 (+6) 9 (+7) 5 (+3)
Brecon & Radnor (Nov 14) 15 (+5) 27 (-9) 31 (-15) 8 (+6) 17 (+15) 1 (+1)
Carmarthen West & South Pembs (Dec 14) 29 (-4) 33 (-8) 4 (-12) 16 (+6) 14 (+11) 3 (+3)
Vale of Glamorgan (Feb 15) 32 (-1) 38 (-4) 4 (-11) 12 (+6) 10 (+7) 3 (+2)



  1. Harry Hayfield

    6% swing to Lab in Cardiff North, 2% swing to Lab in Carmarthen West and 1.5% swing to Lab in Vale of Glamorgan. I hope that this will prove, once and for all, that the concept of a “National Uniform Swing” simply does not happen!

    • Roger Scully

      I don’t think anyone has ever thought we were going to get a uniform swing, Harry. But I still think the concept is useful, in providing us with a baseline measure against which we can judge individual local swings.

  2. Welshguy

    Interesting to once again see those significant swings to Plaid Cymru in individual constituencies not represented at all in national polls. Similarly, they’ve got some promising constituency predictions from

    The party seem to have received a small boost in the past couple of weeks, presumably from Leanne Wood’s TV appearances, but only enough to bring them back up to the kind of levels they were polling a couple of years ago, and nothing to suggest the kind of significant national swing that the individual constituency polls seem to be suggesting.

    The only thing that’s certain is that both sets of polls can’t be right! It will be fascinating to find out which turn out to be the more acurrate.

  3. Jason Morgan

    Difyr unwaith eto!

    With regards Election Forecast, that Welshguy has alluded to above, I’d be interested to hear anything you might have to say on it, Roger. Nobody seems to be sure as to how useful it actually is on a constituency level. I know some people, specifically Plaid Cymru supporters and members on Twitter, have been raving about it (which come election day may make them look quite silly), while others have ignored it completely – I’m a lot more sceptical than the first group but not as dismissive as the second.

    I think the methodology, from what I understand of it, maybe a bit unreliable, but at the same time those updating the predictions are experts in the field and know what they’re doing…


    • Roger Scully

      Well, it’s quite right that the people behind Election Forecast are serious scholars. But I would still urge caution on their individual seat projections, which in a few cases have seemed to throw up strange numbers. This form of election forecasting model is still quite a new business, and something of a work in progress, I think.

  4. Phil Davies

    Can you give us a few details of the Welsh Barometer poll next week Roger? Sample size, medium, dates, etc.? Assume it’s same as usual but can you confirm?

    • Roger Scully

      We haven’t confirmed all the questions yet, Phil. Should be the normal sample size (something over 1000). Will be published on Wednesday evening; sampling will be done earlier that day and the previous day – basically, as late as possible before the election, in order to guard against error because of late swings.

  5. Welshguy

    Frustrating that Ashcroft or someone hasn’t done a constituency poll for a Plaid Cymru seat or target (or indeed, any more Welsh seats, to see whether those constituency results are just a fluke or whether electoral forecast are really onto something). As I’ve mentioned before, the only way the constituency results could be reconciled with the national polls would be if PC is seeing a big drop in its heartlands to compensate for a rise elsewhere – although there’s no other reason to expect that to be the case.

    Is it possible there’s a flaw with the Welsh Political Barometer’s polling methodology? We haven’t really been able to ascertain how accurate it via an election yet: the Wales crossbreaks of UK polls are worthless in this regard, so there’s nothing really to compare the WPB with as nobody else does Wales-only polls. This is not a criticism – I really appreciate that the WPB exists at all! – but it’s obviously in everyone’s interest that polls be accurate.

    I’ve read once that polls in Wales do not weight for Welsh speaking (should be 20% of the sample, and up to 60% in individual seats), something which could potentially be systematically producing lower poll results for Plaid Cymru?

    I’m inclined to agree with Jason that the electoralforecast results are probably overly favourable to Plaid. On the other hand, again, as I’ve said before, there are plenty of reasons a PC surge of some sort would be unsurprising (disaffected Lib Dems not wanting to vote Labour; SNP/Green/UKIP surges framing the debate away from the Lab-Con dichotomy; Scottish independence referendum inspiring nationalists elsewhere; Leanne Wood as a more appealing leader than previous incumbents; increased visibility from the TV debates).

    It will be fascinating to see what happens on the night and as far as I’m concerned this remains the most interesting unanswered question of the election!

    • Roger Scully

      Yes, it is a tad frustrating. But sadly Wales is not at the top of the agenda for those primarily concerned with the UK-wide election outcome.

      The Barometer polls could be a little out, of course; I did comment in a post some months ago that Plaid seemed to do a little better in telephone polls than online ones. (One colleague, semi-seriously I think, suggested that the dire state of rural broadband in some of Plaid’s strongest areas might partially account for this!). However, although the Barometer polls were only launched in December 2013, YouGov were doing some polls for ITV prior to 2010 and the 2011 Assembly election. You can see the details in the Opinion Polls section of the Blog. In 2010, and on the constituency vote in 2011, YouGov did have Plaid a little lower than their actual performance – though not by much. On the other hand, in 2011 they got Plaid pretty much dead-on for the list vote, while their final poll pre the Euro-election last year was also very close for Plaid. Overall, i think we can say that YouGov certainly do not have a record of over-estimating Plaid’s vote share, but there’s not much evidence of them under-stating it either.

      Furthermore, with opinion polls just as much as with investments, past performance is no guarantee of future results!

Leave a reply

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