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


Two new polls in Wales!

22 April 2021

You wait all campaign for a Welsh poll – and then two come along within one week! With only two weeks until the Senedd election, and with some voters now casting postal ballots, today’s Welsh Political Barometer poll provides the very latest guide to the fortunes of the political parties in Wales. Our new poll continues to show a close, three-way fight between the main parties. But their fortunes do seem to have changed in the last month. Meanwhile, a poll published earlier this week by Opinium (for Sky News) paints a rather different picture.

As always, the new Barometer poll asked about voting intentions for both the constituency and regional list ballots for the Senedd. And as with all our recent Barometer polls, we included 16 and 17 year olds in the sampling, given that they will be able to vote next month. Here are results from the new poll for the constituency ballot (with shifts in support for each party since the previous Barometer poll, published last month, indicated in brackets):

Labour: 35% (+3)

Conservatives: 24% (-6)

Plaid Cymru: 24% (+1)

Reform UK: 4% (+1)

Liberal Democrats: 3% (-2)

Greens: 3% (+1)

Abolish the Assembly: 3% (no change)

Others: 3% (+1)

Our new poll suggests a large rise in Labour’s lead over the past month, and a substantial fall in Conservative support. Labour’s advantage has apparently gone from two points to eleven since the campaign for the Senedd began. As always, we should be cautious about over-interpreting a single poll, and the changes that it suggests. It may be that the previous Barometer poll somewhat over-stated the Conservative position; this one might be under-stating it. But our latest evidence does indicate Labour’s position as the largest party in the Senedd to be rather more secure than did March’s poll. Plaid Cymru will also be encouraged by the our new findings: while their support has not changed significantly over the last month, our results suggest that they remain very much in contention with the Conservatives for second place. That did not look likely in last month’s poll.

A uniform swing projection of the changes in party support since May 2016 suggested by this poll indicates that three constituency seats would change hands. Despite Labour’s strengthened position in this poll, all three of those seats are Labour ones: with the Vale of Glamorgan projected to be gained by the Conservatives, and both Llanelli and Blaenau Gwent to be captured by Plaid Cymru. However, as I have emphasised previously, we should be cautious in interpreting these figures: all three projected results (as well as those of other constituencies projected to be retained by Labour) suggest very narrow margins of victory or defeat, while problems in recent years for Plaid’s constituency parties in both Llanelli and Blaenau Gwent may make such gains less likely than the bare arithmetic suggests.

Now what about the regional list vote? The new Barometer poll has the following results (with changes since last month’s poll once more indicated in brackets):

Labour: 33% (+2)

Plaid Cymru: 23% (+1)

Conservatives: 22% (-6)

Abolish the Assembly: 7% (no change)

Greens: 5% (+2)

Liberal Democrats: 4% (no change)

Others: 6% (+2)

Just as for the constituency vote, we see here a modest improvement in the Labour position that – when combined with a larger fall in reported Conservative support – produces a big increase in the apparent Labour lead: from three points to ten in one month. We should again be cautious about such changes, and the rise in Labour support indicated in our new poll is well within the standard ‘margin of error’. But these figures are consistent with those reported above, and indicate a similar picture: with Labour in a clear lead and the Conservatives and Plaid very closely contesting for second.

Our new poll is also the third in a row to estimate support for the Abolish the Assembly party at seven percent. This is right at the cusp of the support level that might win them regional list seats in the Senedd. The Greens are also, according to our new figures, not very far away from such a level of support. Taking into account the projected constituency results mentioned above, and once more assuming uniform national swings since 2016, the new Barometer poll projects the following results for the Senedd’s regional list seats:

North Wales: 2 Plaid, 1 Conservative, 1 Abolish the Assembly

Mid and West Wales: 2 Labour, 1 Plaid, 1 Abolish the Assembly

South Wales West: 2 Conservative, 2 Plaid

South Wales Central: 2 Conservative, 2 Plaid

South Wales East: 2 Conservative, 2 Plaid

From these figures we generate the following overall projected result for the Senedd:

Labour: 26 seats (24 constituency, 2 regional)

Plaid Cymru: 17 seats (8 constituency, 9 regional)

Conservatives: 14 seats (7 constituency, 7 regional)

Abolish the Assembly: 2 seats (2 regional)

Liberal Democrats: 1 seat (1 constituency)

Such a result would equal Labour’s worst-ever performance in a Senedd election (in 2007) and would also equal Plaid Cymru’s best-ever result (in 1999). But we should be very cautious about such projections. The final list seats in most regions come down to very small numbers of votes, and changes in party support that are well within the standard ‘margin of error’ could generate rather different figures.

What we can say with some confidence, as we enter the last two weeks of campaigning for the Senedd election, is that Labour are in the lead. Our new poll also suggests a very close contest for second place between Plaid Cymru and the Conservatives. There continues to be no good news at all from our Barometer polls for the Welsh Liberal Democrats. But for the Abolish the Assembly party, and maybe also the Greens, winning some representation in the next Senedd does now look like a realistic ambition.

And what about the Opinium poll? This was published on Tuesday; sampling was conducted online over quite a long period (9-19 April), with quite a large sample (2,005 respondents); unfortunately 16 and 17 year olds were not included.

Opinium’s results certainly paint a very different picture to those from the Barometer poll. Here are their numbers for the constituency vote (given the lack of previous Opinium polls in Wales I cannot report changes from their previous estimates):

Labour: 40%

Conservatives: 30%

Plaid Cymru: 19%

Liberal Democrats: 4%

Others: 7%

These numbers are clearly rather different from those produced by YouGov, with Labour and the Conservatives both higher and Plaid Cymru rather lower in support. The one thing that both polling companies definitely agree on is that the Liberal Democrats are performing very poorly. But while Opinium estimate the parties’ support at different levels, they also agree that Labour are roughly ten points ahead of the competition. This does somewhat increase our confidence that the sort of Labour lead suggested by the latest Barometer poll may be approximately correct.

Once more applying uniform national swing since May 2016, these figures for the constituency vote project only two seats to change hands: they suggest the Conservatives narrowly gaining the Vale of Glamorgan and Vale of Clwyd from Labour.

On the regional list vote, Opinium once more offer a very different picture to YouGov:

Labour: 38%

Conservatives: 27%

Plaid Cymru: 19%

Greens: 5%

Liberal Democrats: 4%

UKIP: 2%

Others: 5%

As with the constituency vote, Opinium place Labour and the Conservatives about five points higher than do YouGov, and Plaid Cymru about five points lower. That is curious – not least because, as I have discussed before, previous evidence had suggested that YouGov tended to estimate Plaid Cymru support a little lower than did other polling agencies. However, there are once again some points of agreement with YouGov: Labour are about ten points in the lead; the estimate for Green support is exactly the same as YouGov’s; while there is again agreement on the dismal fortunes of the Liberal Democrats. Frustratingly, though, Opinium did not list Abolish the Assembly as an option for their respondents. The inclusion of Abolish as an option would, I suspect, have pushed reported support for both UKIP and the Conservatives somewhat lower. But we have no way of knowing for sure.

Taking into account the projected constituency results, and once more using the assumption of uniform national swing since May 2016, the Opinium poll projects the following outcomes for the regional list seats:

North Wales: 2 Conservative, 1 Plaid, 1 Labour

Mid and West Wales: 3 Labour, 1 Conservative

South Wales West: 3 Conservative, 1 Plaid

South Wales Central: 3 Conservative, 1 Plaid

South Wales East: 2 Conservative, 1 Plaid, 1 UKIP

This then generate the following overall projected outcome:

Labour: 29 seats (25 constituency, 4 regional)

Conservatives: 19 seats (8 constituency, 11 regional)

Plaid Cymru: 10 seats (6 constituency, 4 regional)

Liberal Democrats: 1 seat (1 constituency)

UKIP: 1 seat (1 regional)

Polls apart, indeed! While Opinium are a deservedly respected polling agency, several factors (their lack of previous experience in conducting polls specifically in Wales; the excellent track record of YouGov’s polls in recent Welsh elections; the lack of sampling among 16 and 17 year olds; and the lack of inclusion of Abolish the Assembly as a regional list vote option) all lead me to suspect that this week’s Barometer poll is more likely to be closer to the true picture of party support at present. But we should certainly not ignore the Opinium findings. And the commonalities in findings between the two agencies (the size of the Labour lead; the level of support for the Liberal Democrats; regional list support for the Greens) should increase our confidence in those findings.

UPDATE 04/05/21: I must thank Iwan Standley who has noticed a couple of errors in the regional seat projections from the last Barometer poll. I’m afraid that I (somehow – I am not quite sure how I managed this) ommitted the projected Plaid seat gains in Llanelli and Blaenau Gwent from the divisor for their regional list seat projections. I am very grateful to Iwan for noticing this, and pointing it out.

The revised seat projection for Mid and West Wales, taking into account the projected outcome of Llanelli, therefore should be: 3 Labour, 1 Abolish the Assembly. The revised projection for South Wales East is 2 Conservative, 1 Plaid, 1 Abolish the Assembly. This gives an overall seat projection of:

Labour: 27 seats (24 constituency, 5 regional)

Plaid Cymru: 15 seats (8 constituency, 7 regional)

Conservatives: 14 seats (7 constituency, 7 regional)

Abolish the Assembly: 3 seats (3 regional)

Liberal Democrats: 1 seat (1 constituency)

Apologies to everyone for this mistake; I can only assume that the excitement of Wales seeing two polls in a week led me astray!

The Welsh Political Barometer poll, for ITV-Cymru Wales and Cardiff University, had a sample of 1,142 Welsh adults aged 16+ and was carried out online by YouGov from 18-21 April 2021.


  1. May Watson

    Many thanks.
    Do you think that UKIP badging themselves as “Scrap the Assembly” might get them some unexpected support? And also, Abolish’s position at the top of the ballot with no logo might decrease their support? If so, a split between UKIP and Abolish might cancel each other our and result in no seats for either.

  2. Jac Pro-biotic

    Interesting polls here. One wonders whether the constant drip-feed of stories about incompetence and corruption from the Westminster government have (at last) begun to dent the Tory vote.

    The area where there is most to play for is the bottom end of the regional list. The Green Party could benefit from both the split in the hard right vote, and from Labour voters (in particular) realising that they have a choice with their regional vote. If, say, one in ten of these Labour voters in the south Wales regions (where Labour is absolutely NOT going to win a list seat) chose to support the Greens with their second vote then things would get very interesting. The Greens would come ahead of the far right parties and should do enough to get the fourth regional seat.

    One key point here is education about how the system works. The drawback is that too many people do not understand the mechanics of how the d’Hondt system allocates the seats, and so Labour voters in south Wales end up throwing their second vote away.

  3. Nathan NG

    The standing problem with all welsh polling is that the regularity of polling is far too infrequent to effectively highlight opinion polling. Having followed welsh politics closely recently, the sense that I get is that the polling data does not fully reflect the actual feeling of voters.

    On the basis of a five poll average the vote shares look very different:

    Constituency –
    Lab = 35.8
    Con = 27.2
    Plaid = 22.4
    Lib Dem = 4

    Regional –
    Lab = 33.6
    Con = 24.8
    Plaid = 22
    Lib Dem = 3.4
    AWAP = 5.4 (excluded from opinium poll)
    GRN = 4.2
    Reform = 0.8 (excluded from 2 of 5 polls)
    UKIP = 1.8 (note excluded from latest yougov poll)

    I haven’t calculated what that would look like in terms of seats but it does present a rather different picture than any one poll suggests and is probably more indicative of the current public mood than a single poll might suggest regardless of how accurate YouGov polling has been to date.

    I think we would have a much more accurate picture of put thinking if we had more regular polls such as in Scotland which appears to have a new poll on voting intentions once every few days, which in of itself highlights a recurring issue of welsh politics around engagement, something that is then reflected in the level of media coverage (no welsh based newspapers in wide circulation or specialised TV programming, ITV’s welsh leader’s debate was not accessible through the internet, whilst live), the nature of policy debates, voter turnout. The turnout issue does misrepresent the actual level of support for parties in Wales, with Welsh Labour and Plaid historically benefitting from poor turnout amongst Conservative voters despite the fact that data does suggest that in terms of actual support the number of Plaid voters is significantly smaller than Conservative or Labour voters.

    So the structural inefficiencies of the regularity of polling and the media coverage in Wales does mean that actual sentiment is probably not properly represented in the polling and even the 5 poll average that I have calculated probably does not map as well as it should on to voters feeling.

  4. Geraint

    A million wasted Labour votes

    The previous post is right. There is a problem with voters not understanding the voting system. In the three southern regions, South Wales Central, South Wales East and South Wales West the Labour party has gained over a million votes since 1999 which has resulted in not a single Labour candidate being elected. I would suspect that a very high proportion of these voters would happily vote for other parties to stop Tories and other right wing parties being elected if they understood the process more clearly.

  5. AP

    It looks like they weighted the Opinium poll results based on how they voted in the previous Westminster (not senedd) election. Thats an odd thing to do given that since 1999 there has been consistently different voting patterns in both, with Plaid Cymru in particular polling better in the senedd than in Westminster. If they’re weighing by the Westminster pattern that would be expected to harm Plaid.

  6. Ddirpytnop

    Are we sure that Opinium weighted by previous voting in the Westminster election? That would inevitably have skewed results. Mind you, failing to sample 16-17 year olds is pretty daft too. It’s almost as though they have never before conducted a poll for an Assembly/Senedd election. Oh, hang on…

  7. Andy S

    There appears to be a trend for Greens in both Wales and Scotland to come ahead of the Liberal Democrats on the list and they are tied in London at 11%. That could make for some very interesting local election results ion England.

  8. Jac

    The more far left/right separatist parties the better – split the vote

  9. AP

    Ddirpytnop – I’m not sure what the exact methodology was but the poll included a question on how they voted at Westminster but not at the Senedd. It would be odd for them to ask the question but not use it to weigh; regardless they can’t have weighted by past Senedd vote as they didn’t ask.

  10. david daviesp

    from dd
    Satisfactorily Labour are pulling ahead mostly because their members and supporters canvass extensively whereas
    the Tories rely on financial donations in order to provide written propaganda which is less work for their fewer supporters.
    Obvious admiration for Mark Drakeford is shining through too. Can’t wait for Labour’s forming the next government; that
    Andrew RT Davies is a disaster is plain to see but who else do they have?

  11. Christian Schmidt

    For the Greens I think the big question is how concentrated their vote is. Previously it wasn’t much, with little difference between regions and different regions being the best. Their polling is stronger, but if their gains are equally distributed it doesn’t look enough in my view. If their gains are particularly concentrated in 1-2 regions (Cardiff/SWC?) they could gain a seat…

    With Abolish i think the key issue is establishing how many of those who dislike the Senedd will turn out to vote. I must admit i was surprised by UKIP in 2016, but it was just before the EU referendum when Farage was in the UK news every day. With Wales not having that much of its own media, how much are low-likelihood voters aware of Abolish if they don’t participate in a poll?

  12. Mike Smith

    I am afraid opinion polls in Wales are always pointless and meaningless. They all disagree and frankly it is because Wales is now so disjointed you would have to probably poll about 10000 people throughout Wale to get ay real sense of opinion. It is good, cheap clickbait fodder for the local media but that is about all.

  13. John Ogilvie

    In response to the comment by david daviesp, I note that the YouGov poll has Labour on 26 seats, a fall of three compared to 2016. david daviesp sees this as a sign of voter admiration for the Labour leader. On the other hand, the YouGov poll has the Conservatives on 14 seats, a rise of 3 seats compared to 2016. And the Opinium poll has the Conservatives gaining 8 seats. david daviesp sees Mr Andrew R T Davies as a disaster. david daviesp should get a job as a public relations officer for the Labour Party in Wales.
    I live in a Scottish constituency which was rock solid Labour until the SNP gained it in 2011. So far I have had no Labour canvassers calling at my door this year. Indeed, over the 38 years I have lived at my present address I have only been canvassed once by anybody from the Labour Party and that includes all elections. So much for Labour members and their supporters canvassing extensively.
    Incidentally while I was an undergraduate at Aberystwyth I used to attend meetings of the Labour Party and was a Committee member of the Fabian Society. But I am an older and much wiser man, now.

  14. gary smith

    It may take yet another election to break the Labour grip on Wales. It has been a sad track record as Wales is simply so much poorer compared with England. Brexit could make wales poorer still BUT the main problem is neither Brexit nor Covid 19 but rather endemic corruption in Job allocation. Far less professionals operate in Wales NHS and in Teaching and Education and in the Assembly as Labour is a ‘jobs for the boys Party’ all so sad. Chronic Trde UNions still hold influence. Why is S.Kinnock the MP in Port Talbot knowing so little abiut the Steel industry and why and how did N.G Kinnock get ANYWHERE except losing TWO important UK elections for Labour and then enhancing and promoting Tony Blair who lied the whole Uk in a deadly and very costly War against a regime of NO threat at all to the UK. Iraq is now run by Iran. Well done Tony. Much misery caused and many people died.

  15. AP

    @Mike Smith – The last YouGov Poll before the 2016 election was pretty much spot on for the result.

Leave a reply

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