Comparing the Recent ICM and YouGov Polls in Wales

The new Welsh Political Barometer poll is the second to be published in just over a fortnight. Outside of the 2017 general election, I can’t ever recall having two Welsh polls published in such short succession. Verily doth our cup runneth over…

The two polls have been conducted by two of the most widely-respected polling agencies in the UK. However, they were conducted by very different polling methods. YouGov, as regular readers of this blog will be now be very familiar with, draw samples from their large panel who are ‘interviewed’ via the internet. ICM are among the other companies that have moved into internet polling in recent years. But for this year’s BBC Wales poll, as with previous polls the company has done for BBC Wales, ICM used the somewhat more traditional method of telephone interviewing.

There have been many discussions of the relative merits, and potential problems, of both internet and telephone polling. I am not going to rehearse those arguments here: I have discussed them at various points on the blog previously. My concern in this blog post is simply to see whether there are any significant differences in the picture of current Welsh politics that the two polls have given us. Both asked about voting intention for both Westminster and the National Assembly: we can therefore directly compare these results.

First of all, the following table shows the numbers that the two companies produced (with in brackets the seat projections that these numbers generated via the standard uniform national swing assumption) for Westminster voting intention:



Westminster Voting Intention (Projected Seats)

Party ICM YouGov
Labour 49 (29) 46 (28)
Conservatives 32 (7) 33 (8)
Plaid Cymru 11 (4) 11 (4)
Lib-Dems 5 (0) 4 (0)
UKIP 2 (0) 4 (0)
Others 3 (0) 3 (0)


Clearly, there are no large differences here. All the parties are measured within three points of each other by YouGov and ICM, and there is only one seat out of forty projected to have a different winner. (Preseli Pembrokeshire). The largest difference is that ICM have Labour three points higher. But both agree that Labour are well ahead, and in a very strong position.

What about devolved voting intentions? The following table shows what the two companies found for the two (constituency and regional list) ballots:


  Constituency Vote Regional List Vote
Party ICM YouGov ICM YouGov
Labour 40 39 36 36
Conservatives 22 28 21 27
Plaid Cymru 24 22 22 21
Lib-Dems 6 4 6 4
UKIP 5 5 8 6
Others 3 2 5 7


Again we find a pleasing consistency between the two companies. On both ballots, all of the parties are estimated to have very similar levels of support – in all cases within two percentage points, which is well within any standard polling ‘margin of error’. That is, with one exception. The two companies differ substantially in their estimates of support for the Conservatives – with YouGov putting the Tories six points higher for both ballots than do ICM.

What is particularly curious about this is that there is no substantial difference between the two agencies in their ratings of Tory support for Westminster. Nor are there great differences in how the two companies estimate support for the other parties in the devolved context. Why, then do we see this one specific difference – across both ballots – for the National Assembly?

Looking at the details of the poll, it is clear that the ‘retention rate’ of the Conservatives – that is, the percentage of those who supporting them for Westminster who also then support them for a devolved election – is higher with YouGov than ICM. It is roughly 80 percent for the former, and more like 60 percent with the latter. Sadly, we don’t have enough instances of ICM polls on voting intentions in Wales to see whether this is a consistent pattern, or just a one-off result. We would need such information before we could attempt, with any confidence, to understand why the two companies differ so much in their estimates of Conservative devolved election support in Wales.

What many people would like to know is which agency is right. That is something that we cannot know for sure – there is no way of knowing what is the ‘true’ figure. It is unlikely that there have been substantial real changes in the standing of the parties in the short period between the two polls. The two sets of polling results are the best efforts, by two of the UK’s best polling companies, to estimate that true figure. But we do know that the recent record of YouGov in Wales is very good: in the last Assembly election, general election and in the Brexit referendum, their final poll was close to the final outcome. So I would be surprised if the YouGov poll was very wrong. But then what has electoral politics offered us in recent years but surprises?

But the different polls do generate somewhat different understandings of the current standing of the parties in Wales. In the YouGov version, the Tories are in a clear second in the devolved context; according to ICM, Plaid Cymru are possibly just edging ahead of the Conservatives at the devolved level. But both have Labour well ahead for both Westminster and the National Assembly, and both also put the Conservatives in a strong second place for Westminster.


  • J.Jones

    I find that difference between ICM and Yougov in Welsh voting intention quite interesting if you take it alongside the difference in answers to the constitutional questions between ICM/BBC St David’s day poll and the September YouGov poll which found that 28% of the Yougov sample polled wanted more powers for the Assembly against 44% of the ICM sample wanting more powers. Both polls were the same for independence (7%,6%).
    The question that Roger should be asking is what effect does face to face interview have (telephone) on the desire of the respondent to appear “More Welsh” against the computer, private more honest (?) response.
    I think that there was a paper that touched on this generated after the 2001 census.

  • John R Walker

    5% and 7% for Others on the Regional List – is this support building for the Abolish the Welsh Assembly Party (ATWAP)? I really hope so… With the Lib-Dems and UKIP tanking, ATWAP now deserve to have their own dedicated support question in these polls. After all, they are now the only party with an USP. We, the mushrooms, deserve to know!

  • J.Jones

    “What many people would like to know is which agency is right. That is something that we cannot know for sure – there is no way of knowing what is the ‘true’ figure”

    That’s why we have fearless and impartial academics to analyse and hypothesise and ideally run experiments along these lines:-

    What we have instead is confirmation bias where the aforementioned YouGov poll on the constitutional question elicited not a peep from Roger but the ICM poll which, coincidentally, showed that the population of Wales wanted….much the same as Roger wanted, gets a handy few pieces of analysis written on it.

  • Al Daniels

    5 and 7 or 2 and 3?

    ATWAP supporters are the same people who were anti Welsh in 97.

    They aren’t winning over any new voters and the recent devolution anniversary poll showed every age group in Wales (up to 65) supporting devolution overwhelmingly.

    Why have them as a polling option when they will start to die off in coming years?

Leave a Reply

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