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


The Electoral Review of 2015, Part 2

11 January 2016

For this second part of my review of the electoral year in Wales, I turn to the state of the parties as revealed by the opinion polls. With the establishment late in 2013 of the Welsh Political Barometer, we now have more regular opinion polls conducted in Wales than has typically been the case in the past. But this can only be counted as good news if those polls are vaguely accurate! How informative have the polls in Wales actually proven to be?

Judging the accuracy of opinion polls is often difficult because of the lack of objective bench-marks against which to compare the polls’ findings. However, an election outcome provides one very good benchmark against which we can assess the accuracy of at least the final, immediately pre-election, polls on electoral voting intention.

The only company polling on general election voting intentions in Wales in the period immediately prior to the election was YouGov. That means that we can’t compare their performance with that of other polling agencies. However, we can compare the performance of the polls in Wales with that in the other British nations:

  • In Scotland, the final polls produced by the four main companies working there (YouGov, Survation, Panelbase and Ipsos-Mori) were very close indeed to the final result: they got the broad picture absolutely right, and their mean average error per party was only 0.74%, which is a very creditable performance indeed.
  • The final GB-wide polls (the vast majority of whose respondents would, of course, have been from England) all had the Conservatives too low and Labour too high, by around 3% in both cases. Most of these polls also mostly had the Liberal Democrats a little too high – estimating their support at around 10%, rather than the 8.1% they actually obtained across Britain.
  • In Wales, I think the performance of the final YouGov poll might be described as ‘good though not flawless’. (Although I acknowledge that, as I work with YouGov and ITV-Wales on these polls, I might not be viewed as unbiased!) None of the parties’ vote shares was estimated incorrectly by more than 1.2 percentage points; the mean average error was only 0.82%. That is a high degree of accuracy. The main problem with the poll was the direction of the error, in estimating the Conservatives’ support levels a little too low and Labour slightly too high. When this was combined with the Tories’ very effective campaigning in the key seats, it produced an outcome in terms of seats that was contrary to what the polls had long been leading us to expect.


In response to the election, the major polling companies have all been reviewing their methodologies. YouGov published a detailed report based on their investigations in December 2015, while the British Polling Council enquiry into the performance of the polls is due to report early this year. I’ll try to keep blog readers up to date with the main developments on this front as and when they occur.

YouGov have remained the only company conducting regular published polls in Wales since the general election. What have these polls been suggesting for May’s National Assembly election? The following table shows the figures, for both the Constituency and List votes, for all the polls published since the general election:


Constituency Vote

Poll Lab Con Lib-Dem Plaid UKIP Others
ITV-Wales/WGC/YouGov, June 2015 35 23 5 20 14 3
ITV-Wales/WGC/YouGov, September 2015 39 23 6 18 13 2
ITV-Wales/WGC/YouGov, December 2015 35 23 5 20 15 3


Regional vote

Poll Lab Con Lib-Dem Plaid UKIP Others
ITV-Wales/WGC/YouGov, June 2015 32 22 5 20 14 7
ITV-Wales/WGC/YouGov, September 2015 34 24 5 18 14 6
ITV-Wales/WGC/YouGov, December 2015 34 23 4 18 16 6


These figures clearly show Labour in a clear lead. But they also show the party some way below its vote share in 2011, on both ballots. Unless the party is able to generate a rise in its support levels over the next four months, the likelihood must be that Labour’s Assembly election campaign in 2011 will have to be largely a defensive one – focussing on retaining as many of the gains of 2011 as possible rather than looking to make further ground. Constituencies currently held by Labour in the Assembly but lost to the Conservatives in the general election – Cardiff North, the Vale of Glamorgan, Gower and Vale of Clwyd – may well prove particularly challenging for Labour to hold.

The Conservatives’ poll ratings in Wales remain, as they have done for several years, impressively robust. At the moment they must, I think, be strong favourites to finish the Assembly election as the second party in both votes and seats. Their performance in the 2015 general election also gives them a very strong basis from which to campaign in a number of constituencies that are currently Labour-held.

Plaid Cymru start the electoral year apparently little further forward than in 2011. They will need a significant surge in support to come close to realising their ambitions in the Assembly election. However, they begin the campaign with two potential assets. One is that, as revealed by the last Welsh Political Barometer poll, their supporters appear somewhat more motivated to vote in the Assembly election than those of many of their political opponents. Their second potential asset is their leader. In 2015, Leanne Wood’s profile and popularity grew significantly, though without that bringing her party many electoral gains. Plaid must hope that, in the context of an Assembly election, they can more effectively convert regard for the leader into electoral support for her party.

UKIP could well be one of the big stories of the 2016 National Assembly election. They have, in truth, talked up their chances of winning a seat in both the previous Assembly elections yet under-performed on the day. But in 2016 we are currently on course for UKIP to win not just one seat in the Assembly but several. Unless their support collapses, or they completely fail to get their vote out, we must now expect UKIP to be a significant presence in the fifth Assembly term. Whether the Liberal Democrats will still be there alongside them is another question. Back in 2011, Wales was a relative success story for the party – amidst a simultaneous local elections collapse in England, devastation of the party’s representation in the Scottish Parliament, and the heavy loss of the AV referendum, Wales stood out for the party holding almost all its ground. But recent times have been tougher for the Welsh Lib-Dems. In both 2014 and 2015, the Liberal Democrats’ vote share was lower in Wales than either England or Scotland. If such patterns persist, we may emerge from the 2016 election with the National Assembly still containing four parties, but the Lib-Dems having been replaced by UKIP.

A final thing, however, to bear in mind. The inaugural Welsh Political Barometer poll, in December 2013, put Labour on 41% for the European Parliament election the following May, and apparently on course to win three of Wales’ four European seats. Even a poll done only a month before the election had Labour on 39%, and seemingly very likely to win at least two of the seats. In the end, Labour won only 28.15% of the vote, and one European seat. In my next blog post I’ll discuss that in more detail. The point for now is that opinion poll leads – particularly in low-turnout elections, which those for the National Assembly will, sadly, probably continue to be – can sometimes be less secure than they appear. All that is solid about an opinion poll lead can melt into the electoral air…


  1. Harry Hayfield

    The latest forecast I have is pretty much in line with all the others: Lab 29, Con 13, Plaid 11, UKIP 5, Lib Dem 2 (in other words about the same as now in terms of a majority or lack of it)

Leave a reply

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