Uncategorized

Ratio Swing Numbers for the Latest Welsh Barometer Poll

I have promised to keep publishing here Ratio Swing (RS) seat projection figures, as well as those for the more traditional Uniform National Swing (UNS) method, on the various Welsh polls that come our way. I reported both sets of figures on the blog for the BBC/ICM poll last week; but it has been ‘helpfully’ pointed out to me that I omitted to publish the RS projections for the latest Welsh Political Barometer poll. So, here they are.

For Westminster, RS produces the following projected outcome for the Barometer poll:

Labour: 28 seats (gaining Cardiff North and Cardiff Central). This is the same projection as we get with UNS.

Conservatives: 8 seats (losing Cardiff North, but gaining Brecon & Radnor). Again, this is the same projection that we get with UNS.

Liberal Democrats: 0 seats. As before, we see the Lib-Dems doing worse on RS than with UNS. On UNS projections of the Barometer poll, the Lib-Dems retain Ceredigion; on RS projections they lose the seat.

Plaid Cymru: 4 seats (holding their existing seats and gaining Ceredigion). This is a one seat better projection for them from the Barometer poll than we get with UNS.

Ratio swing continues to produce a more negative picture for the Liberal Democrats, particularly with regards to their chances of holding Ceredigion. Which method is the more credible? I’ve discussed this before, without coming to any definite conclusion. UNS certainly has problems, which the Ceredigion seat actually illustrates rather well: provided Plaid Cymru rise no higher than 13% in the national polls then it is literally impossible for UNS to project them to gain Ceredigion, even if the Liberal Democrats fall to 0% in the national polls! (The latter, I should probably add, is not something that I am expecting to see happen). On the other hand, ratio swing projects a party to lose an equal proportion of its votes in every seat, and thus will almost certainly over-state the likely vote loss experienced by a party in heartland seats that it is fighting very hard to retain.

Of course, the biggest limitation with both methods is that they do not adjust for local peculiarities and conditions. The effects of a popular local candidate, or a particularly effective and energetic campaign, are not things that we can adjust for in nation-wide projections of opinion poll figures. The best that we can realistically hope for with either UNS or RS, I think, is for them to provide a baseline or benchmark, against which we can gauge more peculiar local results. For the Liberal Democrats I suspect that what will happen in this year’s general election is that their vote will melt away substantially in most of Wales, but be significantly more resilient in their existing strongholds. The big question for them is whether it will be resilient enough.

What about the RS figures for the National Assembly vote intention figures in the Barometer poll? An RS projection of the constituency vote intention figures leaves Llanelli as the only constituency seat to change hands, going from Labour to Plaid Cymru. All other seats remain unchanged. Once we add in the Regional List vote as well, and work through the projections of the list seats, we get the following outcome:

Constituency List Total
Labour 27 2 29
Cons 6 5 11
Plaid 6 4 10
UKIP 0 7 7
Greens 0 2 2
LDs 1 0 1

Again, this produces only very minor differences from the UNS figures: Labour, the Conservatives and Plaid Cymru are each one seat better off on the RS projection than the UNS one; UKIP, the Greens and the Liberal Democrats are each one seat worse off.

Finally – may I repeat a plea that I have issued here before? Neither the UNS nor the RS figures reported here are ‘predictions’ of the results in the next general election or Assembly election. I am not attempting to say what will happen in an election in three months’ time (still less one in fifteen months’ time). They are simply formula-based projections of the current situation, as best we can judge it from the attempts by the pollsters to measure voting intentions. I’m sorry if this seems pedantic, but I do get just a tad annoyed when I see myself quoted as having ‘predicted’ that party X will win Y seats in some future election. I’m quite happy to be held accountable for any predictions I have actually made; rather less so to be judged against predictions that I have never issued. And you don’t want to have an angry psephologist on your hands – it’s not pretty.

Comments

Leave a Reply

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