.comment-link {margin-left:.6em;}

Saturday, October 31, 2009

On that poll

Having been away for a week there is a lot to catch up with, not least the YouGov poll that was rather inconveniently published early on in my holiday. Much of the reporting of this poll has been verging on the hysterical in my view, but then as there are so few polls published for Wales who can blame the media and the on-line commentariat for acting like they have just discovered an oasis in the desert.

I am traditionally very sceptical of polls anyway. They are a snapshot of a moment in time and often proved wrong. More to the point past experience shows that to get any level of accuracy in Wales you need a sample bigger than the standard 1000 plus voters. As evidence of this I point to the 1985 Brecon and Radnorshire by-election in which nearly every constituency poll was way out.

The point is that despite being a relatively small place, Wales is diverse and difficult to read as a whole. Even more foolish in my view is to try and extrapolate Welsh constituency results from the figures. The point is that in all forty Welsh seats there is all to play for and the fact that the Welsh Liberal Democrats got 12% as opposed to the 18% they had in the actual votes in 2005 does not mean that Ceredigion and Brecon and Radnorshire are lost.

They are the last two seats I would seek to predict on the basis of a national poll especially when the European Elections saw a swing to the Lib Dems in both from the previous comparable election i.e. the 2004 European Election and especially when there are very important local factors in play in both seats. Equally I would suggest that uniform Labour to Tory swings in seats like Swansea West and Newport East would not happen. Instead anti-Labour votes are more likely to go to the Welsh Liberal Democrats, underlining the problems of seeking to extrapolate in individual constituencies from a national poll.

I cannot find the poll on-line but I suspect that if there was an ITV NOP poll for Wales in 2004 then it came nowhere near to predicting the outcome of the 2005 election here. I do not expect this poll to be anywhere close for 2010 either. What I do think is that those predicting the demise of the Welsh Liberal Democrats as a result of these figures may well have to eat their words.

In particular I would note that in the absence of a Wales-wide media the ratings for the leaders very much shows better recognition for those who are ministers and who therefore have a higher profile. I also note that for all the crowing of Plaid Cymru bloggers, they appear to have deliberately overlooked one important finding of the poll. That is, that if the poll is taken on face value the nationalists will actually lose seats at the next Assembly elections and will be lucky to make any advance next year. This is not a good poll for Plaid Cymru either.

So much for the effectiveness of Plaid Ministers and so much too for the bizarre assertion by one blogger than it is a tactical mistake to criticise Plaid Cymru in government because that just draws attention to the fact that they are exercising power. For goodness sake get over yourself. Good government depends on good and constructive opposition and that is something that the Welsh Liberal Democrats in particular are commited to delivering. And we are doing so.
I tend to think you're right about extrapolation. To extrapolate Ceredigion or any of the "Fro Gymraeg" seats is particularly silly as these seats have shown time and again that they have unique dynamics. Carmarthen for instance was the only Labour gain anywhere in the UK in 1970 and the only Labour loss in Oct 74.

I also think that unique circumstances in Montgomery make any extrapolation impossible. I think Lembit will hold on but if ever a result was going to decided by local factors it's that one.

The B&R by election is a poor example of "polls getting things wrong" though. As you yourself say polls are a snapshot not a prediction and the polls in B&R may well have have been an accurate reflection of the situation before Arthur Scargill's stupid and ill-judged intervention in favour of the Labour candidate.

The good news about YouGov is that it is intended to be a regular poll. Whatever you think of this poll as a baseline it does mean that in future we should heve meaningful data on trends.
There is also stuff on other sites showing that the democraphics for the poll are based on UK wide demographics and not strictly Welsh demographics.

Wales has a higher % of CDE types as opposed to the higher number of ABC1 in England. This makes the polls very dodgy indeed as it overstates the Tories by some way.

Then again, it an "Anything you wantgov" poll, so what do you expect.
I saw that but did not use it because it appears that the methodology was sound after all.
It was YouGov themselves who stated that the poll was weighted according to the make up of the whole of the GB population.
The main poll before the last Westminster election was a joint one by the three TV companies:

Poll result:
Lab 39%
PC 31%
Con 15%
LD 15%

Election result
Lab 37% (wrong by - 2)
Plaid 20% (wrong by -11)
Con 19% (wrong by + 4)
LD 11% (wrong by - 3)

Even though the poll overestimated the Lib dem support, nobody predicted that you would win Ceredigion even on 15% of the Welsh vote!
I don't recognise those figures Alwyn. I suspect they are the 2007 Assembly election outcome rather than the 2005 general election. The same might go for the poll. Can you confirm your sources please? Thanks
Sorry opened my mouth before I put my brain in gear. The poll I quoted was for the 2003 Assembly Election, not the 2005 Westminster election.

Sackcloth and ashes, hide in a corner and never ever comment on poll results again time I fear!

There was a dreadfully inaccurate Welsh poll before the 2005 election too, I shall dig it out in order to try and salvage my reputation!
According to YouGov 44% of Lib Dem supporters will vote No in a referendum and just 38% will vote Yes.
What are you going to do to get our own supporters to support party policies?
I think I will start by ignoring nonsense polls like that, ignoring partisan comments like yours when we all have a responsibility to work together and then get on with campaigning for a yes vote, persuading no voters, irrespective of which party they might vote for, of the merits of a yes vote.
Not WHOLLY relevant to this blog, but thought you'd like to see this video, brings back some memories, especially for you, I'm sure as 'you were there' yes?:


Post a Comment

<< Home

This page is powered by Blogger. Isn't yours?