There's been a chance to digest now the main findings of the YouGov Welsh poll. There are some interesting tidbits in the questions on attitudes to devolution (overwhelmingly positive, but with some oddly contradictory results on the prospect for a referendum on further powers), and in the Labour leadership election (which can be summarised as who, who and who?)
I'll concentrate, though, on the voting intention findings, and what that might mean for predicting individual seats. There's some worthwhile analysis in a series of posts by Daran Hill on WalesHome, the political blog voted Wales's most popular, and from a semi-outsider. For any Welsh-speakers there’s also the highly recommended Vaughan Roderick.Running the figures through UNS gives Labour 20 seats, the Tories 12, Plaid 5, LD 2 and Independents 1.
A couple of questions, though, largely from my ignorance I suspect. A blogger raises the question of weighting. Maybe it would be worth exploring whether the weightings used are entirely suitable for a Welsh context, for instance the lack of specific weighting for Plaid party ID, the slightly different demographic structure of Wales, and whether the newspaper weighting is right (I'd suspect, for instance, that Plaid supporters would be significantly more likely to read 'other' newspapers). I'm not suggesting for a moment that these issues haven't been thought of, it's just that they might be interesting to explore.
Also, the regional breakdowns. The sample sizes are small, and some of the results are a bit counter-intuitive. Perhaps it would have been more interesting to use the Balsom 3-Wales model (which, crudely put, divides Wales into psephological zones based on predominant self-definition of national identities) as a basis for any regional calculations. It might tell us more about regional variation if the Valleys, for instance, were counted together rather than being divided into 3 regions and consequently lumped in with places like Monmouth.
But I'm not really quibbling. Thanks very much to YouGov for this innovation, and to my employer (that's Aberystwyth University, not University of Aberystwyth, get with the alphabetically-inspired brand identity, Kellner!) for helping to commission it.
Sure it's one poll, from one organisation, but it's an useful tool to use alongside the marginals poll and recent local and euro elections to try and predict individual seats. Once the dust has settled I'll do a couple of pieces on individual marginals, but here are some prelimiary thoughts. I think Labour could be in big trouble in the north-east (while possibly holding a couple of their southern semi-marginals); the Tories' advance will be more geographically concentrated than in 79-83 (which will help them make up for the absence of the factor which delivered unusual numbers of Welsh marginals for them in 83 - the SDP); Plaid's slightly underwhelming headline figure hides some real opportunities in individual seats. For the Lib Dems this poll is pretty horrible, but I find them tough to predict. I'm almost coming off the fence to forecast the end of Lembit, but I'm not convinced yet.
One more thing. In the cross-tabs there are some interesting patterns if you compare Westminster and Assembly voting intentions. The figures for people who are intending to vote Tory for Westminster but Plaid for the Assembly is farily significant. I wouldn't expect it to be nil, but there traditionally isn't much crossover in support. Are we, perhaps, seeing the emergence of the anti-Labour tactical voter, who will switch votes according to the context of the election? My, wouldn't that make things interesting in Gower and Llanelli.