Shooting in the dark: Polling and predicting elections in Wales

I’m convinced that there exists a picture, taken the day after the first National Assembly elections in 1999, of a small group of beaming young politicos, each clutching a fistful of notes. It’s possible that my memory is playing tricks, and at some point I’ll check the archive, but by my recollection the story was that some Plaid staffers, buoyed by their phone canvassing data, decided to put a few quid at stratospheric odds on their party to win Rhondda and Llanelli. In many ways, that tale (even if it’s apocryphal) speaks volumes about political betting and prediction in Wales. It’s an insiders’ game; you need to keep your ear to the ground, read the spin, trust your antennae, and sundry other clichés.

What’s missing, of course, is information; the sort of information that comes with polling. While there have been solid studies of questions such as attitudes to devolution, reliable voting-intention polling just doesn’t happen. Wales-only polls appear only sporadically, generally at Assembly election time, sometimes undertaken by NOP for ITV Wales. Back in ’99, the few opinion polls that were carried out had Labour cruising to an overall majority, and their record has scarcely improved since. Neither has their frequency. The BBC aren’t allowed to do polls which include intention to vote questions, ITV Wales hardly do any political coverage any more, and the Trinity Mirror-dominated Welsh press don’t seem to want to help either. We do have Beaufort Research, which can add a political question to its regular ‘caravan’ surveys of Welsh opinion (sometimes at Plaid’s commission), but all this is a far cry from a robust polling methodology. Extrapolating trends from tiny regional sub-samples or council by-election results is even more of a mug’s game.

Things might be about to change. YouGov have established a Welsh panel, and will unveil the first findings on Tuesday morning. The numbers have been crunched by Richard Wyn Jones and Aberystwyth psephologist Roger Scully, and Peter Kellner will be on hand at the launch to discuss methodology. All very interesting. I hope to get the results pretty quickly, and I’ll update this thread if and when. Apparently, they’ll also be unveiling results on attitudes to the Assembly having full legislative powers, and the Labour leadership contest (for which current betting odds are here).

I suppose we shouldn’t get our hopes up too much. It takes time to work on the methodology, and Wales may just be a tough place to poll – especially if you subscribe to the Balsom thesis that there are, in psephological terms, 3 distinct zones in Wales which behave differently. But YouGov may help us to read the tea-leaves with a bit more confidence, especially when combined with the Politics Home marginals poll. There are issues with this too of course. As its authors admit, the marginals poll is a blunt instrument in that it takes the ‘Welsh marginals’ to be one group. It’s a problematic assumption that Llanelli will exhibit similar swings to Brecon & Radnor, or that Ynys Môn will behave like any other constituency in the known universe.

But for what it’s worth, this year’s marginals poll indicated Conservative gains in Aberconwy, Bridgend, Cardiff North, Carmarthen West & South Pembrokeshire, Gower, Newport West, Vale of Clwyd and Vale of Glamorgan. Plaid was projected to gain Ynys Môn, Ceredigion, and the new seat of Arfon, which is notionally Labour but where Plaid have a sitting MP.

Some quite interesting seats weren’t polled – including both the main Lib Dem targets of Swansea West and Newport East. In addition, the European election results (which in Wales were counted on a constituency basis) give a different picture in some areas. In June, Labour also lost out to the Conservatives in Cardiff South & Penarth and in the three north-eastern seats of Alyn & Deeside, Clwyd South and Delyn. There was less good news for the Tories in Aberconwy, where they were beaten by Plaid (albeit on the old boundaries), but they polled the most votes in the Lib Dem seat of Brecon & Radnor, and next door in Montgomery where the Lib Dems were beaten into third by UKIP. Plaid also beat Labour in Llanelli, a seat the marginals poll had narrowly in the red column. And there’s one Labour-held constituency I haven’t mentioned at all where Labour are reported to be phone-canvassing heavily.

So there may be some betting value out there in the under-researched Welsh constituencies, and if readers wish I’ll put up a couple more articles on the prospects in the marginals over the next few weeks.