Peter Kellner of YouGov has an excellent piece in the Sunday Times looking at polling in marginals Before Clegg (BC) and After the Debate (AD). It overturns another big election assumption: that the Tories will perform better in marginals.
He uses a combined sample of 10,000 from past YouGov polls to examine the 115 Lab-Con marginals that should turn blue with a swing of eight per cent.
In this election, during the BC days, the Tory lead in these seats was 4 per cent. Compared with 2005, that represented a swing to the Tories of 6.5 per cent in these target seats, compared with a swing of 5 per cent nationally.
Now look at the difference Clegg has made.
The AD pattern is different. Our sample of 2,220 in these target seats now puts Labour one point ahead. The swing since 2005 is down to 4 per cent in the Labour marginals — the same as the national swing. Not only is the prospect of big Conservative gains from the Lib Dems slipping away; the bonus swing the Tories had been enjoying in the Labour marginals has also disappeared.
The Lib Dem surge has hurt the Tories with special force in Labour-Conservative marginals. The 10-point gain in Lib Dem support in these seats has been overwhelmingly at the Tories’ expense.
Clegg’s rise is making the trusty BBC two-way swingometer more irrelevant. Its predictions are far less useful. But 3-D swing is brain-numbingly complex and no one quite understands how it will play out. There were convincing arguments suggesting it could help the Tories (by splitting the Labour vote) or Labour (because all the new Tory support melts away).
This evidence indicates that, at this stage, the Tories are suffering the most. Admittedly the polling could easily change. There could be a hovering pencil moment that benefits the Tories or Labour. Lib Dem support may evaporate once people realise they are in a Con-Lab marginal. But, whatever the uncertainties, there can be no doubt that Kellner’s verdict will make most unpleasant reading for Lord Ashcroft and the Tory team.
Jim Pickard
Kiran Stacey
