Shouldn’t today’s polls make Conservatives criticise Lord Ashcroft?

I’m puzzled by some of the coverage in today’s News of the World / ICM poll of voters in Labour – Conservative marginal constituencies.

The Conservative share of the vote in that poll is 40%. At the last general election their share of the vote in those seats was 38% (assuming that the same seats have been polled as were polled in their marginals poll last autumn; that’s certainly how the newspaper’s report reads). That’s a gain of only 2%.

However, the last national ICM poll – in today’s Sunday Telegraph – has the Conservatives on 39%. That’s a gain of 6% on the general election.

In other words – the poll shows the Conservatives doing worse in the key marginal seats than they are doing across the country as a whole. The obvious conclusion? Lord Ashcroft’s much-vaunted key seats operation isn’t nearly as good as some Conservatives think (as indeed its results in 2005 demonstrated).

Curiously Nick Sparrow of ICM comments that, “Today’s results suggest the marginals situation is worse than is indicated by the national polls.” It’s not clear to me how this really stacks up. For example, if you look at swing rather than just Conservative share of the vote, the poll gives a 5% swing to the Conservatives in the marginals but there is a 7% swing in today’s national ICM poll. I’ve emailed Nick for his comments on this, but the comments thread is open to all…

UPDATE: Nick Sparrow has been in touch. It looks as if the News of the World was wrong when it said, “The last time we ran our rule over the 145 marginal seats six months ago…” (my emphasis) as actually the seats polled weren’t the same as were polled six months ago:

No they are not the same seats. When we are asked to do this type of poll we first look at the sort of swing suggested by the national polls. In this case we estimated the Conservatives on about 40% (+7%) and Labour on about 30% (-6%) to get an estimated national swing of 6.5%. So we chose all seats where Labour came first and the Tories second where a 10% swing would lead to a change in party MP, so as to include all seats that could possibly fall even with a larger than national swing of 6.5%. It is in these seats we found a 9% swing.

Vote share we got this time is as follows 40% Con, Lab = 32%, Libs = 16% and others = 12%.