ICM poll of Lib Dem target seats

The Guardian has a new poll of Liberal Democrat target seats here. I think this is the first such study of these seats, which have suddenly become key to the election result. ICM polled the first 42 seats on this list, seats where the Liberal Democrats need a swing of up to 6% to win.

The share of the vote in these seats back in 2005 was CON 36%, LAB 23%, LDEM 35%. In ICM's poll today they found support at CON 35%(-1), LAB 18%(-5), LDEM 39%(+4). That equates to a swing of 4.5% from Labour to the Lib Dems, and a swing of 2.5% from the Conservatives to the Lib Dems. This isn't actually very good for them if you compare it to the national polls - the last ICM GB poll was showing the equivalent of a 7.5% swing from Labour to the Lib Dems, and a 3.5% swing from the Conservatives to the Lib Dems - in other words, the Lib Dems are doing worse in their target seats than nationally. On these figures, the Lib Dems would gain 10 seats from the Conservatives, and 11 from Labour.

It gets more interesting though if you look seperately at the Conservative-held LibDem targets, and the Labour-held LibDem targets. As Julian Glover rightly warns in his commentary, only 15 of these seats are Labour held so the sample size isn't huge and some caution is necessary, but it appears to show that the Lib Dem advance in marginals is wholly concentrated in Labour held ones: taken separately, responses in Con-v-LD seats shows no discernable swing to the Liberal Democrats, but a swing of about 8 points in Lab-v-LD seats. That would result in the Lib Dems taking

about 28 or so seats from Labour, but few if any from the Conservatives. If this finding is at all accurate, it will be key to the result.