ICM in Scotland and Ashcroft in the marginals
This coming Wednesday is exactly one year until the Scottish independence referendum, so there are a couple of Scottish polls in the Sunday papers. ICM in the Scotsman on Sunday have referendum voting intentions of YES 32%, NO 49%, Don't know 19%. It looks as though there is also a new (presumably Panelbase) Scottish poll in the Sunday Times, though there is no sign of it yet.
Lord Ashcroft also has a new poll out, this time of a group of marginal seats. It's not the battleground poll of lots of different groups of marginals, this one is focused up on the the 40 most marginal Conservative seats - the same group he polled back in 2011, and effectively the seats that would decide between Labour or the Conservatives being the biggest party within a hung Parliament (though it includes 8 Con vs Lib Dem seats, not just Con v Lab marginals).
The key thing to look at here is whether the marginals are behaving like the country as a whole, and what we can tell about the Lib Dem v Conservative battleground, something the national polls don't really tell us much about. Firstly, looking at the Con v Lab marginals, the vote shares are CON 29%, LAB 43%, LDEM 8%, UKIP 11%. The Conservative lead across this group of seats in 2010 was 3 percentage points, so this reflects a swing of 8.5 points, so actually larger than the swing the national polls are currently showing (which is about 6.5 points), good news for Labour.
Now looking at the Con -v- Lib Dem seats. As in 2011, Lord Ashcroft has asked voting intention twice in the poll, first asking a standard voting intention question, then asking people to think specifically about their own seat and asking how they would vote there. This is something that was first used in the big PoliticsHome polls of marginal seats back before the last election - it makes hardly any difference when you ask people in most seats, but makes the world of difference when you ask people living in seats where the Lib Dems are in contention, presumably picking up tactical voting considerations.
On the standard question, voting intention in those Con -v- Lib Dem seats is CON 33%, LAB 24%, LDEM 18%, UKIP 14%. That's the Conservatives down 8, Labour up 11, Lib Dems down 21(!). However, ask the localised version of the question and it shifts to CON 32%, LAB 18%, LDEM 29%, UKIP 12%. Clearly some Labour voters are still willing to vote tactically for the Liberal Democrats. This reflects a swing of only half a point from Lib Dem to Conservative in these seats. If the same happened in seats the Liberal Democrats were defending against the Conservatives, the Lib Dems would probably be extremely happy.
