Conservative Home has two new Constituency polls conducted by Populus for Michael Ashcorft and looking at the constituencies of Nick Clegg and Chris Huhne. Full tables are on Lord Ashcroft’s website here and here.
Taking Eastleigh first, Chris Huhne’s seat is a Con/LD marginal. In 2005 it was an ultra-marginal with only 568 votes in it, in 2010 Chris Huhne extended his majority to 3864 (7%) – the shares of the vote were LD 47%, CON 39%, LAB 10%. Lord Ashcroft’s poll has currently voting intention in Eastleigh at CON 42%(+3), LAB 21%(+11), LDEM 31%(-16) – suggesting the Lib Dem vote collapsing towards Labour and letting the Conservatives through.
Moving onto Sheffield Hallam, this is currently a pretty safe Lib Dem seat for Nick Clegg, with the Conservatives currently in a distant second place. The topline figures for general voting intention in the Populus poll are LDEM 33%(-20), LAB 31%(+15), CON 28%(+4): an even bigger collapse from the Lib Dems to Labour, but as Labour start off in third place Nick Clegg narrowly holds on.
Before anyone gets too excited though, constituency polling in Lib Dem seats is extremely difficult. As noted in Lord Ashcroft’s article, in the large scale polls YouGov did for PoliticsHome in 2008 and 2009 voting intention in Lib Dem seats shifted massively once you asked people to think of their own specific seats and they took into account tactical consideration and their own particular MP. It’s clear that some people answer voting intention polls based on their national preference, even if locally they may vote tactically or on the record of their local MP, and this problem is most severe in seats with Liberal Democrat MPs.
This means that we can’t tell if the results above are a sign off massive unwind in tactical voting for Lib Dems, or a massive failure of polls to pick up tactical voting for Lib Dems.
In Lord Ashcroft’s poll there’s a nod to this – they ask how people would have voted in their constituency where Chris Huhne/Nick Clegg is MP had they known the Lib Dems would form a coalition with the Conservatives, prompting with all the candidate names. Amongst those saying how they’d have voted, the figures in Eastleigh are CON 35%(-4), LDEM 42%(-5), LAB 18%(+8). In Sheffield Hallam they are LDEM 43%(-10), LAB 27%(+11), CON 20%(-4). It’s still not perfect, people may think they’d still have voted X in May but would have changed their minds since (in response to Labour’s change of leader perhaps), but it does suggest there is still a strong personal vote there that will help Clegg and Huhne.
(For methodology anoraks, looking at the cross breaks there are some odd figures. 14% of people who said they voted Labour in Eastleigh in 2010 say they would have voted for Huhne if they’d known about the coalition. That seems like strange behaviour. My suspicion is that those are people who “supported” Labour in 2010, but actually tactically voted for Huhne. However, Populus weighted the data to the actual shares of the vote in Eastleigh based on people’s 2010 recall. In their shoes I think I’d have changed the wording of the vote recall question to ask people specifically about their own constituency and weighted using that)