Lord Ashcroft returns to the Lib Dem battleground

Lord Ashcroft released another batch of constituency polls earlier today, this time revisiting some of the Lib Dem seats where he had previously found close battles. In Lord Ashcroft's previous polling in Lib Dem seats he's often found wide variation from one seat to another, and it's the same here - in some seats the Lib Dems are holding on against the national tide, in other seats they are doing very badly indeed.

Camborne and Redruth was an ultra marginal between the Conservatives and Lib Dems at the last election. In June 2014 Ashcroft polled the seat and found a close three way battle between the Conservatives, Labour and UKIP. The picture in this poll is far less exciting - a very lead 13 point Tory lead.

North Devon and St Austell and Newquay both had narrow one point Tory leads last time, this time they had more comfortable seven and six point leads for the Tories.

North Cornwall, St Ives and Torbay all saw much less movement. Torbay and North Cornwall both had neck-and-neck ties when Ashcroft last polled them, this time he found a one point Lib Dem lead and a two point Lib Dem lead. St Ives has gone from a one point Lib Dem lead to three points, showing almost no swing from LD to Con at all since the general election.

Turning to the two Lib Dem seats where Labour is the main challenger, Julian Huppert in Cambridge has now opened up a nine point lead over Labour, reducing the swing from LD>Lab to just three points, which would be exceptionally good in a LD/Lab seat.

Finally Lord Ashcroft re-polled Sheffield Hallam, Nick Clegg's own seat. Naturally this is the poll that got the most attention, as he continues to show Clegg trailing Labour. Voting intentions were CON 16%, LAB 36%, LDEM 34%, UKIP 7%, GRN 6% - a whopping great swing of 19.5% from LibDem to Labour. The Lib Dems criticised the poll for not including candidate names, saying this would have boosted Clegg. Lord Ashcroft pre-empted the criticism by saying that he already asked the constituency specific question and feared putting candidate names in the question would give too much prominence to that as a factor and would risk showing too much of a candidate effect. Both are perfectly justifiable arguments - the reality is we don't know. Constituency polls have been very rare in the past, so we don't have lots of constituency polls with and without candidate prompting from previous elections that we can compare to results to make a judgement. There is simply no evidence that would allow us to judge whether candidate prompting in constituency polling is less or more accurate.