On those Lib Dem constituency polls
Yesterday we had a selection of ICM polls in Liberal Democrat seats, initially released anonymously but eventually confirmed as having been commissioned by Lord Oakeshott. Oakeshott has now resigned, but left saying that Cable knew about the polls. The political row rolls on, but I'm just going to look at the polls.
Lord Oakeshott commissioned 6 constituency polls. The first was in Twickenham, and only asked about current voting intention. The other five were in Sheffield Hallam, Redcar, Wells, Cambridge and Inverness, Nairn, Badenoch & Strathspey and also asked about voting intention with alternate leaders. The tables are here, here and here.
All six showed the Lib Dems losing (to the Conservatives, Labour or SNP respectively). However, I'm always slightly wary of constituency polls in Liberal Democrat held seats - the effect of incumbency and tactical voting is far higher for Lib Dem MPs, and when you ask a generic voting intention I think many people give their national preference, rather than how they would actually vote in their own constituency. In most seats this is only a marginal difference - in Lib Dem held seats it can be substantial, as repeatedly shown in polls of Lib Dem marginal seats using a two-stage national-then-constituency voting intention questions (see here by Lord Ashcroft, and here by YouGov). It's also worth noting that ICM didn't do their usual reallocation of don't knows according to the party they voted for the last election, and given there are a lot of past Lib Dems now saying don't know that makes a difference. ICM specifically provided the data necessary to do the calculation ourselves in their tables, and with normal reallocation the Lib Dems would have been ahead in Twickenham and only one point behind in Sheffield Hallam.
The second part of Oakeshott's polling was to ask how people in those seats would vote with different party leaders, which in these seats suggested they would do better with Vince Cable as leader. This is in contrast to a YouGov poll in the Times today which asked a nationwide sample how people would vote with Vince Cable as leader - in YouGov's control question asking how people would vote with the existing party leaders the Lib Dems were on 8%, if the leaders were Cameron, Miliband and Cable the Lib Dems would still be on 8%. No change.
That aside, how would you vote if X was leader are incredibly crude tools. When a would-be leader is little known to the public they are as good as useless. Even when the would-be leader is relatively well known to the public, like Gordon Brown was pre-2007 or Vince Cable is now, they are of dubious use.
Essentially, an ordinary poll respondent who doesn't closely follow politics might know what the would-be leader looks and sounds like, might have a pre-existing positive or negative view, they might have a vague perception of competence or incompetence. What they don't know is what policies that new leader would announce, what direction they'd take the party, how the media would react to them and so on.
If the Liberal Democrats change leader it probably won't be the personality and image of the new leader that makes the difference... it will be everything else that comes with it. Would it mean them leaving the coalition? Would it mean them repudiating the coalition and their role in it? Would it mean them opposing some of the policies they supported until now? And what would be the impact of that - would it win back supporters they've lost, or risk alienating their pro-coalition supporters too? While the departure of Nick Clegg may be a requirement for a change of direction, the big strategic question for the Lib Dems is really what attitude they take to their record in government - not which politician is voicing it.
