Two new polls this weekend have caused less of a stir than last week’s ICM poll but are actually rather more interesting.
Scotland on Sunday cover a Scottish Opinion poll on voting intention for the Scottish Parliament elections next year, which has the SNP ahead on 33% to Labour’s 29%, the Lib Dems on 19%, Tories 10%, Greens 5% and SSP 2% – though it did not distinguish between the constituency and list votes, which is important.
The poll clearly shows a surge for the SNP, Lib Dems and Greens on the 2003 results, at the expense of Labour, the Tories and SSP. Most polls proved over-generous to Labour last time, though 42% of people are undecided and opinion was volatile in 2003, so Alex Salmond’s claim to be planning his first days in office is a little premature.
Nonetheless, the possibility of an SNP administration should concentrate minds, not least that of Blair as he ponders departure.
Meanwhile, the Telegraph reported on their latest YouGov tracker poll, which like MORI shows Labour on 31% but unlike them has the Tories and Lib Dems unchanged on 38 and 31 points respectively, with the defecting Labour supporters switching to minor parties instead.
Anthony King suggests that YouGov’s larger sample size makes it less liable to outlying results than MORI.
The Telegraph‘s claim that voters prefer Cameron to Brown as PM 43-36 is, however, not strictly accurate. The question asked was actually a “forced choice” question – would you prefer Labour under Brown, or the Tories under Cameron, with no other option.
Cameron’s lead is consistent with previous polling on this question, but actually on the “best PM” question, Cameron has only a 2 point lead over Blair and in other polls is usually behind Brown. Blair’s approval rating has plummeted, but Cameron has yet to reap the benefit directly on this question.
Perhaps more damaging for Brown is the steady decline in economic confidence, which does seem to back up some of ICM’s findings.
YouGov have also published aggregate data – findings of all their polls over six months, giving a sample large enough to give a statistically reliable breakdown by region, gender, etc.