Angus Reid marginals data
Political Betting has the latest results from a new AngusReid poll. Their topline figures are CON 38%(-2), LAB 26%(nc), LDEM 19%(+1). As ever, Angus Reid show a lower level of Labour support than other companies (for my thoughts on why see here), but the narrowing gap between the Conservatives and Labour echoes that elsewhere.
Angus Reid have also produced some figures for groups of marginal seats - note that these are properly weighted seperately from the main poll, and not just cross-breaks. It was also a larger poll to ensure they were decent sized samples.
Amongst 150 Labour/Conservative marginals these figures showed a swing of 12% from Labour to the Conservatives, a lot higher than the swing recorded in the country as a whole. As far as I'm aware every poll of marginal seats in the last year or so has shown the Conservatives enjoying a larger swing in the marginals to varying degrees. I'm always slightly wary of the exact difference (and the extent we can draw meaning) but the pattern is at least strongly consistent across studies by all different pollsters.
Amongst Liberal Democrat held seats the shares of the vote, with changes from 2005, are CON 33%(+4), LAB 16%(-3), LDEM 39%(-7). Here I would urge more caution - I have severe reservations over marginal polls carried out in Lib Dem seats.
Essentially, I have rarely if ever seen such a poll that doesn't report a big drop in Lib Dem support, even when subsequent elections don't reflect that. Often they also show increases in Labour support even when the party is down. For example, in November 2004 Populus did some private polling for Michael Ashcroft in Lib Dem held seats being targetted by the Tories - they found the Lib Dem support down by 8 points. The following year the Lib Dem share of the vote in those seats remained static.
My guess is that this is because people do not factor in local candidates or tactical decisions when answering voting intention questions (or perhaps do not consider these things at all until an election is called). Naturally enough, these factors weigh the heaviest in Lib Dem constituencies.
In the PoliticsHome marginal seat polls in 2008 and 2009 I used very heavy prompting, and asked people two versions of the voting intention question to push people towards really considering how they would vote locally... and it massively increased Lib Dem support (and showed them holding their own against the Tories in some parts of the country). I'm not necessarily saying those were more accurate - it's never been tested at an election and for all I know I could have overegged it - but I always take voting intentions in CON -vs- LDEM marginals with a great deal of salt.
