ComRes/Indy/ITV - 35/29/21
ComRes's poll tonight has topline figures of CON 35%(-1), LAB 29%(-2), LDEM 21%(+2). There is no sign of the Conservative manifesto boost here... but neither should there be, ComRes's daily polls are running a day behind YouGov's ones and the fieldwork for ComRes was done on the 12th and 13th, so much would have been before the Conservative manifesto launch and the press reaction to it. ComRes's daily polling is just Mon-Wed, so we won't have another GB poll from them until after the leaders debate (though they are doing some sort of instant reaction poll straight after tomorrow's debate).
There is also a new marginals polls in the Telegraph, carried out by Crosby/Textor in 100 marginal seats. We don't have tables for the poll or details of the methodology and obviously we do not have a national poll to compare it to so as to draw comparisons between marginal and other seats, so we can't tell much from it. Even if we did though, the results are not particularly enlightening - it shows a swing of 6 points in marginal seats, so not vastly different from any of the other marginal seat polls we've seen.
UPDATE: According to Mike Smithson the ComRes daily poll is a "rolly-poll" like Populus did in 2005 - each day they interview a fresh sample of 500 people, with the figures reported each day being a combination of the two most recent samples - this means half of each day's figures are actually the same as the previous days figures. The methodological implications are firstly that the polls are not independent - if you get a funny looking poll and the next days agrees with it, it doesn't really support it, because it's half the same poll (it also means one really dodgy sample will pollute multiple polls, in 2005 Populus had the misfortune of getting one ropey sample over a bank holiday weekend, which lead to them having outlandish results in their poll for several days as it worked its way through). The second problem it raises is that they only have 24 hours to contact each day's sample of 500, leaving hardly any time to re-contract people who do not answer the phone.
UPDATE 2: There is also a Harris poll in the Metro. Topline figures are CON 37%(nc), LAB 29%(+2), LDEM 18%(-4). I can't see any fieldwork dates in the Metro report, so don't know when it was actually done.
