ComRes show collapse in Conservative support

ComRes has a new poll in tomorrow's Independent with, for want of a better description, frankly odd results. The topline figures with changes from ComRes's last poll are CON 30%(-10), LAB 22%(+1), LDEM 18%(nc).

These results are clearly grossly out of line with other companies - why? The reason isn't a change in opinion, this poll was conducted at exactly the same time as MORI's poll. Over at Political Betting Mike Smithson is focusing on ComRes's past vote weighting, which has shifted significantly in Labour's favour this month. Unlike Populus and ICM, whose political weightings are practically static from month to month - thus fulfilling the purpose of ensuring that the political make up of each month's sample is stable - ComRes's seem to change from one month to the next.

I'm struck by the high proportion of "others" in the poll. UKIP are at 7%, the Greens at 8%, the BNP at 3%, SNP and PC at 3%. These are all perfectly believable and in line with other companies. However, unlike anyone else ComRes also have 9% voting for "other others" - far out of line with everybody else.

I'm reminded of Populus's last review of their methodology: people who said they were going to vote for "another party" used to be included in others, even if they didn't know who for. Populus changed their policy so if people didn't name a party they were classed as don't know. If ComRes include people who don't know what "other party" they'll vote for, it's a potential reason behind that big "other other" score.

Alternatively of course, MORI could be wrong (the changes in party support are, after all, just as extreme as ComRes's) and the Conservatives could suddenly have seen their support collapse over the weekend. Personally, however, I'd be amazed if YouGov, Populus or ICM produced figures to support this poll.