New ICM, Populus and TNS-BMRB polls

There should be several polls out tonight. First up is the Guardian's monthly ICM poll, which has topline figures of CON 33%(+2), LAB 41%(nc), LDEM 14(nc), Others 12%(including UKIP 5%, Greens 2%). Changes are from the last ICM poll just before conference season began. The poll shows a slight increase in Conservative support, but this is probably a reversion to the mean after the last ICM poll, which showed an unusually large Labour lead by their standards.

ICM also asked whether people thought Cameron & Osborne or Miliband & Balls were best able to handle the economy, showing a sharp narrowing of the Tory lead. Cameron and Osborne are on 31%(down 9 since July), Miliband and Balls on 27%(down 2). Later this week we have the 3rd quarter GDP figures, so it that shows something interesting enough to be noticed beyond the business pages it'll be interesting to see the impact.

While I am here, kudos to Tom Clark at the Guardian for putting the ICM poll in the context of other companies polls in his write up and explaining the methodological reasons for some of the differences.

Earlier on today we also had the weekly TNS-BMRB poll, which has topline figures of CON 30%(+1), LAB 44%(+2), LDEM 8%(+1), UKIP 7%(-3), Others 11%(nc). Changes are from last week. Fourteen points is a high Labour lead by most companies standards, but is actually pretty normal for TNS-BMRB's recent polls - they tend to show some of the bigger Labour leads and have the Labour lead between 13 and 16 points in their last few polls.

We may also get the monthly Populus poll for the Times tonight (the Guardian's Tom Clark was polled for it!). I shall update if it surfaces.

UPDATE: The actual poll isn't out yet, but the Guardian also mentions a new poll commissioned by Lord Ashcroft in Corby, which is going to show Labour 22 points ahead there. That would be a very strong performance, a thirteen point swing since the general election. Full details of that are due tomorrow morning.

UPDATE2: Having been complementary about the Guardian's coverage of their ICM poll I may now have to be less so about the Times's coverage of their Populus poll. Of course, since the Times is behind a subscription wall the end of their coverage may be hedged with all sorts of caveats, but the beginning of it talking about "slashing Labour's lead" doesn't look good!

Anyway last month's Populus poll was a pretty obvious outlier, showing a startling 15 point Labour lead when Populus had previously been showing Labour leads of around seven points and when no other company was showing similar huge swings to Labour. This month things are back to more usual results for Populus, with topline figures of CON 35%(+5), LAB 40%(-5), LDEM 9%(-1). Big shift from Labour to Conservative and pehaps a sign of a slightly reduced Labour lead, but the size of the shift is essentially meaningless.

At first glance the the polls appear to be giving a contradictory picture, but they really aren't. Part of it is because of the methodological differences (companies like TNS-BMRB and YouGov tend to show bigger Labour leads than companies like Populus and ICM because they treat things like turnout and don't knows differently) and part of it is because of sample error and reversion to the mean after some unusual results. The bigger picture though is that the polls were showing Labour averaging at around about 9-10 points before conference, and are showing Labour leads of around about 9-10 points now.