Unlike the YouGov and Populus polls in recent days an Ipsos-MORI poll in the Observer doesn’t show any sign of a conference boost for Labour – the topline figures with changes from the last MORI poll, conducted directly prior to the Labour conference, are CON 34%(nc), LAB 41%(-1), LDEM 16%(+2) This is an increase for the Lib Dems, but pretty static for the two main parties. There is no indication yet of the exact dates of the fieldwork.

The poll also underlined what is Gordon Brown’s undoubted strength since the Northern Rock crisis – an overwhelming 60% of people think Gordon Brown is best able to handle a crisis compared with only 13% for David Cameron and 9% for Sir Menzies Campbell. Rather cuttingly for David Cameron, amongst Tory voters 37% trusted Brown most in a crisis compared to 32% for Cameron.

A BPIX poll in the Mail on Sunday also shows the Conservatives on 34% and Labour on 41%, but with the Lib Dems down on 12%. It’s the lowest Lib Dem score we’ve seen for a long while, but that’s to be expected. We don’t know what weightings they do use, but we do know from their polls last year that they tend to produce figures with the Lib Dems considerably lower than in other polls.

UPDATE – the full results of the MORI poll are now up on their website here, with one very interesting finding. A week or two ago we were told that Labour’s private polling showed them 7 or 8 points ahead, but asked how people would vote in an autumn election it shot up to a 14 point lead. We have no way of testing such claims of course, but MORI asked a similar question in this poll. From CON 34%, LAB 41%, LDEM 16% in the normal voting intention poll, asked how people would vote in an autumn election the figures changed to CON 35%, LAB 40%, LDEM 16%. It’s not a huge shift – Labour’s lead is only 2 points lower – but it contradicts the idea that Labour would have an even bigger lead in an immediate election.

When YouGov gave Labour an eleven point lead in the week I think most people were somewhat sceptical, and rightly so – it was a snap poll with less than the normal sample size conducted in a hurry straight after Gordon Brown’s conference speech. A new YouGov poll in Saturday’s Telegraph however confirms the change and is backed up by a Populus poll showing a ten point lead.

The topline voting intention figures from YouGov, with changes from the mid week poll, are CON 32%(-1), LAB 43%(-1), LDEM 15%(+2). Compared to the last full YouGov poll from before the Labour party (probably a more sound comparison, given that the last poll was just a snapshot poll with less than the usual sample size) the changes are CON minus 1, Labour up 4 and the Liberal Democrats minus 1. The poll was conducted between Wednesday and Friday.

Populus’s poll has headline figures of CON 31%(-5), LAB 41%(+4), LDEM 17%(-1). Until now Populus had been showing slightly higher figures for the Conservatives than other pollsters, putting them in the mid thirties while everyone else had them down in the low thirties – this brings them into line with the sort of figures other companies have been showing.

Of course these polls are still likely to be showing a short term boost from Gordon Brown’s speech and the publicity given to it, but we can at least be confident now that it’s a real change, not just the random vagueries of a small poll conducted in a hurry.

A double point lead for Labour must make the pressure for an early poll almost irresistable for Gordon Brown – supposedly he is making the decision this weekend – and no doubt there will be more polls during the weekend for him to ponder over, at the very least a new Ipsos-MORI poll is due on Sunday.

While the polling evidence is strongly in Labour’s favour, the evidence in the other direction is from the local council by-election results mid-week which were very positive for the Conservatives, especially in marginal seats like Corby and Dover. How do we square opinion poll results showing a huge Labour lead with local council by-elections showing swings to the Conservatives?

Some people swear by local by-election results as a way of predicting elections, I have always been rather dismissive of them. They are on a much lower turnout, they are often skewed by particular local factors (a big Labour victory in Worcester a week ago was probably partially due to the revelation that the Tory candidate wrote ran a website devoted to erotic fiction, I’d warrant some of these by-elections had there own local causes too!), people vote differently at local elections to national elections (if nothing else, there is a tendency for the Lib Dems to do better and Labour worse) and to a large extent they are probably decided by local campaigning strength and ability, rather than the bigger national picture. Added to that there aren’t actually very many of them and the picture you get from week to week varies according to where happens to have had a by-election.

If, for the sake of argument, they are a pointer, then why the difference? Perhaps it is a sign that the Conservatives are indeed doing better in marginals, and the huge Labour leads in the polls are because of core Labour voters in the inner cities being enthused. Perhaps it is because the swing to Labour is largely amongst people who probably will not, in the event, vote. I’m sticking with the explanation that they actually just aren’t a very good predictor.

I don’t see any reason to think that the polls aren’t an accurate reflection of current public opinion. If you are a Conservative supporter, then I wouldn’t advise you to take succour from local by-election results. The only comfort I can offer is that these figures are from polls conducted immediately after the publicity boost of Labour conference, if the Conservatives have a good conference then may yet see their own boost in the polls.


Labours press office seem to have made great things of the cross breaks of the last YouGov poll, since it showed a whopping 17 point lead amongst female voters. I’ve seen various articles by commentators* who have been briefed about it showing how women has defected en masse to Labour. Is it true?

The graph below shows the Labour lead in all YouGov’s polls this year – the black line is their overall lead, the blue line the lead amongst men, the pink line their lead amongst women.


As you can see, the lines bounce about a bit – they have only half the sample size of the overall figures, so that’s to be expected – but generally speaking Labour has had a LOWER lead amongst women over the last 9 months. Those last figures stick out like a sore thumb – it could be a sudden surge in Labour’s support amongst women, but I suspect it’s more likely to just be the result of it being a snap poll with half the normal sample size. You should be careful about reading too much into the internal breaks in polls anyway, if you see something really strange and unusual in the internal breaks of a poll, the chances are its wrong.

That aside, the early election speculation rumbles on. The Guardian’s coverage of this poll included the news that “Some party whips were also claiming that an early poll might mean the contest could be fought on old parliamentary boundaries, a way of preserving as many as 15 seats for Labour”. If they are claiming that they’ll be sorely disappointed, the boundary changes went through almost three months ago now.

Plus, over on the UK Polling Report Election Guide, I’ve added an article on what could be some key seats to watch at the next election.

A snap YouGov poll with fieldwork conducted on Monday afternoon through to Tuesday – so directly after Gordon Brown’s conference speech – has headline figures of CON 33% (nc), LAB 44% (+5!), LDEM 13% (-3).

The Conservative figure is believable enough, but the Liberal Democrat and Labour figures are extreme – the highest Labour figure recorded by YouGov (and indeed by any pollster) since 2002 and the lowest YouGov Liberal Democrat score since the depths experienced after Charles Kennedy’s removal. On a uniform swing these figures would give Labour a stonking majority of 140, put the Lib Dems down on 31 seats and the Conservatives on 197.

How seriously should we treat it? Well, it’s a snap poll, taken while Brown’s speech was still rumbling in respondents’ ears, there is very clearly a conference speech boost to Labour’s support, largely at the expense of Liberal Democrat voters. Gordon Brown’s first speech may very well give Labour a lasting boost in the polls, but this isn’t it – my guess is that this is just the transitory boost from the speech, exaggerated by doing the whole poll within 24 hours straight afterwards.

If you look over at the voting intention graph, you’ll see a similar sudden spike in Labour’s support at their last conference that put them equal with the Conservatives from being 7 points behind a week before. It faded away, I expect this one will too. That doesn’t change the fact that it’s really going to pile on the pressure for and speculation around an early election (despite the fact the poll also showed that only 29% of respondents wanted one this autumn) and it’s a horrid backdrop for David Cameron going into his party conference next week. If another poll doesn’t come out telling a different story, he’ll be addressing his second conference as leader with the albatross of an 11 point poll deficit hung around his neck. Ouch.

A new Ipsos-MORI poll in the Sun on Monday has voting intentions of CON 34%(-2), LAB 42%(+1), LDEM 14%(-2). There are no dates available for the fieldwork yet – sometimes MORI’s polls do take slightly longer to reach the papers, especially the monthly face-to-face polls, so this may have been taken prior to the ICM and YouGov polls that were published over the weekend.

The 42% recorded for Labour’s support equals their highest level of support so far this Parliament – a level last reached in a MORI poll back in 2005 before David Cameron became Conservative leader. 14% is the lowest the Liberal Democrats have recorded from MORI this Parliament and, though we won’t know until we find out the fieldwork dates, it might well have been despite being conducted during their conference. As I said in my predictions for what would happen in an early election, I suspect the Liberal Democrats are suffering as they once did through a lack of media profile, and that in the event of an election campaign they would probably do better than this.

The poll also supposedly contains a varient on the voting intention question asking how people would vote if Gordon Brown offered a referendum on the EU Treaty which gives Labour a 17 point lead. When, a few months ago, all the polls used to contain a question asking how people would vote if Gordon Brown was leader I got bored to my back teeth of religiously adding a caveat about them being purely hypothetical questions and that the public aren’t actually very good at predicting how they’ll react to future events. Given how accurate they were at predicting a hypothetical future then, I trust everyone will give the appropriate pinch of salt to this hypothetical voting intention question.

UPDATE: The poll was conducted between the 20th and 22nd of September, so is bang up to date. There was also a question that asked how people would vote if Brown ruled out a referendum on the EU treaty which showed Labour’s lead slipping to only 1 point. I’d dismiss this question along with the one above! More interestingly the poll also suggests a far, far lower lead for the no camp when people were asked how they’d vote in a referendum on the EU treaty, finding that 38% would vote against it and 32% in favour. This is but a single poll, and others have all shown a much bigger lead for the No camp, but it does suggest that such a referendum could potentially be winnable for Brown.