An ICM poll in the Sunday Mirror has voting intention figures, with changes from their poll taken at the very height of the Northern Rock crisis, of CON 33%(+1), LAB 39%(-1), LDEM 19%(-1). The change in this poll alone is not significant, but it is similar to ICM’s findings from before Northern Rock so it does echo the YouGov poll from yesterday which showed there was pretty much no effect on voting intentions from the brief banking crisis.

Everything I’m hearing is that it’s increasingly likely that there will be an election called during the next fortnight, possibly next week. My own best guess was that, if Brown was going to call an election this year (which I didn’t think he would!) he’d have done it the day after the Conservative conference with polling day on November 1st. All the same, if he does call an election, what is the result likely to be?

First off, what are the parties standings in the polls? Apart from Populus, who have the Conservatives up at 36, the Conservatives have been around 32, 33, 34 in the last few polls. Labour have, having fallen back at the end of August, gone back up to the high thirties. The Liberal Democrats are, with the exception of ICM, in the mid-teens (and I’ll come to them in more detail later). Overall the postion appears to be something like CON 33%, LAB 39%, LDEM 16-17%. On a straight uniform swing that would produce a House of Commons with 201 Conservative MPs, 379 Labour MPs and 42 Lib Dem MPs, a Labour majority of 108.

I sometimes ponder whether to keep a running prediction on the website somewhere, there are two reasons I don’t. Firstly, voting intention polls really are hypothetical, they ask how people would vote in an election tomorrow and there isn’t an election tomorrow. If there was, then the last three weeks would have been full of campaigning and equal treatment of the parties on the news broadcasts. Secondly I don’t think there is any nice formula you can plug numbers into and get a prediction – swing calculators need to be tempered with insight and I don’t think an election call today would produce a Labour majority of 108. Here’s why:

1) The Liberal Democrats won’t do THAT badly. A hallowed piece of psephological wisdom is that Liberal Democrats always pick up support during an election campaign. It isn’t necessarily true and it depends where you draw the line from. They didn’t in 1987 because they had a rubbish campaign, they might have in 2005, but it depends where you draw the line. Compared to the polls immediately prior to Tony Blair announcing the election, they went up by about 2 points, if you compare it to their poll ratings earlier in the Parliament though they fell slightly. If you go back a couple of elections the increase is partly artifical – pollsters like MORI used to start prompting with party name half way through the election campaign when candidate names were known, giving the Liberal Democrats and their predecessors an artifical boost of a couple of points. All that aside, the reason for what Lib Dem boost there has sometimes been is that mid-Parliament people forget about them as a viable option and forget they may actually vote tactically for them in their local contest, but during an election campaign when they are doorstepping people and the BBC is compelled to give them fair coverage they pick up support. In the last Parliament Charlie Kennedy was a high profile leader and the issue of Iraq got them plenty of publicity, they didn’t have a lack of publicity to start with so couldn’t recover from it. This time round they’ve been relegated to the political sidelines for much of the Parliament, so with TV coverage guaranteed they will probably bounce back somewhat.

My suspicion is that YouGov show lower levels of Lib Dem support because their sample more accurately reflects who actually votes Liberal Democrat, that lots of them don’t have any identification with the Lib Dems and are protest votes and tactical votes and hence far more likely to forget about them away from elections. But those same people who are more likely to switch to the Lib Dems are there in YouGov’s samples. My guess is that the gap between ICM and the others will narrow, and Lib Dems will come up to about 19/20% in all the polls…and that the gain in their support will come at the expense of Labour.

2) The double incumbency bonus. All MPs build up some sort of personal vote from people they’ve helped and people who they’ve come across in their work. It probably isn’t that much, a couple of thousand votes at most, but it’s enough to make the difference in marginal seats. It will be most noticable for new MPs who gained their seat at the last election. In theory a new Conservative MP elected at the last election will have built up a personal vote in the past two and a half years, meanwhile, unless the defeated MP from 2005 is contesting the seat again, Labour will have lost the personal vote they enjoyed last time round. The projection above includes Labour gaining 24 seats from the Conservatives. Some of those are purely notional gains of seats they already hold like Portsmouth North, but in many other cases like Kettering, Shipley, Hemel Hempstead, Gravesham, Enfield Southgate they are seats that were taken by the Conservatives at the last election. These will have a double incumbency bonus and the Conservatives will probably hold on to some seats they would theoretically lose on a straight swing.

3) Where exactly is the increase in Labour support? Uniform swing calculations assume just that – that the swing is uniform. If what has actually happened is that the Conservatives have gained in the south and Labour have gained in the north and the midlands, it gives a misleading picture. There are more marginal seats in the south than the north, so if a party is doing disproportionately better in the south, they’ll do better in terms of seats than the uniform swing suggests.

We don’t have much evidence to judge this by – there is only the aggregated ICM data since Brown became leader that suggested the Conservatives were doing much better in the south. That alone isn’t much to go on, but it does tally with evidence from before Brown’s accession that suggested David Cameron had increased Conservative support in the south, but fallen flat in the north. If this pattern does hold true (and even more so if Brown’s appeal is a mirror image to Cameron and he’s done disproportionately well in the north) it will help the Conservatives. If the Conservatives do very badly in the north it may not harm them much in seats, they don’t have many to lose.

What holds true for broad regions of the country, could also hold true with demographic groups. If Gordon Brown’s increased support is based on core Labour voters who had sat on their hands at the end of Blair’s tenure returning home, his increased support may be concentrated in inner-city Labour heartlands where it will be of no use to him.

4) The campaign. This one can cut either way of course, but I’d be remiss not to mention the actual campaign and what difference it could make. The 2005 election campaign had barely any effect on people’s voting intentions: realisticially the parties may as well have stayed at home and just burnt several million pounds. Polls over the last couple of months though have been very volatile, from a Conservative lead Brown opened up a large Labour lead, which in turn faded away when crime rose up the agenda and then shot back up at the sight of economic problems. Politics is in flux – public opinion may not be so unchanging as it was in 2005.

Labour will want to campaign on Gordon Brown being the strong, reliable leader that the country needs when there could be trouble ahead, contrasting him with a weak and shallow David Cameron. At the moment Brown himself and the economy, despite the wobbles, are their strong cards. He’ll be saying that the country needs a strong, experienced hand at the tiller, not some flip-flopping, inexperienced, jejune pr boy – now give him the mandate to do the job.

There was some suggestion that the Conservatives would paint an early election as Brown running to the country in advance of impending economic disaster. If they did it would be suicide, we’ve seen that in economic troubles people trust Brown. The Conservatives need to try and fight the election against Labour, not against Brown. Polls show positive ratings for Brown, but the government’s approval ratings is still strongly negative. While Labour will want the election to be a choice between strong Gordon and weak Dave, the Conservatives should try to completely ignore Brown and make it an election about choosing an alternative route to the tired, worn out Labour government that’s wasted away 10 years with nothing to show. To do that they desparately need to offer an external narrative, explain what they want for Britain, rather than the internal narrative about changing the Conservative party they’ve offered for 2 years. They need to say what they are changing it for.

A good campaign from either of the main parties (or campaign calamities from either) could change the picture, people aren’t yet certain about Brown, views can change.

5) How accurate are the polls? The short answer is that they are very accurate these days – NOP got it bang on in 2005. The fact remains however that the tiny errors that the rest did display were all in Labour’s favour. There can be no criticism of individual pollsters’ performance in 2005, they were all within the margin of error, as close as they can reasonably be expected to be. However, as a collective group the errors were all in favour of Labour. If the polls were perfectly accurate, there should have been some slightly too favourable to Labour and some slightly too favourable to the Conservatives, a random distribution either side of the real result. There wasn’t. Don’t get me wrong, this will not be 1992 and the polls will be very close to the real result. What errors there are will only be in the region of percentage point or two, but the difference between Labour being 4 points ahead or 6 points ahead is a very large one in terms of seats.

All of these are non-quantifiable to some degree and many could work both ways – it’s just as likely that Labour will “win” the campaign as it is the Conservatives will and we can’t judge regional differences in support – it could be that it’s Labour who are doing better in the southern marginals. However, just taking what I think is the likely swing to the Liberal Democrats during the campaign and the slight tendency in the polls to favour Labour means Labour’s lead at an actual election could end up being around 3 points or so, the same as in 2005. The double incumbency effect will help the Conservatives in some of the very close marginals, perhaps worth another 5 seats or so, boundary changes mean Labour are defending only a notional 36 seat majority anyway…

On the present polls an election next month will see Labour returned as the largest party, but beyond that nothing’s certain.


A new YouGov poll in Saturday’s Telegraph has voting intentions (with changes from the poll last weekend) of CON 33%(-1), LAB 39%(nc), LDEM 16%(+1). The previous YouGov poll was largely conducted before Northern Rock, this was conducted between the 19th and 21st of September making it the first full voting intention poll conducted entirely since Northern Rock.

It’s pretty clear that the poll shows no significant change from the Northern Rock affair, the Liberal Democrat conference has also had a minor effect at best, with a rise of a single point. While the leads aren’t hugely different – 8 points as opposed to 6 points – it is actually painting a different picture to ICM – ICM showed the largest Labour lead yet, this poll is still down from the peak of the Brown boost. It’s probably not methodological differences: since Gordon Brown became Prime Minister YouGov have tended to be the pollster who produce the largest Labour leads, certainly they’ve tended to show larger leads than ICM. Perhaps its just because ICM’s poll was conducted at the very height of the crisis and people rallied to Gordon Brown as the strong leader… and now the crisis seems to be over they’ve returned to their previous views.

Specifically on Northern Rock 52% of people thought the government handled it well, with 37% disagreeing. The vast majority of people don’t seem to be worried about the security of their savings, 74% say they are not very worried or not worried at all, and only 3% say they are very worried. The effect of Northern Rock on other underlying scores is negative for Labour, but only marginally so – Brown’s lead as best Prime Minister is down 2 points from the last YouGov poll to ask the question, but that still leaves it at an inpressive 22 points. Labour’s lead on the economy is down 4 points, but that still leaves it at a healthy 8 points.

All things considered, Labour appear to have emerged unscathed from the Northern Rock incident. A six point lead must be on the margin of Brown calling an October election – it doesn’t take much of a Lib Dem campaign recovery at Labour’s expense or a couple of campaign tumbles to move it into majority losing territory. On the other hand, what was previously a very soft Labour lead has now been tested against a banking crisis and survived. Tim Montgomerie and Iain Dale are both reporting that the Conservative party is on red alert for an election announcement as soon as Monday.

This is interesting – an ICM poll in Wednesday’s Guardian has voting intentions of CON 32%(-2), LAB 40% (+1), LDEM 20% (+2). It was conducted between the 13th and 16th, so a lot of it would have been after the Northern Rock difficulties started to dominate the news agenda. It’s early of course, public opinion can take a while to digest events, but as an early sign it suggests the sign of economics wobbles is boosting Labour support rather than damaging it.

It seemed to be echoed by a snap Populus poll taken on Monday afternoon, after a day in which the TV news was full of footage of huge queues outside the stricken bank. The full results aren’t up on the Times’s site yet, but Peter Riddell’s analysis mentions that only 20% of people blamed the government (and I’d warrent those are largely Conservative supporters anyway) and that the proportion of people who think Brown and Darling are the best team in an economic crisis is still up at 56%, down from 61% before the recent troubles (and Cameron/Osborne are also down, so Brown’s lead is actually bigger than before).

The 8 point lead in ICM’s poll really does put the possiblity of a snap election back on the table. All along my prediction has been that there will not be an election this year, that there simply wasn’t the time available for it to be demonstrated to Gordon Brown’s satisfaction that a Labour lead was truly consistent, and wasn’t a temporary bounce and wouldn’t fade away in the heat of an election campaign. Iff this sort of lead is echoed in several more polls in coming days though (and the very fact of this large lead, and the Northern Rock affair means that more polls will probably be commissioned) then perhaps, just perhaps…

UPDATE: The Guardian headlines on leader approval figures and the supposed swing against Cameron. This is actually rather small. Assuming the wording of the questions was the same, when ICM asked leadership approval questions last month for the Sunday Mirror David Cameron’s leadership was approved of by 38% of voters, with 42% disapproving. This poll has 37% approving of his leadership, an insignificant change, and 45% disapproving – a change in his net rating from -4 to -8. Over the same period, Gordon Brown’s approval ratings have dropped from +40 to +32. Alas, Sir Menzies Campbell, whose net rating is now marginally above Cameron’s, was not asked about last time, so we can’t draw out any changes. No doubt that Brown is still riding high and Cameron’s in the doldrums, but it’s always worth looking at the figures rather than the headlines.

ICM also has figures for best party on issues, which seem to show Labour head on everything, even Conservative core issues like crime and immigration (albiet, by very small margins). The most notable figure there though is a huge 25 point Labour lead on the economy. For a poll taken almost at the height of a banking crisis, this certainly supports the hypothesis that people have rallied to Gordon Brown as a safe pair of economic hands.

It’ll be interesting to see what happens in the next poll – this was conducted before the crisis stabilised, so Labour may well get a further boost from being seen to have successfully handled the problem. On the other hand, if people have rallied to Brown at a time of economic crisis, they may clear off again once the problem has passed. On top of all of that, we should remember it is just a single poll, we may all be getting excited about an outlying poll that just happened to co-incide with a mayor news story!

A ComRes poll in the Independent has headline voting intention figures, with changes from their last poll, of CON 34%(-2), LAB 37%(+1), LDEM 15%(nc). Like YouGov yesterday it shows a widening of the Labour lead (or in this case, a re-establishment of it). While the changes in YouGov’s poll alone weren’t enough to be confident that anything had changed, the addition of this poll begins to suggest a pattern of labour widening their lead once again.

Unfortunately the poll was conducted on Monday and Tuesday last week, so is actually older than the weekend’s YouGov poll and considerably prior to Northern Rock’s difficulties. We await the first poll to give us any real idea of what political impact, if any, the economic wobbles of the last couple of days has had.

Note that in the Independent’s coverage of the poll they say that, if repeated at a general election on a uniform swing, these figures would reduce the Lib Dems to 13 seats. This is wrong: the Independent’s figures are based on the Electoral Calculus calculator which uses a modified proportional swing, on a uniform swing these results would produce a House of Commons with 229 for the Conservatives, 353 seats for Labour and 39 seats for the Liberal Democrats.