Thursday 18 December 2014


A couple of rules: We know that overall trends are more important than individual polls. We know that while politics is not an exact science, certain historical patterns tend to repeat. What we cannot yet say with a high degree of certainty is how the most finely balanced electoral contest in decades will play out. Here are some thoughts on how you might at least get a sense of the possibilities.
The ‘poll of polls’
You are usually on safer ground looking at an average of polls, as this reduces sample error (the natural margin of error in even a perfectly representative sample) and may even out the different “house effects” individual pollsters have in their methodologies.
Even the poll of polls should be treated with caution, though. If most polls include similar inaccuracies, systematic error can occur – where they are all biased in a particular direction. The classic example is the 1992 General Election, when most pollsters collectively and incorrectly pointed to a victory for Neil Kinnock’s Labour, but this year’s Scottish independence referendum also showed systematic biases across all pollsters, with the poll of polls average significantly underestimating the ‘No’ vote.
Systematic error is not normally large. The problem this year is that relatively small differences in vote share for the Conservatives and Labour could lead to wildly different outcomes. Even a small degree of systematic error could be the difference between David Cameron or Ed Miliband standing on the doorstep of Downing Street in May.
Reversion to the mean
Reversion to the mean says that if a measure is over- or under-performing its long-run trend, it is likely to revert towards that trend. In the context of an election, this means that parties tend, over the course of an election cycle, to climb or fall towards their vote share at previous elections.
A very simplified example of this process at the last election is shown below. It uses the December 2009 polling average and compares it with the results at GE2005 and GE2010 (excluding Northern Ireland):

GE2005 final result (GB only)
December 2009 poll average
GE2010 final result (GB only)
Lib Dem
What we see here is that each party’s vote share moved back towards its previous result (the Lib Dems overshooting it after a late surge). This time around, though, there are a couple of jokers in the pack.
Junior coalition partners suffer
Junior coalition partners face a torrid time. The nearest parallel can be found in Germany. In 2009, Germany’s Lib Dem sister party (the FDP) won a strong 14.6% of the national vote, and went into coalition with Angela Merkel’s larger Christian Democratic Union (a centre-right party). At the following election, their share fell nearly ten points to just 4.8%, despite the government being relatively popular.
The German experience suggests that the Lib Dems could face a similarly severe collapse and that the above projection of 13.2% for the Lib Dems may still be an optimistic estimate. The FDP polled a static 4-5% throughout the run-in to the 2013 election.
A vacuum for UKIP
While UKIP has not stolen hoards of voters from the Lib Dems, it is the collapse of the Lib Dems more than anything that has created the space for UKIP to thrive. Many commentators have made the point that the rise of UKIP is not about policy but about their being a focal point for anti-Westminster sentiment – a role in which the pre-2010 Lib Dems and other “outsider” parties thrived.
The size of the non-Labour/Conservative vote at general elections has increased steadily at every election in recent memory, from just 22% in 1992 to 33% in 2010 (these figures are for Great Britain only; the dotted line projects this forward):
It is reasonable to assume that this trend will continue, given declining party loyalty among younger generations of voters and demographic changes. (More bluntly, at each election, some of the older two-party tribalists have died, and younger, less loyal voters have reached voting age.)
The two-horse race
So the Conservatives and Labour are competing for about two thirds of the total vote. This is reflected in their current polling shares of around 30-34%. Below are some rough seat forecasts based on different permutations of figures in this range (assuming Lib Dem 8% and UKIP 16%). We have used UK-Elect v.9.1 to run the projection.

Con vote % (GB only)
Lab vote % (GB only)
Con seats
Lab seats
LD seats
Hung parliament (Con largest)
Hung parliament (Con largest)
Hung parliament (Lab largest)
Hung parliament (Lab largest)
Lab majority
Lab majority
Lab majority
The most striking feature of these projections is the imbalance in the system. Some will argue that the Labour figures could be depleted by a strong SNP showing in Scotland (some have even suggested, erroneously in our view, by as many as 40 seats) – that could happen, but none of these losses would directly benefit the Conservatives, so barring an extraordinary and isolated collapse in Scotland, the imbalance persists.
The fundamentals
And yet the suspicion remains that Ed Miliband can never be Prime Minister: that voters balk when confronted with the image of him striding into Downing Street; that he is weird, otherworldly, unsuitable. Anyone who has conducted a focus group on British politics will be familiar with these attitudes.
The question is whether these views have already been priced into the polls. It is tempting for those who dismiss Mr Miliband to think that a further collapse in Labour support is on the horizon. But there have been plenty of ineffective party leaders in the past, and there is little evidence that their presence precipitates a last-minute collapse in support (as opposed to an already depreciated stock value). Michael Foot’s disastrous performance in 1983 had already been foreshadowed by polls in the summer of 1982.
So Mr Cameron’s biggest hope will be that the polls are systematically understating his party’s support by two percentage points (see Shy Tory phenomenon), and overstating Labour’s by the same figure. He will hope that he then climbs from an adjusted current position of 33% to 35%, with Labour dipping from an adjusted position of 31% to 29%. This would give him a slim majority or pole position in another round of coalition negotiations.
Mr Miliband will simply want to avoid losing any ground. The debate around 35% and 40% strategies is over. For Labour it is now a minus–two strategy – avoid slipping any further than two points behind the Conservatives, and hope that the Government continues to make mistakes.
Bet on a weak government
In the meantime, given the volatility of the system, betting on an actual outcome this far out is pretty difficult. With four possible outcomes, nothing stands out as being more likely to happen than not happen, with the exception of a hung parliament. Instead we can look for common threads running through each likely outcome, and see if together these give us something we can confidently predict.
The one that strikes us is the likelihood of a weak government. Not all coalitions are created equal: the current Conservative-Lib Dem coalition has been remarkably stable, enjoying a large working majority of over 360 seats, and the (relative) goodwill of senior figures in each party. But a Lib Dem collapse in 2015 would make viable two-party coalitions much harder to form, and the remaining options (Scottish and Welsh nationalists, Northern Irish parties, and UKIP) would make strange bedfellows for a party with a conventional agenda.
Expect a parliament of management rather than reform. Expect the constant threat of backbench rebellions. Expect there to be a battle in each party between consensus-driven politicians, who seek to make a difficult parliament workable, and adversarial figures seeing an opportunity to fight a second election on more favourable terms. And your most lucrative bet may be on another election well before 2020.
Follow ComRes on Twitter for the latest polls and analysis:
Author:Andy White

Head of Innovation
Be prepared for GE2015 with the new ComRes Election Toolkit

The 2015 Battlebus is an online survey of 1,000 adults living in the 40 most marginal constituencies where Labour and the Conservatives share first and second place between them and battle head-to-head to get their candidate elected. This survey offers unique access to the opinions of those voters who will win or lose the election for the main parties - all at omnibus price levels.
In the run up to the election, and whilst the parties are drafting their manifestos, this research tool is ideal for ensuring that each of the parties know the importance of your policy issues to those who will be decisive in getting them elected. This can be very powerful for lobbying material or for generating media hits.
Future MPs survey

Based on forecasts using our long-term voting intention surveys and careful psephological analysis, we have formulated a method of gauging the most likely composition of the House of Commons after May 2015.
Guaranteeing a sample of 100 of those MPs and PPCs who are most likely to take a seat in Parliament after 2015, ComRes is offering its clients a chance to gain vital insight into the levels of support for policy issues post 2015, enabling organisations to be on the front foot for when the new Parliament sits.
More information here.

No comments:

Post a Comment

Comment is open to all feel free to link to this blog.