Showing my working… How I arrived at my May 2015 general election predictions

by Stephen Tall on January 16, 2015

I offered up my predictions for the next election in my ConservativeHome column this week. I thought I’d explain here how I arrived at my guesswork that this is what we’ll wake up to on 8th May:

Conservatives: 35% (291 MPs)
Labour: 32% (283 MPs)
Lib Dems: 12% (32 MPs)
Ukip: 11% (3 MPs)
Others: 9% (41 MPs, including 22 SNP MPs)

First, I’ll explain the poll-shares…

The Conservatives are currently averaging 33% in the polls (UK Polling Report). It is highly likely that will increase by 7th May. In the last national election, May 2014’s Euros, most of the polls under-stated the Tory vote (and over-stated Ukip’s) and I think we may see something similar this time. However, I find it hard to imagine the Tories exceeding the 36% they achieved in 2010. So 35% it is.

Labour is currently averaging 34% in the polls. However, there is a well-documented tendency for oppositions to lose votes in the year leading up to an election — the average loss of support is 6%. Labour was polling c.37% in spring 2014. Ordinarily, then, I’d be predicting Labour would fall towards 30%, or perhaps even lower, especially as some of the anti-Tory tactical vote drifts back to the Lib Dems in their battleground seats. However, I think Labour will be protected by some unwinding of the Ukip and Green votes back to Labour as decision-day nears. So I’m reckoning it’ll net out at about 32%.

The Lib Dems are currently averaging 8% in the polls — so how do I justify my estimate they’ll climb up to 12%? Here I’m placing my faith in ICM, historically the most reliable pollster around. Its methodology of re-allocating some voters who say they don’t know this time according to who they say they voted for last time means its polls are part-snapshot, part-forecast. They tend to be kindest to the Lib Dems — over the past year, the party has averaged 12% with ICM. What has happened in previous elections is that pollsters begin to converge as polling day draws near. Lib Dem voters who are least likely to say they are certain to vote for the party make up their minds later; and my party is more likely to benefit from tactical votes in key seats. Of course, no-one knows if what’s held true in previous elections will also hold true this May, but I’m guessing it will. So 12% it is.

Ukip is also hard to call. However, as Lib Dems know to our cost, third party surges often fade by the time voters reach the ballot box. Perhaps it’s the electorate’s caution at not wanting to ‘waste’ their vote. Perhaps the voters enjoy the flirtation but recoil from the consummation. Perhaps it’s the failure of the smaller parties’ organisational capacities. Whatever it is, I find it hard to believe Ukip will sustain ratings in the high-teens, and that some Tory (and Labour) defectors will re-rat when it comes to the crunch. So Ukip will have to settle for 11% (getting on for triple their 2010 performance).

Secondly, what does all this mean for bums on seats in the Commons? Well, despite the exactness of my prediction I’m not at all sure. I haven’t attempted a seat-by-seat analysis. But I think the following…

In the Tory/Labour battle:

If the Tories poll 3% ahead of Labour it’s very unlikely they won’t end up as the largest party, even allowing for the anti-Tory tilt in the current electoral boundaries. However, the Ashcroft polling indicates Labour is still doing well in the key marginals, and I think Labour’s ground-game is probably superior. Overall I’m not expecting a great shift in seats from Tory to Labour; after all, I’m predicting only a small national Tory to Labour swing of 2% compared to 2010.

Lib Dems:

For all the talk of Lib Dem incumbency, I’m not actually predicting there will be very much of it: the almost-halving of my party’s national rating since 2010 is likely to see an almost-halving of the party’s number of MPs. Yes, we’ll hold on to many seats we ‘ought’ to lose according to uniform national swing; but I suspect we’re in for a grim night in three categories of seats: Scotland, where Labour is in second place, and where the incumbent MP is retiring.


Ukip will win a lot of second places, but not even a handful of MPs — the party has been unable to curb its polarising toxicity, which will make it hard to break the 30-35% threshold needed to win, especially in a general election where turn-out is higher than at by-elections. It’s possible Nigel’s ‘People’s Army’ will snatch more seats than I think by sneaking through the middle in three-way contests. But I think their best chances will come in those (few) seats where they’re standing a well-known, well-organised and long-standing local campaigner.


One place I found very hard even to guess at the results is Scotland. It’s not just that I don’t know it’s electoral politics well; it’s also that the SNP’s performance is genuinely impossible to gauge at this stage. The handful of Westminster voting intentions polls we’ve seen indicate Labour to SNP swings in excess of 20%, enough to produce a nationalist landslide. I don’t think it will be quite that dramatic for the reasons set out on the blog here. However, a combination of an insurgent and battle-hungry SNP, voters’ ‘buyers’ remorse’ from the referendum rejection of independence, and the smaller sizes of Scottish constituencies mean that a big upset is possible. I’ve guessed at the SNP securing 22 of the available 59 seats, which would mean they would still trail Labour quite significantly, but, frankly, who knows?

There you have it, then. The thinking behind my predictions. How does it hold up?