The scales have tipped. After favouring the Tories since it launched in October last year, our model now makes Labour the favourite to emerge as the largest party in May.
It’s still a very close call – our forecast gives Labour a 52% chance of winning the most seats to the Tories’ 48%. We did have Labour as slight favourite once before – back in July – but this is certainly their best forecast so far.
So what’s happened? Well, the Tories have dropped a point in the polling average this week. But it’s not as if their 32% share is especially bad compared to previous weeks – in fact, they’ve been on 32% almost half the time since October, and on 33% for most of the other weeks.
Nor have Labour made any gains. They haven’t moved from 36% in the average for two months, and are actually down a point or two from where they started 2014.
But – crucially – yet another week has gone by without the Tories improving. Our model expects the Tories to make up ground and Labour to lose it, because that’s what the historical tendencies suggest. Every week that doesn’t happen, the forecast moves slightly away from the Tories and towards Labour.
And that’s been the overall trend since October: the Tories have made no progress in the polls, partly because of their failure to win back those who’ve defected to UKIP. As the General Election draws nearer, that failure’s led to them losing their position as favourites.
Our central forecast is now for the Tories to win the most votes (with 35.0% to Labour’s 32.4%), but Labour to win the most seats: 299 to the Tories’ 295, though still 27 short of a majority (coincidentally, that’s the number the Lib Dems get). So, in that scenario, a Labour-Lib Dem coalition would command only the slimmest of majorities, and a Tory-Lib Dem one would fall short.
Date of forecast: 12 September 2014
Days till the election: 237
Inputted current average poll shares
Others (inc. UKIP): 24%
– UKIP: 15%
Forecast Election Day Shares (with 95% Prediction Intervals)
Con: 35.0% (±7.2, i.e. 28% – 42%)
Lab: 32.4% (±5.4, i.e. 27% – 38%)
LD: 11.5% (±7.8, i.e. 4% – 19%)
Implied point estimate shares for:
– Others (inc. UKIP): 21.1%
– UKIP: 13.2%
Forecast Election Day Seats (with approximate 95% Prediction Intervals)
Con: 295 (219 – 383)
Lab: 299 (216 – 371)
LD: 27 (22 – 34)
(Prediction intervals assume LD & others shares at central forecast, Con & Lab shares vary as per prediction intervals)
Central forecast: Lab largest party, but short of a majority by 27
Approximate probabilities of key outcomes
Con largest: 48%
… with a majority: 22%
Lab largest: 52%
… with a majority: 26%
Hung Parliament: 52%
… with Con largest: 25%
… with Lab largest: 26%
(probabilities may not sum due to rounding)
2 thoughts on “Forecast update: 12 September 2014”
To play devil’s advocate a little, at what point assuming the trends don’t change would you judge that the historical tendencies on which this model is based simply don’t apply at this election? If the model continues to gradually move towards the polls rather than the other way round, it must eventually falsify its own hypothesis. Is there a way to judge when this happens objectively/mathematically?