Friday, 25 October 2013

A long-range forecast for a 2015 British General Election based on current polls and historical polls and votes


I've just started publishing forecast from a polls based method for forecasting the next general election here. I'm hoping to update the forecast weekly. 

As I say on the website: "The methodology is described in a working paper. The approach is broadly to predict the next election based on current opinion polls and the track record of polls in previous electoral cycles allowing for change in opinion in the run up to the election. The method allows for three main phenomena: historical tendencies for the Conservatives to over perform and Labour to under perform their vote intention figures in the polls when it comes to election day; governments being more likely to recover and oppositions fall back in the run up to an election; and a tendency for parties to move back towards their long-run average level of support. All three suggest a Conservative recovery and a Labour set back from autumn 2013. The statistical regression methodology generates estimates of uncertainty and so prediction intervals (range of likely outcomes) and probabilities for key events are also provided below. The forecast represents a way to think about the implications of current opinion polls for the outcome of the next general election in light of the historical relationship between polls and election results. It is the product of a statistical analysis of the data and not my personal opinion about what will happen."

The first forecast is replicated below. There are two striking features. The first is that the Conservatives are forecast to do much better than they are currently polling, mainly because they are in government but also partly because they are historically at a relatively low point (from which regression to the mean effects suggests they should recover) and as a party they have tended to outperform polls at elections. Labour are predicted to do correspondingly worse and so the forecast for the top two is almost symmetrically opposite from the current polls: 40:32 instead of 33:38.

The second of the two most striking features is that the prediction intervals for shares of the vote are enormous.  For the main parties it is clear that there is more uncertainty in the Conservative vote but even the forecast Labour vote could be out by as much as 6.6 points: a huge political difference.  At first glance these prediction intervals may seem to encompass all foreseeable outcomes and more. For the Liberal Democrats the interval from nothing (a hard boundary that had to be invoked!) to 26% seems ridiculously large. While these may seem hilarious at first sight, remember that not all points within the intervals are equally likely to occur. Also as 95% forecast confidence intervals they reflect the historical variation in the votes for these parties and there should be only a 5% chance of a result outside the interval. So they are bound to be very broad to be credible. Even so, the lower bound for the Conservative forecast, at 28%, tells us that it is very unlikely the Conservatives will do much worse than they currently stand in the polls, which is informative. Similarly, the Labour prediction interval suggests it is extremely unlikely that Ed Miliband will do as well as Tony Blair in 1997 or 2001 but it would not be surprising (given polling in previous elections) if he did worse than Gordon Brown or Michael Foot. 

The forecast election-day seats are as you would expect them given the forecast shares. They are a probabilistic estimates of the kind used in election night forecasting. A classic uniform change prediction would produce slightly different figures but not by much, especially given the large prediction intervals for seats that follow from the large prediction intervals for votes.

The estimated probabilities for key outcomes are perhaps the most helpful feature of the forecast so far from an election. These show that the Tories have an 88% chance of being the largest party but only 57% chance of an overall majority. These are rather one sided at this stage in the cycle given there is plenty of scope for change in party support before the election. Even immediately before the election there is considerable uncertainty given the historic record of the polls as shown in Table 1 of the working paper.

Having some estimate for the probability of a hung parliament (currently 28%) is helpful to understand the operation of the electoral system. A Liberal Democrat recovery to May 2010 levels would increase the chances of a hung parliament to over 50%.

Date of forecast: 25.10.2013
Days till the election: 559

Inputted current average poll shares 
Con : 33
Lab : 38
LD  : 11

Forecast Election Day Shares and 95% Prediction Intervals
Con : 40.2 plus or minus 11.8 i.e. between 28 and 52
Lab : 31.8 plus or minus 6.6 i.e. between 25 and 38
LD  : 11.8 plus or minus 14.5 i.e. between 0 and 26

Forecast Election Day Seats
Con : 337
Lab : 265
LD  : 21
Con majority of 24

Forecast Election Day Seat approx. 95% Prediction Intervals
Assuming LD share at 11.8 and allowing Con and Lab to vary as per interval above.
Con between 219 and 471
Lab between 140 and 376
LD between 14 and 30

Probabilities of key outcomes
Pr(Con majority) = 57%
Pr(Lab majority) = 15%
Pr(Hung parliament) = 28%
Pr(Con largest party) = 88%
Pr(Lab largest party) = 12% 

17 comments:

  1. Steve.

    Surely many historical precedents are not valid because we are in a unique situation - there will be no incumbent government defending its position. All parties will be campaigning on the basis of change.

    Secondly what is going to happen to the 2010 LD>CON switchers. In your model they seem to evaporate.

    Third point the detail from almost all polls shows that very little of current LAB support comes from CON converts.

    ReplyDelete
  2. This model is flawed as it takes no account of UKIP's rise in popularity (even allowing for a possible tail off at the GE). Nor does it allow for the fact that a significant number of unhappy Lib-Dem voters will be registering their dis-satisfaction with the Coalition & voting Labour. As Mike Smithson says, this will be a unique election. Such generalised models based on previous voting patterns are, therefore, unlikely to be correct.

    ReplyDelete
  3. One question - there is less probability for Labour to be largest party (12%) than to get a majority (15%). How come?

    ReplyDelete
    Replies
    1. Because as Lab votes increase, Lib Dem votes go down and there is less chance of a hung parliament and more chance of an outright Lab victory.

      Delete
  4. Historical tendencies? Well you forgot the historical tendency that incumbent parties have only once increased their share of the vote since the 50s.. 40% for the Tories from a 2010 starting point of of 35%? What a crock of shite!

    ReplyDelete
    Replies
    1. With a coalition there is no single incumbant

      Delete
  5. To what degree does the model take into account the point frequently made by Mike Smithson, that the loss of half the Lib Dem vote from 2010 has mostly gone to Labour, and there appears to be little sign that it will either go back to the Lib Dems, and very little sign that it will migrate to the Conservatives?

    ReplyDelete
  6. This is the most ridiculous political projection that I have ever seen. It's the equivalent of the us polling firm Scott Rasmussen. This is important because in 2010 mid-terms, scott was the best polling firm but in 2012 presidential he was the worst. Why the disparity? Simple, he thought the white vote and the registered republican vote would be higher than it actually was. His 2010 model collapsed. Using a 2010 model to form a 2015 uk prediction is doing a scott rasmussen. The number of registered labour voters has changed because the liberal democrats betrayed thousands of voters. We can guess a figure or a % of the number of converts but with each seat we have very little idea.

    So let's look at a random seat in the east midlands that would easily be held by the conservatives if this was correct.

    Loughborough is ranked 52 in labours target seat.
    Tory vote - 41.6% (2010) 37.0% (2005) 35.0% (2001) 38.0% (1997)
    Labour v - 34.5% (2010) 41.0% (2005) 50.0% (2001) 49.0% (1997)
    Lib Dem - 18.0% (2010) 18.0% (2005) 13.0% (2001) 12.0% (1997)

    So comparing the results, you can the tory vote has ranged from 35-41.6%. very consistent. But the labour vote has ranged from 50%-34.5% with the lib dems taking 6% of the vote. Another factor to consider is that in 2005/2010 the lib dem vote didnt change but labour lost nearly 5% of their vote to the bnp. Looking at council elections, the bnp vote has gone back to 2001 levels and labours vote has gone back to where it was between 2001-2005.

    Looking at loughborough as one example you see seats like erewash (target 43), nuneaton (38), high peak (69) nw leicestershire (107) have the same numbers involved. Look at the local council results in nw leicestershire and see the changes in vote from lib dem to labour, bnp to labour and the tory share of the vote has stayed static. NW leicestershire is labour target 107!!!! Yet just looking at the numbers, the factors on the ground, the collapse of the lib dem/bnp vote it is a definite winnable seat.

    Here is scott rasmussen after the 2012 general election, http://www.outsidethebeltway.com/scott-rasmussen-tries-to-explain-why-his-2012-polling-was-so-bad/

    ReplyDelete
  7. As far as I can see this 'research' is fundamentally flawed and is relying on 'ifs, buts and maybes'.It ignores the upsurge in the Ukip vote and from what I can see those people who have deserted the Tories for Ukop are even more adamant they will not be going back.
    You also don't mention the LD to Labour switchers who make up four groups 1) Straight LD to Labour defectors 2) Protest voters 3) Former Labour voters who left Labour for the Lib Dems and have now returned 4) Tactical voters. NONE of these groups will ever vote Tory and they are extremely angry that in 2010 they voted Lib dem to keep the Tories out and got the Tories anyway and the Lib Dems have help form one of the most right wing nastiest governments. You are also not taking into consideration the rising popularity of Ed Miliband. True we have to see if this continues, but it it happening. You also seem to be totally underestimating him and Labour. Their energy policy will prove repeatedly a winner, but they will have others they will unleash just prior to the election.

    If David Cameron could not get a majority in 2010 when he had every single thing going for him, how is he now going to persuade 40 constituencies to vote for him?

    You seem to be ignoring Ashcrofts polls in the Tory Labour marginals too, 20,000 people polled and Labour were found to be ahead in 32 out of the 40 seats the Tories need to win the election - Labour poll 14% ahead.

    And your numbers above look way out too.

    I believe your predictions are way off the mark.

    ReplyDelete
  8. This comment has been removed by the author.

    ReplyDelete
  9. An interesting model from Fisher but it totally ignores the fact that most of the rise in the Labour vote has come from the LDs so there is little room for swingback to the Tories, so the Labour vote is unlikely to fall below 35/36% to the 32% he forecasts. I agree with him that the Tories could eke a majority, but at most it will probably be less than 10 and mainly come from squeezing the UKIP vote!

    ReplyDelete
  10. I haven't looked too closely at the maths but as others have said, intuitively these predictions just feel very wrong. If anything, there's too *little* uncertainty rather than too much - or alternatively, the predictions for the parties are way out in the first place. These results claim that Labour have a less than 5% chance of getting above 38%, their current average poll score, which just can't be right. It is at least concievable, with a more than 1/20 chance, that Labour could improve their polling position between now and 2015.

    The suggestion that there is an 88% chance of the Conservatives being the largest party is just embarrassing. Perhaps you haven't considered the ways in which the UK electoral system tends to help Labour and hurt the Conservatives? If you are right and most other political commentators are wrong, you could make a lot of money by backing this possibility on the betting exchanges. But I think you're going to have egg on your face when 2015 comes and Labour end up as the largest party, or perhaps even a majority.

    ReplyDelete
  11. This piece rightly looks at historical data, which very few pieces have done, and largely reaches a similar conclusion to that which Leo Barasi's research did (referenced in mine here http://thecentreleft.blogspot.com/2013/10/with-labours-lead-narrowing-next.html and here http://labour-uncut.co.uk/2013/01/09/the-polling-facts-about-labours-mid-term-lead/), that current analyses are failing to anticipate the natural swing towards the incumbent towards an election.

    I think this looks about right - my only comment would be that it could in theory overstate that swing if it does not take into account changes in polling methodology after 1992. I have not read the paper in detail, so it may.

    ReplyDelete
  12. An interesting blog (and paper), thank you for sharing it. One comment to add to the others. If the Liberal Democrats do poll at 11.8% (or thereabouts), is a top estimate of 30 not rather low? They might well get 30 or less in the end, but given the importance of local parties and incumbency to the party, there is every chance that they will be able to target effectively and mitigate losses predicted nationally.

    ReplyDelete
  13. I have a lot of faith in regression models. Very accurate when determined correctly. I expect when this report is revisited at election time it won't be far wrong

    ReplyDelete
  14. Hi there,

    Reading your working paper I noticed that you use week of poll data rather than eve of poll data for comparison with actual votes. Was there a reason for this choice and would substituting eve of poll data reduce the observed difference between the polls and the election results? I'm also wondering whether (as one would hope) polls have become more accurate over time, and if that's been accounted for in the expected bias towards Labour in your model? Many thanks in advance.

    Jonnie Marbles

    ReplyDelete
  15. It is an interesting statistical game but, in a post code democracy, it is never certain if the polling figures ever represent the significant constituencies. So not only are polling figures affected by the questions asked, they are affected by which post codes those polled live in or, more importantly, don't live in. This is before tomorrow's headlines are taken into consideration.

    ReplyDelete