Stephen Fisher, University of Oxford
This page gives details of
the current forecast from a new polls-based method for forecasting a general election
on 7th May 2015. The method was revised in February 2014 and is described in a working paper.
The original October 2013 working paper is here. There is also occasional commentary at
my Elections
etc. blog. To receive Tweets with updates of both the forecast and
commentary, follow me @StephenDFisher.
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 historical tendencies: governments being more
likely to recover and oppositions fall back; parties moving back towards their
long-run average level of support and/or the level of support at the previous
election; and lastly and by far the least important tendency is for the
Conservatives to over perform and Labour to under perform their vote intention
figures in the polls when it comes to election day. 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.
Date
of forecast: 11.07.2014
Days
till the election: 300
Inputted current average poll shares
Con : 31
Lab : 36
LD : 9
Oth : 24
UKIP:
14 so UKIP share of combined Others vote is 58%
Forecast Election Day Shares with 95%
Prediction Intervals
Con : 34.9 plus or minus 7.7 i.e. between 27 and
43
Lab : 31.9 plus or minus 5.7 i.e. between 26 and
38
LD : 12.8 plus or minus 8.6
i.e. between 4 and 21
Implied
point estimate shares for:
Others combined: 20.4
UKIP : 11.9 (based on share of
combined Others vote.)
Forecast Election Day Seats
Con : 295
Lab : 296
LD : 31
Lab
largest party, but short of a majority by 31
Forecast Election Day Seats with
approximate 95% Prediction Intervals
Assuming
LD share at 12.8 and Other share at 20.4 and allowing
Con and Lab to vary as per intervals above.
Con
between 213 and 389
Lab
between 207 and 372
LD
between 25 and 38
Approximate probabilities of key outcomes
Pr(Con largest party) =
49%
Pr(Lab largest party) =
51%
Pr(Con majority) = 23%
Pr(Lab majority) = 25%
Pr(Hung parliament) = 52%
Pr(Hung parliament with
Con largest party) = 26%
Pr(Hung parliament with
Lab largest party) = 26%
The inputted current polling average is that from UK Polling Report,
chosen because it is sensible, convenient and frequently updated. But this is
not a claim that this is necessarily the best estimate of current opinion and
it would be reasonable to use other methods of averaging recent polls. On
occasion I will adjust the polling average to make it more up to date.
I try to update the forecast every week, and mostly on
Fridays.
