By Stephen Fisher, John Kenny and Rosalind Shorrocks
Since our update last week there have been several new forecasts, most notably including the YouGov MRP (multilevel regression and post-stratification) model. That was a nowcast rather than a forecast, but the same is true of most of our “forecasts”. More on differences between forecasting models below, along with some observations about intriguing question wording effects for citizen forecasts.
But first, overall, the seats projections overall have tightened for the Conservatives, who are down from a 353 average last week to 346 this week, while Labour are up from 209 to 218. The Liberal Democrat forecast total has dropped yet again (from 23 to 19). Now they are estimated to return fewer MPs than they had going into the election (20), but still more than the number of seats they won in 2017 (12).
|Seats||Betting Markets||Complex models||Simple models||Average|
There is now remarkably little difference between the betting markets, complex and simple models in the expected size of the Conservative majority. Particularly striking is that on average the complex models differ by only a seat for each party from the simple uniform change projections based on the average of the opinion polls.
Variation among complex forecasting models is shown in the table below, which is ordered by the forecast Conservative seat tally. Only one model suggests the Conservatives losing their majority and none suggest a majority of more than a hundred. To that extent they are less variable than the opinion polls over the last week: out of the nine uniform change projections from the most recent poll for each pollster, one (BMG) suggests a hung parliament and one (Opinium) suggests a majority of nearly a hundred (98).
|PM and Pendulum: Lebo and Fisher||311||268|
|Regional change: English & Bailey||349||214||17||49|
|PM and Pendulum: Norpoth||352||228|
|Citizens: Murr, Stegmaier and Lewis-Beck||360||190||16|
|Best for Britain/Focaldata MRP||366||199||17||44|
To the extent that the MRP projections are distinctive, they are on the better side for the Conservatives and particularly on the weaker side for the Liberal Democrats. But the main message from them, and especially from the YouGov model, is that that they reinforce expectations of a Conservative majority, unlike last time when the YouGov model uniquely suggested Theresa May might lose her majority.
(As usual, please let us know if there’s something you think we’ve missed or there are mistakes.)
The main reason the forecast seat totals are less favourable for the Conservatives this week is the narrowing of the Con-Lab lead in the polls from 12 points last week to 10 points this week. This is due to a net transfer of a couple of points from the Liberal Democrats to Labour. If that has been tactically efficient then it could benefit both those parties, but it may not have been.
|% GB Vote||Poll aggregates||Betting markets||Models||Average|
Despite the tightening of the polls and seats forecasts, there has been relatively little change in the predicted probabilities of a Conservative majority. For the betting markets, this may be because the election is now just a week away.
|Probabilities||Betting markets||Models||Polls||Citizen forecast||Volunteered||Average|
|Boris Johnson next PM||0.78||0.71||0.74|
|Jeremy Corbyn next PM||0.24||0.18||0.21|
One noticeable change, however, is an increase in the citizen forecast of a Conservative majority from 32% to 43%. This is only partly because citizens in general are becoming less sceptical of a Conservative majority. It is mainly because of a new citizen forecast from ICM. Their survey question about voter expectations appears to be crucially different from the Deltapoll and Opinium questions. ICM gives respondents two separate options with a Conservative majority: one larger than 50 seats, and one fewer than 50 seats. The combined percentage of respondents who say that either of these is the most likely outcome is 59% (excluding don’t knows). Opinium and Deltapoll, by contrast, provide a single Conservative majority option (of any size). In response to their questions 36% (Opinium) and 34% (Deltapoll) say that a Tory majority is the most likely outcome. ICM’s question also does not distinguish between a hung parliament with the Conservatives as the largest party, and one with Labour as the largest party. Deltapoll and Opinium give these as separate options. The differences in the citizen forecasts between these pollsters are especially interesting given that ICM have tended to give smaller Conservative leads on vote intention than either Opinium or Deltapoll. This suggests that citizen forecasts might be quite sensitive to question wording effects. As a consequence our average citizen forecast will be sensitive to which pollsters have fielded which questions.
The idea of combining forecasts from different sources has a good track record, though it has to be admitted that our attempts for the 2017 general election and the 2016 Brexit referendum did not work out well. See here for a review of our 2017 forecasts. Also worth noting is the experience of the pollyvote.com combined forecast of the US presidential elections.
The basic approach is to combine forecasts by averaging them within each category and then average across categories. Since the different sources do not all present similar figures that can be averaged on a like for like basis there are various judgement calls we have had to make on how to treat the data.
The main changes in method this week are: New models this week are YouGov MRP, Best for Britain/Focaldata MRP, 326 Politics and the Citizen’s forecast model from Andreas Murr and colleagues. Lord Ashcroft’s citizen forecasts have been removed as they are now older than 2 weeks and a new citizen forecast from ICM has been added.
For vote shares, we use the various available polling averages. There are various different polling averages. Some admittedly are quite sophisticated, allowing for house effects, but they are nonetheless estimates of current public opinion and not future votes. We only use polling averages that have been updated in the last 7 days.
To estimate a pseudo-probability of a Conservative majority from the polls, we use uniform change projections for each of the most recent polls from each pollster in the last week. For these we also take estimates for the SNP and Plaid Cymru shares from the most recent polls in Scotland and Wales. The proportion of these projections with a Conservative majority is taken as a pseudo-probability, and similarly we consider the proportion pointing to a majority of over 100 as a pseudo-probability of a landslide.
The pollsters vary considerably in the lead they attribute to the Conservatives. Since nearly all the pollsters are doing polls at least once a week and since we take just the most recent poll from each pollster, our calculated probabilities of a Conservative majority from the polls should not be, and do not seem to be, much affected by any changes in the composition of our poll lead average.
We have divided statistical models into simple (poll average plus uniform swing seats projection) and complex (anything more elaborate but not necessarily particularly complex). Within these categories we simply average the available estimates of seats and shares. We have not excluded any models based on quality, but they do have to be statistical models as opposed to personal guesses. We only include models that have been updated in the last 7 days.
For the seats forecasts we are using just the mid points of the spread betting for the markets. Note that the markets might imply fewer or more seats forecast for the main parties than there are in Britain. This is because the markets are separate for each party and do not need to be consistent collectively.
Betting markets for shares of the vote are typically done in bands. We take the mid points of the (up to) three bands with the highest implied probabilities weighted by their implied probabilities from the betting odds, averaging across bookies.
Some polls ask people what they think that the outcome will be on December 12th. We call these ‘citizen forecasts.’ Different pollsters use different survey questions but they can be combined to generate pseudo-probabilities. We use the proportion of poll respondents who think there will be a Conservative majority, excluding don’t knows and re-percentaging, as the pseudo-probability of a Conservative majority. We similarly use the proportion of poll respondents who think that Boris Johnson will become the next Prime Minister, and similarly generate a pseudo-probability for this outcome excluding the don’t knows. We take the most recent such poll from each pollster in the previous two weeks, although it should be noted that only a few pollsters ask such questions.
These come from the Good Judgement Project, which encourages people to forecast the outcomes of various events in order to develop and improve their forecasting skills. The people providing their predictions to the Good Judgement Project are not necessarily experts, but nor can they be seen as a representative sample, and so we report their predictions separately as ‘volunteered’ forecasts.
Estimates come from morning on 4th December 2019.
Bailey and English
Authors’ own uniform change projections
Polling Averages (less than a week old):
Bailey and English
Authors’ own calculations