Sunday, 30 April 2017

Updated GE2017 Forecast - Conservative Majority of 116.

Today saw the release of two further polls with field dates of 28th April. Both had the Conservatives well ahead (again), though there is some evidence of a narrowing in their lead over the past few days - back to similar levels as we were seeing at the beginning of the campaign.

The (7 day moving average) uniform swing model using today's new polls produced the following result:

Conservatives 383, Labour 183, Lib Dems 7, UKIP 0, SNP 55, Plaid 3, Greens 1.

The 'polls plus' model was updated today to include a bespoke swing adjustment parameter for Scotland. This new mechanism is based on the latest available Scottish polls ( Survation and Panelbase: April 21st, and YouGov: April 27th), and the 2016 Scottish Parliament results.

It hopefully gives the 'polls plus' model a better assessment of how swings are behaving in Scotland, which with the added dynamics of the SNP (vote share previously not modeled) and the independence question, we should expect to be quite different than in England and Wales.

It also updates new locations where parties are standing down in favour of incumbents/main challengers - this time UKIP in the South West for the Conservatives.

When applying the 'polls plus' model (with the new Scottish swing parameters), the result did change somewhat, but the total Conservative seat projection remained the same.


Conservatives
383
Labour
184
Lib Dems
12
UKIP
0
SNP
48
Plaid
4
Greens
1

The 'polls plus' model moves the Liberal Democrats up by 5 seats, Plaid and Labour up by 1, and the SNP down by 7. Overall, the Conservatives (if this projection were to be correct) would sit on a Commons majority of 116 seats.

The updated projection tracker graph shows the Conservatives maintaining steady in their seat projection, a long way ahead of the Labour party (who do however move up to their highest point yet in the projections).




Conservatives Gains in Scotland

According to the polls plus model, the Conservatives are indeed set to take Moray with about a 3% margin from SNP Westminster leader Angus Robertson (a result suggested by YouGov's analysis of their recent Scottish poll).

In fact, the model predicts the Conservatives will take 5 seats in Scotland from the SNP, adding to their solitary win in 2015 (Dumfriesshire, Clydesdale and Tweeddale). This would be hugely significant for the Conservatices, cementing their position as the SNP's main challengers in Scotland, increasingly pushing Scottish Labour to the fringes, and taking a 'scalp' to boot in the form of the SNP's leader at Westminster. The table below shows current projected Conservative Party wins in Scotland.


Projected Scottish Conservative Wins

Constituency
Winner in 2015
Margin (%)
Berwickshire, Roxburgh and Selkirk
SNP
19.44
Dumfries and Galloway
SNP
9.72
Dumfriesshire, Clydesdale and Tweeddale
Conservatives
21.57
Moray
SNP
2.83
Perth and North Perthshire
SNP
3.44
West Aberdeenshire and Kincardine
SNP
3.48

The polls plus model increases Conservative wins in Scotland, but projects some losses to the Liberal Democrats in England (hence their overall seat projection does not move between the two).


The New Marginals

A new regular feature of the forecast will be showing the line at which new marginals between Labour and the Conservatives would be formed, according to the forecast. In short, each forecast will show the 6 most marginal battlegrounds between the country's two largest parties, according to the projected overall results. This line currently sits between Gower (Conservatives, 27 vote majority) and City of Chester (Labour, 93 vote majority).

According to today's forecast the margin would move deep into Labour territory: between Bury South and Wirral South, both of which were won by Labour with around about a 10% margin in 2015. 

Crucially, and perhaps indicative of the kind of the night that the Conservatives are projected to have, the model is currently forecasting the Conservatives to win in Tooting - current London Mayor Sadiq Khan's old Commons seat.



Constituency
Predicted Winner
Predicted Margin (%)
Mansfield
Labour
0.78
Dudley North
Labour
0.51
Wirral South
Labour
0.50
Bury South
Conservative
0.07
Tooting
Conservative
0.18
Hyndburn
Conservative
0.23


A word of caution should be made at this point - though the projections are expected to give an accurate overall picture, I would urge caution when reading into constituency-level predictions. There are a huge amount of local factors which can always sway the result here or there which it is impossible to control for in a model such as this.

Think of individual seat predictions as more informative of the kind of nationwide result that the model is projecting, rather than actually trying to make a call in each seat.

In other words: don't put money on the Conservatives taking Tooting.

But if they did, it the Conservatives would be well on their way to a 100+ margin.

Thursday, 27 April 2017

Updated Forecast - Conservative majority of 118

Today's YouGov poll (Con 45, Lab 29, Lib Dems 10, UKIP 7, Greens 3) has moved the 7 day rolling average downward somewhat for the Conservatives, but they still enjoyed a healthy lead.

The updated uniform swing generated the following result: Conservatives 387, Labour 178, Lib Dems 7, UKIP 0, SNP 55, Plaid 4, Greens 1.

Adding the contextual swing conditions in the 'polls plus' model produced the following result:


Conservatives
384
Labour
179
Lib Dems
11
UKIP
0
SNP
53
Plaid
4
Greens
1

This would give the Conservatives a majority of 118 (correctly calculated this time, thank you to those on Twitter who noticed that I got that particular sum wrong!). Down on yesterday's forecast, but still very healthy and would constitute a solid vindication of Theresa May's snap-election gamble.

The graph below shows today's forecast moving back towards the initial result projected two days ago. 



Labour need a very significant shift in the polls in order to prevent the Conservatives from achieving exactly this sort of result. Unfortunately for them, it does not appear to be forthcoming.

Wednesday, 26 April 2017

Forecast Launch - Conservative Majority of 134

The PME Politics GE2017 Forecast is officially launched this evening (a bit later than intended).

Using a 7-day rolling average of pooled polls (those released by British Polling Council members within the last 7 days), the model estimates an empty 'uniform swing' result before then also applying a range of contextual conditions (as reported on the methodology page) onto the swing in each seat in order to produce a final, 'polls plus' projection.

The model has been ran on both data from yesterday and today, so as to give an early sense of its sensitiveness to polling changes.

Today's uniform swing model reports the following result: Conservatives 387, Labour 178, Lib Dems 7, UKIP 0, SNP 55, Plaid 4, and Greens 1.

The polls plus model, adding various degrees of conditions to the swing, produced the following result:


Conservatives
392
Labour
172
Lib Dems
11
UKIP
0
SNP
52
Plaid
4
Greens
1

This constitutes an increase of 62 seats for the Conservatives, and would give them a majority of 134. Labour would be down 60, and the Lib Dems up 3. UKIP would be down 1, the SNP down 4, Plaid up 1, and the Greens static.

Such a result would be a disaster for Labour, and a resounding success for Theresa May. That said, unlike other forecasts this model does not have the Conservatives breaking the 400 seat mark.

The same forecast was ran on data yesterday (thus not including the Ipsos Mori and Panelbase polls from today). The result was: Conservatives 382, Labour 180, Lib Dems 12, UKIP 0, SNP 53, Plaid 4, and Greens 1. The graph below shows the change from yesterday to today's predictions.



The two polls from today (both with the Tories pushing 50% thus shifted the Conservatives up a few seats at the expense of each of the Lib Dems, the SNP, but mainly from Labour. Updates will follow as soon as new polling data becomes available.

Forecast Launching Today

I am currently in the final stages of putting together my GE2017 Forecast Model. The model will predict overall seat numbers (not vote shares).

Information on the methodology is up and available on this page.

The forecast will be updated with each new poll, which will be accompanied by a Blog post and a graph tracking the forecast across the election campaign.

This is my first attempt at forecasting, so I am very excited to see how it goes and welcome all comments and criticism on both the model and presentation.

Tuesday, 18 April 2017

#GE2017 - An Opportune Moment and the Death Knell of Fixed Term Parliaments


And so, it begins. An early General Election has been called by the Prime Minister, and Parliament will not block it. The country will go to the polls in an Election which will be dominated by what kind of Brexit we want, and what kind of post-Brexit Britain we want to be.

Many are shocked, "stunned", but this decision was not surprising. Inner-circles have been talking about the strong possibility (and in recent weeks – probability) for a long while. The signs were fairly obvious – for instance, the Conservatives have been raising money and selecting candidates and appointing staff for the past few months, far earlier than they ever would have done for a 2020 contest.

The only real surprise is that it was not after May 4th, which some of us (those who always believed the election was coming this year) thought was going to be May’s final litmus test on whether or not to call a nationwide vote. Even then, others were thinking May could be calling it for as early as May 4th itself.

Despite the amazement of many commentators and pundits, calling this election was both logical from a tactical perspective and confirms the absolute and unquestionable authority of the Prime Minister in British politics (which many had thought reduced or impeded by the Fixed Term Parliaments Act and other constitutional arrangements – or lack of).

Strategically, it is absolutely no coincidence that the day after the Conservatives smash through the 20-point lead barrier, a General Election is called. If the election was held today, on some estimates the Conservatives would reach 400 seats. Labour are weak, and May is strong. The economy looks good (minus the plummeting pound), and the public still back Brexit.

But she is also smart enough to know that this would have not likely lasted too long. Or at least that it would have been foolish to presume that the state of play as it exists now would be the same in 2020. Politics moves fast, and with recent results in mind, nowadays we need to be expecting the unexpected. Perhaps a change of Labour leader, or more likely a big shift in public opinion towards Brexit and the negotiations could have been on the horizon.

There is clearly a large amount of unease and uncertainty surrounding the public's perspective of the Brexit negotiations, polls suggest that they are setting high expectations, and clearly are not in favour of May's plan to leave the EU without a deal in place. There was always the danger that once people realised that they would have to queue at customs to get into France, or pay extortionate roaming fees to text relatives in Spain, that there would be a quick shift against the government’s position on Brexit. 

Better to call an election now, and use a full 5 years to negotiate and ride over the Brexit storm, than to be facing a General Election within months of formally exiting. It makes sense.

An early General Election to stamp authority, silence dissenters, and gain a clear command of the Commons also appeals directly to Theresa May’s style. She has always made use of a very top-down, authoritarian line of management. Whether it be in the Home Office or in Number 10. Dissent within her own ranks and threat to overturn the final Brexit deal by opposition parties have never sat well with her. The need to negotiate and concede, governing with a slim majority, is just not May’s way of doing things. What she wants (and needs) is a clear, strong, majority in the House of Commons.

Finally, the amount of hype surrounding the Fixed Term Parliament Act and its ability to restrain May’s decision to call an early General Election among some commentators has been shown to be completely overblown. 

As a colleague and friend put it, the Act might have provided some additional “friction”, but any suggestions that it could seriously get in the way of an early General Election were always completely nonsensical. Despite what some were saying, this was always going to be easy.

The British Prime Minister, so long as they are the head of both the British executive (Cabinet) and legislative (the Commons) will always be able to call elections as and when they like. Yes, formally speaking May has not ‘called’ and election, and Parliament officially have to vote on it, but this is the equivalent of nit picking and rubber stamping. It is all moot, and that is all the Fixed Term Parliaments Act has ever been.