Friday, 26 May 2017

Updated #GE2017 Forecast - Conservative majority of 64 Seats

**Note: Today's forecast contains three important updates. Firstly, an extra mechanism is introduced to try and counter for the traditional overstating of Labour Party and underestimation of Conservative Party support in British polling. Secondly, there are updates to Welsh and Scottish swings to take into account more recent polls. Finally, since we are moving closer to election day, the rolling average function has been tightened from 7 to 5 days.

This week, polling averages continued to move in Labour's direction over the week leading up to the Manchester terror attack, and the YouGov poll released last night had the two parties within 5% of each other for the first time in the race to date.

The story does now appear to be that Labour are set to improve on their 2015 vote share, achieved under former leader Ed Miliband.

The uniform swing model projected the following result from its now 5-day rolling average of current polling: Conservatives (344), Labour (223), Lib Dems (6), UKIP (0), SNP (55), Plaid (3), and Greens (1).

This would constitute an increased Conservative seat total of just 13 seats, and a majority of 38.

The 'polls plus' model, now including the 'Conservatives outperforming' mechanism as well as a host of other contextual factors (see methodology page for full information), produced the following prediction:

This result would give the Conservatives a majority of 64, with an improvement of 26 seats on their 2015 result. Though an increase, this would certainly not be the sort of outcome that Theresa May would have been gambling on when calling the early election back in April.

However, that Labour still find themselves so far behind in terms of seats, despite rallying in the polls and surpassing their 2015 benchmark will be frustrating for the party.

The prediction tracking graph below shows how this updated forecast represents a further narrowing of the projected result - and this is despite the inclusion of a new mechanism which certainly gives the Conservatives a boost compared to former predictions.

The predicted post-2017 marginals have thus swung back towards the 2015 line, with the current projection returning Bishop Auckland and Stoke-on-Trent North back into the Labour Party column. The line now sits between Bristol East and Harrow West, according to the 'polls plus' model.

As campaigns resume and further polling data is released, it will be fascinating to see what sort of impact the tragic events in Manchester on Monday have on the campaign. Will Labour's rise in the polls be halted? Can Theresa May escape the baggage acquired from a poorly received manifesto and associated screeching U-Turns?

Friday, 19 May 2017

Updated #GE2017 Forecast - Conservative Majority of 86

Today's forecast features a number of updates to the 'polls plus' model: namely a full account of where the Greens and UKIP are exiting in seats where they fought in 2015, and where incumbent MPs from 2015 are standing down. For full information on the effects that this information has on the model, see the methodology page of this website.

It has been a week since the last PME Politics forecast, when the projected Conservative majority dipped under 100 for the first time since the forecast was launched. Today, the seat projections move in further to give the Conservatives a majority of 86.

The continuing slide of the predicted Conservative majority is important - if this trend continues, we should soon abandon talk of a Conservative landslide (at least in terms of seats).

Over the past week, more polls than not have given Labour a positive swing compared to their 2015 vote share. In other words, in most of this week's polls voting intention for Jeremy Corbyn's Labour has surpassed Ed Miliband's 2015 result.

As such, the Uniform Swing (UNS) model, using the updated 7 day rolling average of Westminster voting intention polls, looks much better than previous for the Labour Party, and gives the following result: Conservatives 370, Labour 198, Lib Dems 5, UKIP 0, SNP 55, Plaid 3, Greens 1.

For the first time the UNS model has Labour close to the McCluskey 'good night' benchmark of 200 seats, reflecting the recent narrowing of the Conservative lead in the polls (though this narrowing is incredibly relative - even the closest polls still have Labour 15 points behind the Conservatives).

The 'polls plus' model does actually take Labour over the 200 mark for the first time in this forecast. The added contextual factors within the model suggest that just a few more Labour MPs will hang on in the face of the 'blue tide' than in the UNS scenario.

The full result from the 'polls plus' model is as follows:

The Lib Dems move perilously close to single figures in today's forecast, currently predicted to be only 2 seats better off than their disastrous 2015 result.

Despite Labour's improvement in the polls, the Conservatives continue to enjoy voting intention shares of close to 50%, thus maintaining a clear - if decreasing - lead in projected seats.

Finally, the 'polls plus' model predicts that the margin between Conservative and Labour seat wins now sits between Darlington and Coventry South. Interestingly, Copeland - taken by the Conservatives in that famous by-election earlier this year - is also close to the projected fault line between the country's two largest parties.

If Copeland were to fall the other way and once again return Trudy Harrison to Parliament, it would be no mean feat given past results in the seat at General Election time. Copeland's precarious position should serve as a reminder to jut how many traditionally solidly Labour seats still remain in jeopardy - polling rise or no. 

Friday, 12 May 2017

Updated #GE2017 Forecast - Conservative Majority of 96

Today's update of the #GE2017 Forecast came with an update to the redistribution of UKIP votes section of the 'polls plus' model (whereby votes are sent to other parties in seats where UKIP have withdrawn).

This section of the model now incorporates every constituency where UKIP are now standing aside (where they previously stood in 2015) in favour of the Conservatives, and the redistribution of their votes has been recalculated to reflect recent analysis conducted by Chris Hanretty of British Election Study data.

Interestingly, the model suggests that this updated list only moves one seat from the Labour to Conservative column (Chorley), such is the impact of their near 50% cut in the UKIP vote share at the national level. The model it seems was already reflecting Tory-UKIP gains by using the present (seemingly observable) swing of UKIP voters to the Conservatives in current polling

The uniform swing model calculated the following result:

Conservatives 376, Labour 190, Lib Dems 7, UKIP 0, SNP 55, Plaid 3, Greens 1.

This sizeable cut in the number of projected gains for the Conservatives reflects something of a revival recently in the Labour Party vote share across national polls. The 7-day rolling swing estimation (which both models use to predict seats) has Labour now just 0.3 points down from their 2015 result.

The story of this election seems to increasingly be solely about where those UKIP voters are going (over half of them appear to have already ditched the party, according to the polls), as opposed to any Labour to Conservative swing.

The 'polls plus' model moved the result to the following prediction:

The result represents Labour's highest seat projection of the forecast to date, reflective of a gradually declining swing away from the party in the polls since late April. It still projects however a healthy Conservative victory, giving them a majority of just short of 100 seats (96).

The forecast tracking graph shows the extent to which the gap between the parties has narrowed - not much in the grand scheme of the result, but certainly a significant shift in the projections.

The 'polls plus' model now projects the Conservative swing into Labour seats to stop at Southampton, Test. Ealing Central and Acton is projected to remain in Labour hands again for the first time since this feature of the forecast began. Chorley would also remain in Labour hands according to the model were it not for UKIP's withdrawal in the Lancashire seat, which is expected to push the Conservatives over the line there.


The news is certainly positive for Labour from a relative perspective, but they are still far from seriously denting the prospects of a significant Conservative majority in the House of Commons come June 8th.

Monday, 8 May 2017

Updated #GE2017 Forecast - Conservative Majority of 146

Today's forecast is the first update to PME Politics predictions since May 2nd. It includes the ICM poll released this afternoon, which gave the Conservatives a record (for ICM) lead of 22 points.

Much has happened since the last forecast, including last week's Local Election results (a resounding Conservative win) and Dianne Abbott's ill fated LBC interview.

Since then, the Conservatives have since apparently reestablished much of a polling lead which appeared to have been cut somewhat over the week previous by the Labour Party.

The rolling 7 day average of polls applied to the uniform swing model today produced the following result: Conservatives 403, Labour 165, Lib Dems 6, UKIP 0, SNP 54, Plaid 3, and Greens 1.

Applying the contextual and political variables to condition the swing according to various parameters in each seat using the 'Polls Plus' model resulted in a final projection of:

The projection has the Conservatives significantly up on the last forecast produced, with Labour someway down and both the SNP and Plaid also down a seat each. The Lib Dems are fairly stable, but seemed to have made some ground in recent polling which may see their forecasted seats rise, if it is sustained into later this week.

The current model forecast moves the Lab/Con marginal line further down the Labour path, putting the Tory Tanks right outside the outskirts of both Manchester (Worsley and Eccles South) and Sheffield (Penistone and Stocksbridge), having already blitzed through the previously safe Labour territories in the South of Bristol and Luton.

Just how might the Conservatives really deliver a result which sees them push towards 400 seats, only a few years after many questioned their ability to even build a majority again after the Thatcher and Major years?

The general pattern of last week's local election contest was of the Conservatives hoovering up UKIP votes in order to bypass the other parties and fly into 1st spot in over 500 council seats across the country. There is every reason to suspect that this may well happen again in next month's General Election, as former UKIP voters continue to abandon their previous party in support of Theresa May and her "Brexit means Brexit" agenda.

The number of Labour held seats in which the total Conservative plus UKIP vote share is greater than the Labour Party vote share is well over 150 (credit to Steve Fisher over at for that one). If the Conservatives are once again successful in mopping up UKIP support, then the sky really may be the limit for the size of their landslide.

Friday, 5 May 2017

Thoughts from #LE2017 Part 4 - Conservatives Sweeping up the 'Left Behinds'

**Edit: in my dazed state at 9am on the morning after the elections, I incorrectly titled the manufacturing figure as 'manual labour'.

The Conservatives gradually winning over blue collar voters in England is nothing new - working class Conservatives have always existed. However, the rate at which working class voters are now seemingly traveling to the Conservatives it is certainly emerging as one of the stories of this local election contest - and will most likely be very prominent in the upcoming General Election in June.

As my colleague Professor Rob Ford pointed out, education continues to make its mark as a dividing line in contemporary British politics:

But as well as this, other indicators not just of social deprivation but social class are beginning to connect with Conservative Party swing. From the graph below we can see a steady upward relationship between Conservative Party swing (from 2013-2017) at the ward level in England and the percentage of workers in manufacturing (according to Census 2011 data). This is evidence of an ever-growing movement of working-class votes to Theresa May's Conservative Party. 

**Edit: in my dazed state at 9am on the morning after the elections, I incorrectly titled the manufacturing figure as 'manual labour'.

Additionally, deprivation also appears to play a role in increasing Conservative swing, with the % of people in each ward reporting to be in 'good health' (again according to Census 2011 data) relating negatively to Conservative Party vote share - in other words, when perceived health status is lower, so the swing to the Conservatives was higher. 

All of this points to one clear picture - a strong connection between social class and deprivation among voters, and Conservative Party swing. Put simply, the Conservative Party is hoovering up the 'left behind' voters, so many of whom flocked in their droves to UKIP just four years ago. 

However, this relationship doesn't hold north of Hardian's Wall. In fact, the opposite is true the other side of the border. For instance, in Scotland the Conservatives actually attracted greater swings in areas with lower rates of unemployment (Census 2011). 

This suggests two divergent trends in Conservative Party voter swings between England and Scotland; while in the former, the party is increasingly attracting blue collar workers in more deprived areas, it appears that the majority of its support in the latter is coming from middle-class areas. 

In each nation, the Conservatives have clearly and purposefully situated themselves on one side of the dominant debates - independence (from the United Kingdom and from the European Union), and it is very much paying off. 

The Conservatives are making inroads in areas no one ever would have reasonably thought possible, and their increasing ability to attract swings from very much working class, blue collar areas in England, whilst simultaneously appealing to a more middle-class, pro-Union base in Scotland is the key to that.

Thoughts from #LE2017 Part 3 - Lib Dem Protestors Return?

It appears that the Liberal Democrats are having some success in both the East and West of England in taking council seats from UKIP (who's torrid time continues - just the one seat won at the time of writing).

Thus far, they have taken seats from the anti-EU party in each of Cornwall, Norfolk, Buckinghamshire, Somerset, and Worcestershire local authorities.

Voters flipping from the UK's most anti-EU, Brexit supporting party to one of its most pro-EU, Remain supporting groupings and swinging seats will no doubt come as a surprise to many.

What we can suggest in response to this puzzle is that perhaps the Liberal Democrats, in taking seats from UKIP, are demonstrating a recapturing of their old 'protest vote' - an element of their support which was very much lost when they joined forced with the Conservatives in government in 2010.

Perhaps, with UKIP - undoubtedly the primary party of protest voting over the past four to five years - fading and voters looking for alternatives, the Liberal Democrats may be beginning to be rediscovered as a 'none of the above' option?

The evidence will be in future polling and survey data, but there does seem to be something to suggest that, in some circumstances at least, the Liberal Democrats are reattracting some of their old base. Protest, 'none of the above' voters were always an important part of that.

Thoughts from #LE2017 Part 2 - Uniform Swing

Another interesting thing to note is that we seem to be seeing much the same pattern from the South West of England right up through to the North East.

What was true of Somerset and Gloucestershire at the beginning of the counts - Conservatives taking seats from the Liberal Democrats despite both largely increasing their vote share, Labour support decreasing slightly, UKIP tanking - is currently holding as we move into the far North Eastern Council of Northumberland.

In short, there doesn't seem to be much regional variation in the patterns observed up until now. Across the country so far, there has been a swing to the Conservatives and Lib Dems from Labour and UKIP, with the main result being Conservatives taking control of seats (and councils).

Patterns in Wales have been slightly different in magnitude but directionally also quite similar (but with Plaid profiting from modest swings rather than the Lib Dems).

Something of a 'uniform swing' across the country - everywhere you look, it's broadly the same story.

As we move through the day, it will be interesting to see if any regions buck the trends. Scotland obviously will provide the most interest in this regard. What will happen with the SNP vote? Is their support also fading? If so, will it be the Conservatives that once again profit from those drifting voters?

Thursday, 4 May 2017

Thoughts from #LE2017 Part 1 - Lib Dem Fightback

As I work on analysing the 2017 Local Election and Metro-Mayor results with the BBC Team, I will be trying to pull out some interesting stories which might tell us something about the upcoming General Election on June 8th.

Number 1 - Lib Dem fightback being restrained by Conservatives

Early on, I spotted a trend of Liberal Democrat loses in Somerset - old South-West heartlands for the party where, if a Lib Dem Fightback is forthcoming, then it should be detectable there.

Commentary from the Guardian

The Liberal Democrats and the Conservatives did end up trading seats somewhat on the Council, but overall once Somerset had declared, the Liberal Democrats found themselves 6 seats down despite their vote share increasing by nearly 5% (on 2013).

What happened in Somerset is indicative of the struggles that the Lib Dems will face in between now and June 8th in kicking the engine into gear of their much-discussed comeback: they have increased their vote share, but the Conservatives have passed them at 100 miles per hour on the outside by mopping up UKIP votes (who, incidentally are absolutely crashing nationwide) and going even further ahead (and in some cases taking Lib Dem seats even when they increased their own vote share).

A similar thing happened in Gloucestershire County Council - the Liberal Democrats increased their vote share by 6.5%, but made 0 gains in terms of seats (conversely, the Conservatives took an extra 8 seats to gain overall control of the Council).

What can we read from this in terms of the upcoming General Election? To put it simply, the Liberal Democrats might increase their vote share in the South West by say 5%, but the Conservatives will go up 10%.

This does not translate into General Election seat wins. Lib Dems need a swing from the Conservatives to them, not for the Conservatives to go further ahead.

So something of a Lib Dem fightback could be on in terms of vote shares in the South West, but winning seats looks like a whole other ball game which appears to be out of reach so long as the Conservatives continue to hoover up UKIP voters.

This very much puts the brakes on the Lib Dem fightback in the South West.

Tuesday, 2 May 2017

Updated GE2017 Forecast: Conservative Majority of 120 Seats

NOTE: A mistake in the coding of the 'polls plus' model was noticed today - the Liberal Democrat comeback mechanism was not operating properly. This fix moved the party up by 3 seats in today's forecast, and likely would have moved them up by a similar amount in previous ones. 

Today's ICM poll gives the Conservatives another near 20-point lead over their Labour counterparts, on a 10% swing from their 2015 performance.

This poll moved the uniform swing model to the following: Conservatives 386, Labour 181, Lib Dems 6, UKIP 0, Greens 1, SNP 55 and Plaid 3.

The 'polls plus' model, which adds a variety of conditions to the swing in each seat if it meets certain criteria (explained in full on the methodology page of this website), gave this prediction:

This result would give Theresa May's Conservatives a majority of 120 seats, well over the 100 mark that many believe she is headed for. 

Events today surrounding Dianne Abbott's interview woes will not likely help Labour to close this gap, it will be interesting to see how Labour's polling does after the fact - will there be a backlash against (further) perceived incompetence? 

Elsewhere, the Lib Dem fightback is arguably one of the non-stories of the campaign so far, at least on national polling evidence. Reading the breakdowns (which we must all do with extreme caution), there does appear to be at least some weight in the argument that the 'fight back' will be a localised affair, and not necessarily detectable in terms of uniform swings or shares. 

The 'polls plus' model however does attempt to account for this (re: Lib Dem comeback mechanism above). But it's probably fair to say that a rise of 4 seats on 2015's performance would not be viewed with great satisfaction by the Lib Dem faithful.

The prediction tracking graph shows a steady and sizeable lead for the Conservatives throughout the period analysed. 

In terms of the (new) Con-Lab margins, the model suggests that the battlegrounds have now moved further into Labour territory, with the line now between Mansfield and Dudley North. Interestingly, the prediction suggests that even the 'progressive alliance' being formed in Ealing Central and Acton (controlled for in the poll plus model) will not be enough to stem the blue tide. 

Predicted Winner
Predicted Margin
Stoke-on-Trent North
Batley and Spen
Dudley North
Wirral South
Ealing Central and Acton
Bury South

If seats such as Mansfield, a Labour seat since Frank Bradley Varley took it from the Liberals in 1923 and never once represented by a Conservative MP, are at an alarming risk of falling to the Conservatives, there really are few Labour MPs who should be sitting comfortably.