Nigel Marriott's Blog

An independent statistician using data to understand our world and to predict the future

You are here: Home / Archive / By-Election Forecast #2 – Stoke-on-Trent Central, Staffordshire

By-Election Forecast #2 – Stoke-on-Trent Central, Staffordshire

By

Originally posted on 19th February 2017.  I then added an updated forecast on 22nd February which can be seen at the bottom of this post.

If UKIP want to supplant Labour in the North England, they must win Stoke-on-Trent Central to kickstart this trend and the odds are 2 to 1 that they will win.  That is my forecast after running my by-election model on Stoke-on-Trent Central.  For more details about the methodology, please read my description of how I forecast by-elections in the Brexit era.

My model uses both the breakdown of the 2015 General Election results when Labour held the seat with a 16% majority on a 50% turnout (the 2nd lowest turnout in the whole of the UK in 2015) and the 2016 EU referendum result where I estimate that Leave won with 71% of the vote on a 60% turnout.  This makes this seat the 17th most Leave seat in the UK and it is worth noting that its two neighbours were #22 (Stoke South) & #3 (Stoke North).  In % terms, the increase in number of voters in 2016 was the 22nd highest in the UK and has created a substantial Non-GE segment of 17% of 2016 voters that have the potential to influence the by-election.  Whether these non-voters from 2015 will vote in the by-election is one of the big uncertainties and it will be fascinating to see if they do.

My voter segmentation based on the Ashcroft “exit poll” show Brexiteers account for 59% of all voters in 2016.  Unlike other Labour seats, this seat appears to be one of the few where Labour Leavers outnumber Labour Remainers.  This exacerbates the debate within Labour as to whether to oppose or support Brexit and how to address the cultural divide that has been exposed within the party.  I think the list of Top 30 Remain and Top 30 Leave seats shown below encapsulates Labour’s dilemma as well as anything.  Note that the shaded figures represent my estimates using my Leave Vote Share model and the unshaded figures are actual results.

    

As I have stated before, the following voter dynamics are already apparent from the 3 contested by-elections we have had since the referendum.

Witney & Richmond Park & Sleaford and North Hykeham

  • Pro-EU voters from all parties have defected to the Lib Dems.
  • Lib Dem Brexiteers have defected to the Conservatives.

Richmond Park & Sleaford only

  • Non-GE voters turnout is 1/3 of other party turnout.
  • A small national swing away from Labour of about 2% has occurred.

Witney & Richmond Park only

  • Best of Both voters have defected to Lib Dems, Conservative only in Witney, Labour & Conservative in Richmond.
  • Half of Potential Bregret voters have defected to Lib Dems

The question now is which dynamics will be at play in Stoke Central and are there other dynamics that could come into play?  To answer the second question, there is one dynamic that is of great interest in both Stoke Central and the Copeland by-election which is taking place the same day and that is whether Labour Leave voters will defect to UKIP.  Therefore in my scenarios I have added a 7th dynamic as follows.

Not seen in 3 contested by-elections since Referendum

  • A quarter of Labour Brexiteers defect to UKIP.

There are many other dynamics that could come into play e.g. Green voters defect to Conservatives, but I have decided to only consider the 7 dynamics listed above.  In effect, I am saying that other dynamics not considered cancel each other out i.e. Conservatives defecting to Green would cancel out Greens defecting to Conservatives.

For some of the 7 dynamics I have allowed for a number of variants such as:

  • Either half or all of Best of Both voters defect to Lib Dems and such defectors may come from Conservatives only, Labour only or both Conservative & Labour.
  • The national swing away from Labour varies between 0% and 8% with the swingers splitting equally between Lib Dems and Conservatives.
  • 3 turnout models are used and are based on the 3 contested by elections we have had since the referendum. Richmond Park had the highest turnout whilst Sleaford had low turnout.

The national swing away from Labour is a hard one to get a handle on as from what I see in opinion polls there are considerable regional variations.  By taking the average of YouGov’s 4 polls between 10th January & 13th February, I have calculated the swings from Labour since the 2015 General Election as shown in the table.

In the South (which contains Witney), there has been no change whereas in London (which contains Richmond Park) there has been a considerable swing to the Lib Dems.  In Scotland, the swing is to the Conservatives whilst in the Midlands and North, the swing is split between Conservatives and Lib Dems.

So I am now ready to run numerous scenarios based on various combinations of the 7 dynamics described and summarise the results.  In total, I ran 528 scenarios broken down into 3 broad groups as listed in this table.

The 3 groups of scenarios are:

  • LAST 3 BEs – 72 permutations of scenarios which are based on what was observed in the 3 contested by-elections. These do NOT include the 7th dynamic of Labour Brexiteers switching to UKIP.
  • VARIANTS – 144 permutations which are based on what was observed in the 3 contested by-elections and also include the 7th dynamic of Labour Brexiteers switching to UKIP.
  • OTHER – 312 permutations which were not observed in the 3 contested by-elections.

The table shows the average vote shares across the multiple scenarios and you can see Labour tops the Last 3 BEs, UKIP tops the Variants and both are tied for the Others.  But for me the more interesting figures are the % of scenarios where each party wins.  These show a clear pattern.

  • If Stoke Central follows the pattern of the 3 previous by elections, the Lib Dems have more than 1 in 4 chance of winning the seat due to Best of Both voters defecting to them. Effectively a repeat of Richmond Park dynamics will lead to shock win.
  • Labour’s best chance of winning comes from holding onto their Brexiteers & Best of Both voters and discouraging Non-GE voters from voting.
  • If the 1 in 6 voters from 2016 who did not vote in 2015 turn out to vote in any numbers at all, it is reasonable to assume they will vote for UKIP and UKIP’s chances start to rise.
  • If Labour Brexiteers start to defect to UKIP, then Labour’s chances sink and UKIP will win in almost 2/3 of scenarios.
  • If other dynamics not seen before start to come into play then we will have a 4-way fight with Labour most likely to come out on top.

Given everything we know about Stoke and the large numbers of Non-GE voters available, the fact that this is a Labour Leave seat in the Midlands, I find it hard to believe that we will see a rerun of one of the 3 previous contested by-elections.  Therefore I am saying that the final outcome will come from the Variants group where UKIP’s chances are 2 to 1 in favour of them winning the seat.

Furthermore, I have decided to stick my neck out and choose one scenario as my forecast which is shown in the 3rd row of table and shows a UKIP win by 3%.  This is based on the following assumptions.

  1. Voting dynamics will be closer to the Sleaford by-election since this is not a London seat.
  2. Sleaford saw no defections from Best of Both and Bregretters.
  3. There will be some Labour to UKIP defections among Brexiteers.
  4. Turnout patterns will be similar to Witney but with the addition of some Non-GE voters.
  5. An additional 2% swing away from Labour needs to be factored in.

A defeat for UKIP will give Labour encouragement that they can defend themselves in the North of England and send UKIP back to the drawing board.  Whilst I am predicting a UKIP win, I am also struck by the distinct chances that the Lib Dems have as well.  This will only happen if Labour Remainers defect in great numbers but it is a scenario that should not be ruled out.

UPDATED FORECAST AS OF 22/2/17

After I published this post, I was analysing some polls by YouGov and I came to the conclusion that there is an 8th dynamic that was present in both the Richmond Park and Sleaford by-elections.

Richmond Park & Sleaford only (additional)

  • Half of UKIP Potential Bregret defect to the Conservatives

This is not a large effect but I think it needs to be incorporated in my scenarios.

At the same time, I noticed that some scenarios that favour the Lib Dems had inadvertently been double counted and I eliminated my most extreme scenario in terms of additional national swing from Labour.  After rerunning the numbers, the revised forecast is now shown in this table.

With the 8th dynamic, the odds on a UKIP or Lib Dem win are now lower and the odds on a Labour win are higher.  I am still using the Variant scenarios for my forecast and basically it is neck and neck between UKIP and Labour with UKIP slightly ahead.  A Lib Dem win can still occur but only if we see a repeat of Richmond Park.