Almost all the doorknocking has been done. The campaign stops are finished. The car crash interviews are over. Election Day is upon us.
And Election Day wouldn’t be Election Day without a pseudo-scientific forecast on what the result will be.
Based on the final call from the British Polling Council pollsters, (Ipsos MORI are yet to complete there’s), historical results and reports from the campaign, in this post we take a look at what the vote shares are likely to be, what that will mean in terms of seats and who is set to form a government.
So who is going to win?
There has been a significant narrowing in the polls over this campaign. At the start we were looking at 20+ point Tory leads, now it’s as narrow as one point in some polls.
There has clearly been an uptick in Labour support towards 30%. Part of this is probably the ‘I’m Normally Labour But’ who were not voting/voting Tory because of Jeremy Corbyn’s leadership. These voters have firmed up for Labour as fears of a Tory landslide grew and Corbyn’s ratings have generally improved.
The other part is the drift from other parties. Most notably the falling back of the Lib Dems to their 2015 level, presumably as part of a consolidation of the anti-Brexit/anti-Tory vote. On a smaller scale the Green’s have seen their vote collapse to roughly 1% (down from 4% in 2015). Corbyn’s Labour and the Green’s have much in common, so it’s easy to see how that’s happened.
That brings Labour up to the early 30s, making up lost votes to the Tories and perhaps even advancing a little from 2015. But that doesn’t explain how some polls have them depriving the Tories of the majority and others see Theresa May returning with a comfortable majority.
Why the difference?
The difference revolves largely as to how pollsters model turnout. What is driving Labour’s polling improvement is a large increase in young voters saying they will vote, particularly 18-24 year olds and Labour have a big lead in this demographic. There has actually been very little Conservative downward movement, and they retain a big lead in older voters who historically turn out in much higher numbers.
Pollsters ask respondents how likely they are to vote, usually on a scale of one to ten. Some pollsters like YouGov take these responses at face value. If a young voter says they are likely to turn out, they put them down as likely to turn out.
Other pollsters, like ComRes and ICM, take a different approach, feeling that historical results show younger voters are much less likely to turn out than they say they will. They adjust their weighting accordingly and this gives the Conservatives bigger leads in these polls.
Which is right?
Neither method has put Labour with a hope of forming a majority, or even forming a minority government or coalition. But clearly there is a big divergence with Survation suggesting a Tory lead of just one point, and BMG suggesting a lead of 13 points.
A good rule of thumb is to take the polling average. That is (changes from 2015):
CON 44%(+6) LAB 36%(+5) LD 8%(nc) UKIP 4%(-9) GRE 2%(-2)
But given the big range between the polls an average is difficult because we’re not comparing like with like. An average is a good starting point but is incomplete.
We have to ask ourselves, then, will young people turn out more than before?
There are some arguments to suggest yes. Young people have had a miserable time in politics of late. They strongly supported the Lib Dems in 2010, only for them to lose seats. They supported Labour in 2015 with great expectations of victory only to see a Conservative majority. They overwhelmingly backed Remain only to see Britain leave the EU. The victory of Donald Trump has also certainly agitated the liberal youth as well. Could it be enough is enough and they finally commit to something?
Additionally, Jeremy Corbyn has offered a platform that is at least superficially attractive to younger voters. Just like the middle classes vote for Tory tax cuts, could younger voters be driven out by the abolition of tuition fees? Will Generation Rent come out to support rent controls?
On the other hand, those successive defeats could have lead to demoralisation just as much as it could lead to radicalisation. Even during high turnouts young voters still turn out poorly. Turnout for the 2014 Scottish referendum was 85% overall. But for 18-24 year-olds this was only 54%, compared to 96% for 65+ year olds.
Turnout remains poor even if there are plausibly “young-friendly” choices. Tuition fees didn’t boost turnout in 2010, with only 44% of young people turning out compared to 65% nationally. Milifandom wasn’t enough to boost turnout beyond that figure either.
On the historical evidence, those turnout modelled polls seem to be favoured so that would suggest a few points higher than the average. Our political culture does seem different to 2015 though and does feel like young voters may turn out more. In which case that would put the poll lead not too far above the average.
One way to get to that is discounting Survation polling. Survation is currently suggesting a turnout in excess of 80%. Turnout could plausibly be better, but nowhere near that good. Let’s remove that from the average, which gives us:
CON 44% LAB 35% LD 9% UKIP 5% GRE 2%
Might all the polls be wrong?
Polls had a poor showing in 2015 and didn’t do much better in the Referendum either. But we have previously seen that polls aren’t actually too bad. Pollsters have also made many methodological adjustments since 2015.
That being said there is the historical elephant in the room. Even in their better years, polls have consistently underrated the Tories. Analysts have tried to come up with a loose ‘rule of five’ – expand the Tory lead over Labour by five points.
CON 46.5% LAB 32.5% LD 9% UKIP 5% GRE 2%
That does make the Conservative lead seem a little too high. We have also after all factoring in some underestimating by discounting Survation as well.
However, in actual voting the Tories have done pretty well this year. They enjoyed a big win in the Copeland by-election and the recent local elections. This comes on top of pretty good results elsewhere too. Again both these suggest a Tory lead bigger than the polls.
Let’s settle for a modest adjustment of one point either way, giving a final adjustment of (changes from 2015):
CON 45%(+8) LAB 34%(+3) LD 9%(+1) UKIP 5%(-8) GRE 2%(-2)
Vote shares are all well and good, but under First-Past-The-Post, it’s translating them into seats that matters.
The first step is to use uniform national swing (UNS). UNS takes the change in the share of the vote for party compared to another. The current UNS between Labour and Conservatives based on the forecast is 2.5%. In theory, that means in every 100 2015 Labour voters are now voting Tory. Applied uniformly in every seat across the country this would mean seat totals of (changes from 2015):
CON 349(+19) LAB 217(-15) SNP 55(-1) LD 6(-2) PC 3(nc) GRE 1(nc) UKIP 0(-1) OTH 19(nc)
On the margins
Of course, voters don’t act uniformly with certain geographical areas voting differently to others. To make gains parties need to ensure they get votes in areas where they need them most: marginal seats, those seats that are likely to change hands.
It’s not use Labour gaining votes only to pile them up in their safe seats. Unfortunately for them this seems to be exactly what they are doing.
Labour’s vote is holding up and perhaps even advancing amongst the young, urban and ethnically diverse voters. These voters tend to overwhelmingly be in cities like London, Manchester and Bristol where Labour already do well.
Despite the social care fiasco, the Tories continue to have towering leads amongst older white voters of the lower middle classes. These voters are placed in just the kind of areas the Tories need them, marginal suburban seats in the Midlands, Yorkshire and the North East.
There’s some anecdotal evidence to further back this up. Labour candidates have reported disappointing doorstep responses outside their London heartland.
More compelling is the Tory campaign strategy. The Tories seem remarkably relaxed for a party that might be in danger of losing their majority. Indeed, Theresa May continues to park her tanks on Labour lawns. This suggests their internal polling and candidates are giving feedback they are set to make big gains.
History also suggests the Tories will outperform UNS in marginal seats, just as they did in 2015. Why? Partly it’s because this is where the ground war is fought and the efforts made, voters will have been engaged more. The Tories generally seem to have a good track record of winning over marginal voters too and have clearly focused their campaign here.
Almost entirely arbitrarily let’s add an extra 2.5 points to the swing in Lab/Con marginals. That’s based on how the Tories did in the recent local elections and in the Copeland by-election. That leaves us with:
CON 376 LAB 190 SNP 55 LD 6 PC 3 GRE 1 UKIP 0 OTH 19
This election has seen something of a realignment, with many Leave voters coalescing around the Tories. Many of these were Labour voters, and again, in those Midlands/Northern seats where the Tories are reasonably competitive.
That puts the Tories likely to pick up seats beyond what are considered marginal. Roughly half of the 2015 UKIP vote seems to be going to the Conservatives. That alone could put them over the top in 35 Labour seats, without any need for Lab/Con switchers. It will also mean picking up Lib Dem seats with big Brexit votes.
The effect is slightly limited by the apparent consolidation of the Remain vote behind Labour. Labour also get some of the UKIP carcass with polls suggesting they’ll get about 1/6 of their 2015 vote.
But the overall takeaway is that the Tories will capture some of their more ambitious targets. This will add around four to the Tory total.
CON 380 LAB 187 SNP 55 LD 5 PC 3 GRE 1 UKIP 0 OTH 19
Labour do seem to be performing particularly well in London, with one poll showing a swing away from the Tories. That makes sense given its heavy support for Remain, its diversity and younger population.
If that holds out it would mean some unexpected Labour defences in marginals like Ealing and Acton and Brentford and Isleworth.
The Labour vote also appears to have recovered well in Wales. Again this could easily mean defending themselves against all Tory advances.
Labour’s success with younger and urban voters should be replicated in other strongholds like Birmingham and Bristol. A gain in Brighton Kemptown from the Tories is feasible. Overall this will add around 18 back to the Labour total.
CON 362 LAB 205 SNP 55 LD 5 PC 3 GRE 1 UKIP 0 OTH 19
One subplot is the election in Scotland. Here the Tories are enjoying a major revival with a swing of around ten points from the SNP. That would add a further six to the Tory total.
The Lib Dems are also likely to benefit from unionist tactical voting to give them two more to their total from Scotland. Labour could also benefit from the same, also seeing two extra seats on their total from Scotland. That leaves a GB outcome of:
CON 368 LAB 207 SNP 45 LD 7 PC 3 GRE 1 UKIP 0 OTH 19
Northern Ireland is important too. Not just because it is actually part of the UK, but it also has relevance for the seats needed for a working majority. As Sinn Féin don’t take their seats in Westminster, the more seats they have, the less needed for a working majority.
As we know, few seats are likely to change hands in Northern Ireland. Current polling suggests that only Fermanagh and South Tyrone will switch, flipping from the UUP to Sinn Féin. That would give Northern Ireland results of:
DUP 8(nc) SF 5(+1) SDLP 3(nc) UUP 1(-1) IND 1(nc)
Take those five Sinn Féin MPs away, alongside the Speaker (who only has a tie-breaking vote) and the winning line for an overall majority is 322 seats.
Overall, we get:
(GB*) Vote Share: CON 45%(+8) LAB 34%(+3) LD 9%(+1) UKIP 5%(-8) GRE 2%(-2)
Swing: LAB to CON – 2.5%
Seats: CON 368 (+38) LAB 207 (-25) SNP 45(-11) DUP 8(nc) LD 7(-1) SF 5(+1) PC 3(nc) SDLP 3(nc) GRE 1(nc) UUP 1(-1) Speaker 1 (nc) UKIP 0(-1)
CON Overall Majority of 86(+76)
CON Working Majority of 92(+78)
*Polls always exclude Northern Ireland, so the UK-wide vote share will be a little lower for the main parties.
WHAT DOES THAT MEAN?
In the vote share, both main parties will see significant increases and it will be their biggest combined share since 1979.
The Tories will have the best vote by any party since 1970 and the biggest lead since 1997. Labour will have their best performance since 2005 and would be the first election since 1997 to see their vote share increase.
In terms of the seats, the Tory majority will be a little shy of what Thatcher got in 1987 as indeed would be their seat totals. Though Labour get a good vote share increase, their seat totals would be the worst since 1935.
Theresa May called this elections for three main reasons. 1. To get a wriggle room for Brexit negotiations. 2. To dilute the authority of Tory dissidents that caused u-turns on issues like the Budget. 3. To lock the Labour Party out of power for a generation.
On the first two this result would easily get there. She’ll have enough room to ignore the hardcore Brexiteers and Tory Remainers. A Thatcher-sized majority will also make it much easier to pass what she likes.
On the third objective she would have fallen a little short. There will be some disappointment in the Tory party at the squandering of the massive poll leads that could have really threatened the Labour Party’s existence.
This would be a bad defeat for Jeremy Corbyn with Labour going significantly backwards for the fifth successive election. Returning to power will be an even bigger challenge.
But in fairness this was that first election since 1987 that Labour fought without any realistic chance of winning. They also would have avoided a complete catastrophe and Corbyn would be able to point to a vote share not far of Blair’s in 2005.
This is all predicated on a shaky forecast of course and with the variance of polls it is difficult to commit with any great confidence with polling divergence. Perhaps the young will turn out in droves, pushing us towards a hung parliament. Maybe they won’t turn out at all and the Tories would have been underestimated again, pushing them towards a majority of over 100.
We will have to wait and see.