The 2017 British general election will be the third general election that I’ve followed in detail. I wasn’t expecting one to come so soon after 2015, but here we are. I’m neither British nor do I live in Britain, so my interest in this affair may seem disingenuous to some, but after witnessing my own country fall into the hands of a cartel masquerading as a political party, this election offers politically-minded individuals such as myself the ability to enjoy the “fun” of watching an election from a more detached point of view. Today, June 8th 2017, is election day. Unfortunately, today also looks set to be a bad day for those who, like myself, are sympathetic to Jeremy Corbyn and Labour.
So, what exactly is going on in Britain? On April 18th 2017, PM Theresa May announced her attention to hold a snap election on June 8th. At the time, this plan seemed to make a good deal of sense: a general election would give May the opportunity to bury Jeremy Corbyn’s Labour, stave off potential challenges from within her own party, and push back the deadline for the next GE from 2020 to 2022 — creating a nice gap between that vote and the Brexit deadline in March 2019.
Of course, this clever strategy relies on one important factor: winning. In April, May’s victory over Corbyn seemed inevitable. Under his leadership, Labour had been polling terribly for over a year. A YouGov poll on April 13th, the week before the snap election was called, had the Conservatives at 44% and Labour at an abysmal 23%. Ever since the EU referendum, Labour had struggled to reach 30% in the polls. So, May can perhaps be forgiven for believing that defeating Corbyn’s Labour would be an easy task: that was the conventional wisdom, after all.
Of course, conventional wisdom has a way of falling apart, and according to some pollsters, it indeed has. YouGov’s most recent poll has the Tories at 42% and Labour at 35%, indicating a slight decline for the Conservatives and a very impressive rise for Labour (though Labour has dropped a few points in the YouGov poll over the past few days). A 7-point margin is still a fairly comfortable lead, of course and one must remember that the British GE is a parliamentary election, not a direct one. The voters are not actually voting directly for May or Corbyn, instead, members of each one of Britain’s 650 constituencies vote for a member of parliament, and the resulting layout of parliament determines who gets to be PM. YouGov’s model, which attempts to estimate the composition of the next parliament, currently predicts that Conservatives will end up with 302 seats in parliament, down from their current 330 and well short of the number needed to form a majority government (that number is technically 325, though lower in practice due to Sinn Féin). Labour ends up with 269 seats in this model, with the Scottish National Party at 44 and the Liberal Democrats at 13. If YouGov’s model is accurate, the result will be a hung parliament, with no party possessing a clear majority. It will be up to the parties to then decide on forming a coalition or minority government.
YouGov, however, is far from the only polling agency out there. Survation’s telephone poll — which proved accurate for the 2014 Scottish referendum and the 2016 EU referendum — has the Tories at 41% and Labour at 40%. Before one starts betting on Corbyn being the next PM, though, it is worth noting that Survation’s poll is a massive outlier. ComRes gives the Conservatives a 10 point lead, 44 to 34. ICM Research’s poll is even more bullish for the Tories, giving them a 12 point lead, 46-34. BMG Research’s results are similar — Tories lead Labour 46-33. Britain Elect’s poll of polls currently has the Conservatives leading Labour 43.6 to 36.1, with the Lib Dems a distant third at 8.1. The Financial Time’s poll of polls delivers a similar result — Conservatives 43 Labour 36.
Whilst YouGov’s model has gained a large amount of attention due to its prediction of a hung parliament, other models give the Tories a comfortable majority. Britain Elect’s Nowcast predicts that the Conservatives will increase their majority to from 330 to 353, with Labour falling from 229 to 219. This would mean that despite the narrative of May’s ineptitude and Labour’s rise, Labour would perform even more poorly than they did under Ed Miliband in 2015 when they won 232 seats. Lord Ashcroft’s model is similarly glum, with the Tories capturing 357 seats to Labour’s 222. Matt Singh — who became well-known in 2015 for predicting the polling failures of that election — sees the Conservatives capturing an amazing 374 seats to Labour’s 207. Gloomiest of them all, from a pro-Labour standpoint, is independent statistician Nigel Marriott’s model, in which Labour wins only 202 seats to the Conservative Party’s 375. If Singh is correct, this could be Labour’s biggest defeat since 1983. If Marriott is correct, it would be the biggest defeat since 1935.
So, what explains the volatility of this election? The general narrative, after all, is that Theresa May has run a terrible campaign, characterized by her wooden nature, her refusal to hold conventional debates, and bizarrely cruel policies such as the proposed “dementia tax” and cuts to school lunch programs. Jeremy Corbyn, on the other hand, has benefited massively from the spotlight of the campaign. He’s been able to connect to the voters in a way that he simply wasn’t able to before the election was called.
The tightening of the polls indicates that this narrative has more than a bit of truth to it. However, the sheer deficit with which Corbyn started the campaign with may prove fatal for his leadership of the Labour Party. 36% is much better than 22%, but without the context of the Labour rise it is hardly an impressive figure.
Labour supporters have sought solace in the youth vote as a potential tipping point in this election — probably not one that can give Corbyn a majority, but perhaps one that can reduce or even eliminate May’s. An ICM poll shows Labour leading the Tories 73-15 amongst 18-24 year olds. The flip side of that, of course, is that the Conservatives lead Labour 64-20 amongst the 65 and above age bracket, and that demographic tends to turn out more reliably than the youth vote.
I think that the final result in terms of the popular vote will be akin to Britain Elects’s and FT’s polls of polls — so, around Con 43 Lab 36. Survation’s poll, which puts the gap at one point (41 to 40), relies on sampling a “pre-stratified sample of individuals” that are chosen as being “representative of all persons in a chosen area,” in this case, the United Kingdom. In this way it is somewhat similar to the University of Southern California/LA Times poll that predicted Trump’s victory in the 2016 US presidential election. However, in my view the Survation results are inaccurate, likely caused by the poll accidentally oversampling from people who are more favourable towards Labour on average than the general public. I just can’t see Survation’s results coming true — after all, the other national polls in the presidential election were mostly true in that Clinton did win the popular vote. It was the models based on state polls that dropped the ball.
Speaking of models, I suspect that YouGov’s model, which is an outlier when compared to the others mentioned here, will be proven faulty. YouGov’s Anthony Wells stated that “from the pollsters’ point of view this is an experimental election. We all got it wrong in 2015 and we are all trying different methods to get it right this year.” I actually don’t think that the YouGov poll, which has shown a widening lead for the Tories since Wells’s post was published on June 1st (the latest poll he was working with had a 3 point gap between the two parties, the June 7th poll has it at 7), will be that inaccurate. I do think that YouGov’s model, on the other hand, will be proven wrong. This model works via a different methodology than YouGov’s standard poll and predicts a 42-38 popular vote share for the two parties and a hung parliament with the Tories shrinking to 302 seats and Labour increasing to 269.
Modeling a British general election is more difficult than doing the same for a US presidential election due to the nature of the two systems. The latter is determined by an electoral college composed of 50 states plus the District of Columbia. Each one of these areas can be polled and a model can thus be created. This isn’t to say that the model will be accurate, of course — in 2016, many were not. However, it is still considerably easier than creating a model for an election involving 650 constituencies. Nobody is polling all 650 constituencies independently, so a fair amount of extrapolation and guesswork will be involved in any general election model.
YouGov’s model, the methodology of which is explained here, is created by people far more intelligent and experienced than I am. YouGov’s modeling of the 2016 presidential election was more accurate than most American-made models, after all. However, there are reasons to believe that the geography of the election will more closely resemble the outcomes predicted by the likes of Britain Elects, Lord Ashcroft, Singh, and Marriott.
Labour Uncut’s Atul Hatwal has stated that Labour is likely to lose big in the election, “based on dozens of conversations with activists, candidates and officials who cumulatively had sight of tens of thousands of canvass returns.” In Hatwal’s view, Labour’s rising poll numbers have been driven by a genuine surge in interest in safe Labour areas in and around London. FT’s Jim Packard and George Parker have reached a similar conclusion based on their contacts within Labour: “Asked if Labour was heading for a good night, one well-placed party source replied: ‘No. Not at all. Not one bit. They are all wrong.'” If Hatwal and the FT are correct, Labour may lose several seats throughout the Midlands and the north, similar the February Copeland by-election that saw Labour lose that northern constituency to the Tories for the first time since the 1930s.
It’s worth taking time to note how I’ve barely mentioned the Liberal Democrats in this post, and I haven’t mentioned UKIP or the Greens at all. It seems like this will not be a good year for most third parties — indeed, a return to the usual two-party system may be one of the factors behind Labour’s rising poll numbers. The Lib Dems will probably get a higher share of the popular vote than the Scottish National Party, but the SNP’s strong regional concentration means that they will continue to be the most powerful third party in parliament. It also means that Labour will once again have to do without their former stronghold of Scotland.
My prediction is that Labour is set for a defeat. Whilst I do not think it will be as disastrous as some of the predictions imply, I do think that they will fail to close the gap to the Tories, and that gap may indeed widen despite May’s disastrous campaign due to Labour losses in the north and the Midlands. I would love to be proven wrong, as between the two candidates I favour Corbyn (though the Lib Dems are closest to my ideal). I fear, though, that the recent exultation around Corbyn is very premature, and that the Tories will once again prove their electoral effectiveness.
For fun, I thought I’d take my crack at predicting the results of the general election. I call this the “JaeCast,” though it is not really a model as I’m not really doing any mathematics here. I’m literally just picking numbers based on the other models I’ve seen. Note: I haven’t included the 18 seats from Northern Ireland in this table. Anyways, you can all laugh at me when this prediction is inevitably proven to be laughably inaccurate by the actual results: