A good-but-not-great election for Labour

featured image
Tuesday, June 20th, 2017

GQR Vice President Peter McLeod wrote a weekly column on the UK general election for politics.co.uk, where this article first appeared. GQR was pollster to Labour at the 2015 UK general election but does not work with the current leadership.


The fallout from the UK’s (first?) general election of 2017 will go on for months. Theresa May has gone from commanding to crumpled, and the Conservative Party’s reaction to the loss of their Parliamentary majority will be fascinating if uncomfortable to watch – and worrying for the process of Brexit. In the meantime Jeremy Corbyn, his team and the wider Labour party exceeded expectations. When the election was called perfectly reasonable people believed that Labour could poll below 25%; they hit 40. That’s remarkable.

Nonetheless, Theresa May swiftly announced the formation of her new minority government backed by the DUP. Of course she was perfectly within her rights as a May government, with those 10 DUP MPs voting with them, commands a working majority the same as they had before Parliament was dissolved. Labour excitement at a huge gain in vote share and modest gain in seats is justified; Labour satisfaction with a result 64 seats short of a majority is not.

While there will be a tremendous amount more to learn from this campaign, it’s worth taking a moment to, as the Bayesians say, update our priors. The following now seem like more solid truths than they were before:

·         Snap elections are inherently risky. The two elections of 1974 could both be considered “snap,” and resulted in a loss for the incumbent followed by an extremely narrow win for the new incumbent. Gordon Brown’s decision not to go for an election in Autumn 2007 looks a lot more solid now than it did then.

·         It’s difficult to win with a bad economy. Governments don’t tend to do well when they go to the polls amid falling real wages. The polls told the Tories they could buck that trend, but reality has caught up with them.

·         Moreover, it’s a bad time globally for incumbents. Following losses for the US Democrats and French Socialists, the Conservatives are another party diminished by voters’ frustration with the establishment over weak economic growth, migration and terror.

·         Parties need to make clear offers to the electorate that tell them a story about how they will make the country better. Labour did this and gained votes and seats. The Tories offered vague, empty slogans accompanied by a palpable distaste for the entire process of getting themselves re-elected, and went backwards.

By contrast, the following are shakier propositions:

·         Young people and former non-voters don’t turn out. We need more data on who really showed up to the polls before we can say this conclusively, but it’s difficult to see how Labour got the numbers it did without a big increase at least in youth turnout. It’s another question still whether future party leaders can rely on this: Corbyn seems to have a certain magic with this demographic and you can see them slipping away from Labour again without another very inspiring campaign.

·         Campaigns don’t matter. This is a fairly fashionable view and as discussed above, the economic background to the election can’t be ignored. Perhaps it’s usually true because most campaigns are roughly equal in competence. But we learned on 8th June that when there’s a dramatic difference in message, narrative, execution and enthusiasm, campaigns can make huge impacts. Don’t forget that at the local elections Labour had a projected national share of 27%.

There are also some major questions that remain unresolved:

·         UKIP as a “gateway drug.” One of the more obvious stories of the election was the collapse in UKIP’s vote share. This was expected to benefit the Tories, with even former Labour voters who had gone to UKIP in 2015 potentially drifting onward to Theresa May’s party. This clearly did not happen to the extent the Conservatives needed it to – if it happened at all.

·         Can Labour win from the left? The question of how Labour kicks on from here is consuming. The party is 64 seats short of a majority. Even if it won back the rest of Scotland from the SNP, it would still be behind. Tacking to the centre risks losing the young people’s votes that helped so much. But those hoping to restore the party to government need some strategy to win back English seats from the Tories. They will need to work out whether they have hit the ceiling of what a left wing proposition can deliver, or whether there’s more still to come.

I wrote on the eve of the election that Labour had not solved its problems around trust on public spending and leadership, while immigration could return as an issue depending on how Brexit goes. An election loss means these are all still real issues. Labour was right to be prepared for an election this year and it should remain in a state of readiness as goodness knows what lies round the corner.

Finally, a note on the pollsters: YouGov and Survation emerge with the most credit. The former built a seat and vote share prediction model based on huge quantities of fieldwork (7000 interviews per week) plus the now-famous Multilevel Regression and Post-stratification (or MRP) that converted that data into seat-by-seat estimates. As the results came in, it was quickly apparent that the model was picking seat outcomes with uncanny accuracy, especially in England and Wales. MRP clearly has a future, although don’t expect to see the standard newspaper voting intention poll using a sample of 7000 any time soon. Meanwhile Survation stuck to its guns on a final poll showing a 1-point Tory win in terms of vote share compared to the 2.4-point actual margin. It was the only pollster to pick Labour’s share of 40% and it missed the Tories’ by only a point. The next-best final poll with a standard methodology was YouGov’s giving the Tories a 7-point win. The problem for the other pollsters seems to be that they were correcting for the errors of 2015 rather than measuring the race as it stood at the time. The challenge next time around will be not to replicate Survation and YouGov’s methods, but to pick up on the most decisive factors for the next election. This is of course far from trivial.