Monday 9 May 2016

Marking our own work on 'Super Thursday'


One of the most refreshing and exciting things about academic life is that you don't always have to pretend that you're right all the time. Well, you work or pose as an 'expert', but in reality the legs are firing away under the water, working on whatever you're worried about right now - over and under, over and under.

That's because the academic method itself is designed to help you understand where you got it wrong, firstly so as to get things more accurate next time, and secondly to reveal the moving parts of history, economics, sociology or whatever in a way that isolates the 'wrong' or inaccurate bit of your data, reasoning or imagination. That's a fascinating reveal in its own right, because by exposing the 'wrongness' you might get closer - though not perfectly close - to 'rightness'.

Anyway, enough of the pompous preface: how did we do last week, when we ventured out to make predictions about the UK's 'Super Thursday' of local, city and devolved elections? Where were we wrong, where did we get it right, how inaccurate were we, and what does that tell us about British politics? Here's a list of predictions, along with some ratings - and a grade. Marking your own work can be cathartic, and we hope you think we've been fair - or even harsh - since only from such judgements can better forecasting spring and grow.

Scotland. Well, this was a right blowout. Sorry, but we just messed this up. We thought that the Scottish National Party would continue to govern as a majority at the Holyrood Parliament. They didn't make it. We said that Labour would just make it to second place. They fell well short and, in truth, they got absolutely hammered by a resurgent Scottish Conservative Party that has now eclipsed them to become Scotland's official Opposition. What a pain. The main lines of interpretation were right - the big picture is that SNP First Minister Nicola Sturgeon can go on governing - but the details were all wrong, wrong, wrong. The lesson? We should have listened to the anxious voices from within the SNP - so reminiscent of intelligence from within UK Labour in the run-up to last year's UK General Election - and we should have taken even more note of the relative popularity of Ruth Davidson, the Scottish Conservatives' combative and likeable leader in an age where charismatic and even edgy leadership seems more and more important. Head over to Matt Singh's Number Cruncher Politics if you want to see someone call this one much more accurately, by the way.

Mark: C (55%).

Wales. Here we did much better. Not only did we correctly see from the polls that the United Kingdom Independence Party would gain six or seven Members of the Welsh Assembly (in the end, they got seven), but we said that Labour would secure '27 or 28' AMs. They got 29, partly because they got very, very, very lucky in some of their marginal seats and on the regional list even at the same time as their vote share plunged downwards. YouGov, along with most other pollsters, emerged as one of the heroes of the campaign: their numbers were very close to the final result, and they managed to eliminate the overestimate of Labour's likely vote share that had been so clear in previous campaigns. Well done to them. So: not bad, though what we can take from these results is that very fine margins can make for very large results. Had Labour only managed 26 AMs, as they might well have done given their much lower popularity this time, they would now be sweating about how to govern. As it is, they can pretty much breeze on as before - with the proviso that they are on probation from an electoral system that just pulled their irons out of a very hot fire.

Mark: B+ (67%).

English local government. This one was even better. We picked up accurately that Labour might not get quite the hammering that many social science models were predicting from the opinion polls, and from local by-election results. Something about the result in last year's Oldham West by-election gave us pause, and the Conservatives' rapidly falling polling since their badly-received Budget told us to be cautious, before we piled in behind everyone else. This was proved an entirely accurate instinct: there's an interesting phenomenon taking hold, in which Labour voters complain vociferously about the party's leader and splits on the doorstep, and deride both in focus groups, but just about cling to prior loyalties in the polling booth. That might not last, but it's holding for now. Labour had a very, very, very bad night by historical standards - one of the worst nights for an Opposition in living memory - but they did not quite fall into the black hole that many of their ridiculous and absurdist Westminster contortions might have led you to expect. Again, they got lucky - if you can call almost-certain Opposition until at least 2025 any measure of luck - as their very low vote share did not quite translate into the loss of councillors. We also thought that the Liberal Democrats would move forward just a bit (they did), and that UKIP would do quite well (they slightly under-performed on their own hopes in England, slipping back again in the projected National Share of the Vote). Apart from that UKIP glitch, bringing to mind the lesson that most big talk polling and polling will rather overstate UKIP support, this was a bullseye. It gets a top mark (thanks very much).

Mark: A (76%).

London. Bang on here too in predicting Labour's win in the capital, though this was easier since the polls had put the Conservatives way behind Labour's Sadiq Khan for many weeks, and the Conservatives had accepted for quite a while that they had lost. In the event, the polls got this one almost exactly right as well, so although we thought differential turnout might make Mr Khan's victory slightly smaller than it looked, there was only really a very small difference between opinion surveys and the contents of London's ballot boxes. Labour didn't really move forward much in elections for the London Assembly, standing still on 2015 as elsewhere, and Mr Khan didn't push Labour's vote up much even in the Mayoral contest; but he did manage to make his main opponent, Zac Goldsmith, look ridiculous simply by virtue of keeping his head and looking calm while Mr Goldsmith raged about Mr Khan's supposed (and exaggerated) links to Islamic extremists. There's a tutorial in that, all on its own. Had Mr Goldsmith remained truer to himself, like the redoubtable Ms Davidson, he would probably have won more votes - and more friends.

Mark: A- (72%).

So the final mark? That's basically running at an average of 67%, or a B+. It's better than the B- we'd probably award ourselves for thinking that last year's General Election would end in a Labour defeat, but not a Conservative majority. We'll take that, as a B+ level of accuracy is probably at the limits of the controllable and the forecastable in any case. There's yet another lesson, right there.