Election 2017

Well, not all the seats have declared but we know the overall result, and it is very interesting, in so many ways.

First of all, polling. Conventional opinion polling is a zombie – the walking dead. In a world when people have multiple dimensions of importance, ‘snapshots’ of opinion don’t cut the mustard.

Let’s not forget something: All the major polls used a ‘fudge factor’ to make predictions and again they were wrong. What is this fudge factor? Many people don’t realise that the polling figures those companies quote are NOT the raw percentages – they adjust them based on an estimate (a.k.a. complete guess) as to what the turnout will be in various groups – most importantly age groups. Every election they get it wrong, they adjust the fudge factor for the next election, then find it’s wrong for that….and we continue to lap up this rubbish.

YouGov’s ‘new’ model is closer to the mark. It attempts to use a variety of data sources – including attitudinal data – to get a better handle on likely turnout, and it models at a more disaggregated level – looking at individual constituencies. Kudos. However, alarm bells should ring because YouGov make boasts about their clever use of ‘big data’. Big data models work brilliantly…..until they don’t. Frequently they are over-fit (to use the statistical term describing the fact that they are tweaked to fit past data brilliantly but which makes them vulnerable to changes in key parameters).

Regular readers will know I ran a choice model of BREXIT preferences. It wasn’t set up to predict the general election, though I made one assumption in order to make a prediction at the bookie anyway. Hint: I didn’t win on my national vote bet but I knew damn well that Corbyn was going to deny May a majority. Thus I only got the magnitude wrong; I did win on my local bet: my local (Labour) MP would only have been in trouble had May’s gamble paid off. Thus:

  • My model wasn’t set up to predict a general election – it was run to understand BREXIT votes and it just happened to be the case that I could make predictions, IF BREXIT were a major issue in the vote – which it only partly was;
  • My prediction was in the right direction (unlike all the main polls). Now, YouGov got it just about right, let’s not forget that. But they don’t use choice models to quantify attitudes so have no ‘proper’ way to identify and quantify any changes in how attitudes translate into votes. So one day their model will go wrong. Meanwhile I knew (and predicted on here) that Corbyn’s losses in heartlands by ‘not being too BREXIT’ would be more than offset by gains in seats with big REMAIN votes. I only got the magnitude of the swing wrong.

When friends seemed scared that the exit poll might be wrong again, I cautioned on twitter after the Sunderland and Newcastle results that these differences in the Conservative-Labour balance were obviously driven by the differences in the BREXIT votes between these two constituencies. This showed time and again in individual seats: REMAIN constituencies (and, in Wales in particular, weakly LEAVE ones) went disproportionately Labour with (by historical standards) massively high turnout. (Turnout was no surprise to me – go check my past entries.)


Second of all, the UKIP effect. Theresa May made a fatal error. She assumed UKIP supporters were one of two groups:

(1) Basically hard-right BREXITEERs who could be taken for granted as they ‘had nowhere else to go’ once she made some hard BREXIT statements. They did have somewhere to go. It’s called ‘home’. I knew this. Because my choice model showed that HARD BREXIT support is waning. No, this doesn’t necessarily mean that loads of LEAVE supporters are reconsidering: but I know for a fact they are more concerned about the single market now on average – they’ve moved from HARD to SOFT. That’s what 9 months of debate has achieved. (And what any model like the YouGov one can’t possibly quantify as it doesn’t ask the right questions.)

(2) Former Labour voters who ‘went UKIP’ and ‘just needed a little push to go all the way to Conservative’. This led her to target loads of Midlands and Northern seats. Targetting people in the opposition camp on an issue that is of secondary importance but doesn’t actually define them? This is otherwise known as the ‘Clinton effect’ (and someone at the Times made exactly that joke). She thought that those UKIP supporters who had come from Labour – and there are LOTS – were ready to move over to the Conservatives once UKIP’s raison d’etre was gone, attracted by their ‘small c’ conservatism on social policies and dislike of the ‘Islington Set’. Unfortunately, this strategy failed almost everywhere. Labour is still tribal in lots of areas and lots of people who voted UKIP in 2015 returned to Labour, particularly when faced with a far more attractive Labour economic manifesto than at any time since the mid 1970s.

In the north and north-east post-industrial BREXIT heartlands there were swings to the Conservatives, but they were small – way too small. Ironically, the only seat where this strategy worked quite spectacularly seems to be one that’s local to me – Mansfield. Mansfield is a former mining area that all Brits reading this could predict what is like now – Americans, think of the worst rust-belt town you can. Hence why it has never elected a Conservative MP. Ever. Until today. The gain in the Tory vote is suspiciously similar to the collapse in the UKIP vote. I’m guessing the previous ‘anti-elite’ UKIP vote did, in fact, lead these people to vote Tory (often for reasons related to social views) – all my non-local buses come from Mansfield and suffice to say I’m not keen to be on one on a Saturday night from Nottm city centre…..Nottingham didn’t get the crime capital label for nothing during my youth in the late 1980s. (I love my city really – honestly! I wouldn’t have returned otherwise.)


Third, the screw you effect. Nottinghamians and others in Notts are, on the face of it, really really bloody minded people who pay attention to our MP. As the Conservative Anna Soubry – who just clung onto the seat of Broxtowe – said, this is mad what’s gone on here, with her winning but Gedling going Labour, Mansfield coming over to her side. Now, for anyone (everyone) unfamiliar with Nottingham, my constituency (Gedling) and Broxtowe (other side of city but similar on all sorts of criteria) should go the same way. Yet my local MP, Vernon Coaker (Labour) increased his majority whilst Anna Soubry fended off Labour. I note that right now at approx 7.30am Soubry’s call on May to ‘consider her position’ (aka p*** off in realpolitik) is the tagline on BBC news. I lived in Sydney 2009-2015 so missed a lot of recent Nottm history but now I’m back I was intrigued and Soubry’s interview with David Dimbleby was top notch British TV. When asked to qualify her description of her own party’s campaign as “awful”, she replied “where do I start?” And she enunciated EXACTLY why people like my mother (former Tory) now vote Labour. F*ck with the elderly and you really ARE “the stupid party”. As far as I have seen there are only two sane members of the Conservative Parliamentary Party and they are both fairly local to me: Anna Soubry and Kenneth Clarke. Soubry more or less said half of her own colleagues are lunatics and she should be applauded. She knows why she is getting voted back into Parliament. Maybe she and my local MP (who always bucks the national trend and gets more votes than the Labour national vote would predict) should be running things? 😉