Sunday, 11 September 2016

This Is An Interesting Paper On Picking The Winner With A 95% Probability

Labour leadership candidate Owen Smith is so unbelievably inept, unworldly, arrogant and socially maladroit that the Independent is starting to wonder whether he's actually a Corbynite sleeper agent.

After all, when a politician is quite so gauche and incompetent, you begin to entertain the idea that even conspiracy theories about him could be credible, particularly when even in a party with more third rate politicians in senior positions than ever before in my lifetime, there are evidently less-ridiculous leadership candidates than Owen Smith.

Still, the Owen Smith story will surely be very short-lived - he has almost zero chance of winning the Labour leadership contest, and even less chance of ever seeing his party win an election with him at the helm, so it looks like Britain is going to see a May vs. Corbyn battle in the next General Election (probably 2020).

On that topical note, today I stumbled upon a very interesting paper, in which the author Andreas Murr from the University of Oxford predicts that according to a Bayesian analysis there is a 95 per cent probability that having the larger winning margin in party leadership elections increases the chances of winning the General Election, and that the party leader with the larger winning margin will almost certainly become the next Prime Minister.

Murr was one of the few people last May to predict what most people didn't expect - that David Cameron, not Ed Miliband, will be Prime Minister for the next term. Of course, given that the PM in opposition Theresa May doesn't have a winning margin due to the fact that the withdrawal of Andrea Leadsom in the leadership race left her as the sole candidate, there is difficulty in comparing how the respective 'winning margins' would play out.

However, given that leaders are so important to the party's chances of winning an election, and that the principal goals of a party are to ensure in-group solidarity and cohesion (as much as possible) and to be popular enough to form a government, I suppose the goal of all parties is to have a leader that best represents a kind of weighted average of the nation, rather than a leader that best represents a kind of weighted average of the party.

To that end, then, Theresa May with her crypto-socialism and small 'c' Conservative identity combination is a country mile ahead of Jeremy Corbyn in terms of probability of her party winning the next election, because she much more closely resembles a weighed average of the nation than Corbyn. That probably explains a fair bit about why Theresa May currently has a whopping 42-point lead in favourability ratings over Jeremy Corbyn.