• Submission to Ofcom on media plurality
• Submission to the Commons Select Committee on Culture, Media and Sport

I’ve just finished a very preliminary draft of a paper on the Labour leadership elections. It draws on the same data I noted earlier.

The abstract is as follows:

I analyse the preference data of Labour MPs and MEPs in that party’s leadership election of 2010. I show that the primary dimension of competition is a left-right dimension, and that the second dimension of competition is a valence dimension, separating ‘nice’ (unthreatening) from ‘nasty’ (electorally competitive, potentially divisive) candidates. In both cases, the eventual leadership winner, Ed Miliband, is located close to the (dimension-by-dimension) median voter in the Labour party.

The article is available here, and here are the results for the two-dimensional solution.

• Yes victory: 48.8%
• No victory: 51.2%

Or in other words, pretty even.

Andrew Rudalevige explains the significance of 55% very succinctly at the Monkey Cage. (I’m not aware of any other supermajority requirements in UK politics, but pipe up in comments if you know of some).

It’s an ugly solution. I can understand the need to bind the coalition together — and indeed to make it more difficult to bring down the government — but there are a number of intermediate solutions.

One could require confidence motions to have the support of a majority of members of the House, instead of a majority of those voting. As far as I can see, this would have prevented the 1979 fall of the Callaghan government: Callaghan lost 311-310, but a 50%+1 majority would have been 318.

The second option would be to go the German route, and require that votes of no confidence be constructive. But this wouldn’t protect the Tories against a LibDem defection to a Rainbow coalition.

Neither solution would bind the coalition so totally as the 55% solution does — but neither is so objectionable.

Update: Nice briefing note from the Constitution Unit at UCL:

“Some commentators appear to have confused a dissolution resolution moved by the government, and a confidence motion tabled by the opposition. On no confidence motions tabled by the opposition parties, the normal 50% threshold should continue to apply.”

I don’t know what this means for arcane debates about uniform versus proportional swing; the swings during the night were all over the place. It’s possible that uniform national swing is still a better model than proportional swing, but the stochastic element looked huge. (And of course, it’s the stochastic element that the Curtice et al projection model models).

The Conservatives have now reached out to the Conservatives; former Prime Minister John Major has suggested that having `one or two’ LibDems in the cabinet would be a price worth paying to have stable government quickly.

I would suggest that’s a bad deal for the LibDems other things being equal (and bear in mind that the big other thing is electoral reform). One of the best supported law-like statements in political science — perhaps more law like than Duverger’s Law — is Gamson’s law.

Gamson’s law states that parties receive cabinet seats in proportion to their contribution to the seats in the majority.

At the moment, the Conservatives have 306/7 seats, the LibDems 57. That means that the LibDems have roughly 16% of the seats in a hypothetical Conservative-Liberal Democrat majority.

Now, the last Brown cabinet had 23 ministers. 16% of that is a little bit over three and a half. So a `proportionate’ share of ministers would be bigger than the number Major suggests.

UPDATE: More on this, together with normalized Banzhaf scores (woo!) at the LSE Election Experts blog.

In a letter to Professor Nutt, Mr Johnson writes: “I cannot have public confusion between scientific advice and policy, and have therefore lost confidence in your ability to advise me as chair of the ACMD… I would therefore ask you to step down from the council with immediate effect.”