Getting specific on PR

Judging by the reaction to previous posts here at The Sharpener, you, dear readers, cannot get enough of this electoral reform stuff. So, in the greater service of interested humanity, I thought I would write on the subject (hereafter re-branded the Fair Vote) once more.

But this time I’ll try to answer a specific question: how would the political map of Britain look if we went ahead and introduced this foreign, European electoral muck?

(Actually, that’s another myth: Anglo political philosophers like Mill and Hare brought PR to the world, much lively political debate in Edwardian England centred around PR, and introducing PR-STV for the House of Commons only failed to pass in 1917 by 7 votes after, oh the irony, mass defection to the anti-camp by Liberals.)

I’m basing my calculations on a system of regional, open-list PR and using votes cast two weeks ago as the basis for calculation. Under this system, each voter casts one vote for a candidate only, just like now. There is no preference voting (1st, 2nd, 3rd choices etc.) and no ‘party list’ voting (putting a cross in a box marked “Labour”, for example). Each vote cast for a candidate does count towards the total regional votes for the party (unless he/she is an independent), but is only used for calculating the distribution of local seats to the party, not for deciding which party members take up those seats. So, one voter, one vote: PR needn’t be complicated.

There are problems with using the data from the 2005 election. Most crucially for these re-runs, it is useless for illustrating which specific MPs would have been elected to the Commons. The New Hero of the American Left, for example, wouldn’t have been standing in Bethnal Green and Bow, but (alongside other Respect candidates) in a constituency called, perhaps, London East–Central. It is impossible to estimate how many votes he would have picked up. So, in the tables below, I only show how many MPs from each party in each region would have been elected. Remember: that doesn’t mean it would be in the gift of the parties to bestow the seats. Seats would go to the party members who polled the greatest number of votes in the constituency. (A slight modification on this has the parties nominate a list order in advance, with a quota above which any individual candidate must poll to ‘jump the queue’ within his/her list. Obviously this quota can be designed to give party or the individual primacy.) This system, by the way, also has zero impact on the ability of independents to get elected.

In addition, the raw voting numbers are heavily polluted with tactical voting, and so this little exercise is unlikely to produce the ‘will of the people’ writ large and precise. Still, the results are more valid than attempts to estimate what would happen under preference voting systems like AV, AV+ or PR-STV. That really is guesswork.

To calculate the results, the total votes scored by each party in the regional constituencies (some designed by me, some taken from the Jenkins Report of 1998) are first summed. Seats are then distributed among the parties using the D’Hondt highest average method. This is commonly used in PR systems worldwide, and was chosen for the UK as it tends to favour larger parties (in contrast to, say, the Sainte-Lague method) and, along with small-ish constituency sizes, could be expected to produce outright governing majorities for parties scoring in the mid-40%s, which doesn’t seem unreasonable.

D’Hondt works on a system of divisors (1,2,3,4,5 etc.): the party with the greatest number of votes is awarded a seat, then its total vote is divided by the next number in the sequence. Then another seat is awarded, to that party again if after division their ‘average’ is still ‘highest’, or to the party with the ‘highest average’, whose vote is then divided according to the sequence. And so on. This continues until all seats in a regional constituency are allocated. (Have a look at the tables below — it’s very simple, honest.)

I’ve done the calculations for a sample of proposed constituencies, from small, unitary and urban (e.g. Brighton – currently 3 seats) to large, varied and urban (e.g. London North East – currently 10 seats) or rural (e.g. Cambridgeshire – currently 7 seats). The results present a snapshot of how Britain really looks opinion-wise without the polarizing and distorting lens of FPTP.

The fair vote won’t cure the British polity. Douglas Rae’s maxim – that “like the Sheriff of Nottingham, electoral systems are apt to steal from the poor and give to the rich” – will still hold. But at least we’ll be able to collar the real perpetrators.

Re-runs
Cambridgeshire (7 seats) is constructed from the 4 current Cambridgeshire constituencies (NE, NW, S, SE), plus Cambridge, Peterborough and Huntingdon. It was suggested as a possible top-up region by the Jenkins Report. Currently it has 6 Conservative MPs and 1 Lib Dem. On the table below (and all others to follow), the order the seats are distributed is indicated by (1) for the first, (2) for the second, and so on.

Cambridgeshire — 7 seats
Party	Total votes	Div 1	Div 2	Div 3
UKIP	10,954	10,954

Hull is constructed from the 3 current Hull seats (E, N, W and Hessle). All 3 have Labour MPs with comfortable majorities.

Hull — 3 seats
Party	Total votes	Div 1	Div 2	Div 3

Brighton is constructed from the 3 current ‘Brighton’ constituencies (Brighton Kemptown, Brighton Pavillion, Hove and Portslade). All three are held by Labour but are close Lab–Con fights, especially Hove, won by Celia Barlow by just 420 votes.

Brighton — 3 seats
Party	Total votes	Div 1	Div 2	Div 3
Green	14,905	14,905

Edinburgh below is simply an amalgamation of the 5 new Edinburgh constituencies (W, E, N, S, SW). Four are held by Labour, with one Lib Dem in Edinburgh West. Edinburgh South is very marginal Lab–LD.

Edinburgh — 5 seats
Party	Total votes	Div 1	Div 2	Div 3
SNP	22,517	22,517

Like Edinburgh, Bristol below is an amalgam of the four Bristol seats (E, NW, S, W). Three are held comfortably by Labour. One (W) was a Lib Dem gain in a three-way marginal two weeks ago.

Bristol — 4 seats
Party	Total votes	Div 1	Div 2	Div 3
Green	5,876	5,876

Every one of the 5 seats in Cornwall – Cornwall N, Cornwall SE, St Ives, Falmouth and Camborne, Truro and St Austell – is currently held by a Lib Dem, a couple quite narrowly.

Cornwall — 5 seats
Party	Total votes	Div 1	Div 2	Div 3
UKIP	12,683	12,683

London North East is another top-up constituency proposed by the Jenkins Report in 1998. It consists of: Barking, Dagenham, East Ham, Hornchurch, Ilford N and S, Poplar and Canning Town, Romford, Upminster, West Ham. In fact, a more accurate description of the super-constituency would be ‘Urban Essex’ (Poplar aside). There are 10 seats in all – 6 currently Labour, 4 Conservative.

London North East — 10 seats
Party	Total votes	Div 1	Div 2	Div 3	Div 4	Div 5
BNP	12,124	12,124
Respect	20,783	20,783

Of course, I could go on re-running these forever. Interestingly, even in mono-political areas like South Shields (13 seats, all Labour), introduction of even a limited form of Fair Voting like this brings marked change: 8 Lab, 3 LD, 2 Con is the count. Like all above, a better fit with how the region actually votes. Which, after all, is what an election is for, isn’t it?

(Credit: Cabalamat Phil for the table code).