Monday, 8 February 2010

Fair votes are simple with STV

In this guest post, Denis Mollison continues Next Left's debate about electoral reform, arguing that an electoral reform referendum should be about the Single Transferable Vote (STV) not the Alternative Vote, as the Prime Minister has proposed. Mollison has authored an illustrative new constituency map showing how STV could work in practice, which has been taken up by the Liberal Democrats as they propose an STV referendum in a Commons amendment to the Constitutional Reform Bill.

In a discussion of voting systems on "Any Questions" this week, Brian Paddick said "If I have to choose between simplicity and fairness, I'll choose fairness every time".

There are plenty of arguments for STV on grounds of fairness. But it can also be claimed that STV is simpler than our present system (FPTP) in the aspects that really matter.

Some of the simplicities of STV are evident in the scheme tabled last month by the Liberal Democrats as an amendment to the Constitutional Reform bill. (I must declare an interest here, as the author of this scheme).

Hopefully this amendment will get debated on Tuesday, alongside the more publicised one from Gordon Brown proposing a referendum on AV.

To digress for a moment, here are three reasons not to be cheerful about the Alternative Vote, in response to the arguments made for AV on this blog last week.

Firstly, AV may be a small step in the right direction, as Chris Huhne described it, but because that small step largely benefits the LibDems, it makes any support they give it look self-serving, and thus detracts from their support for more radical reform.

Second, if it is passed in a referendum, that will surely delay more radical reform for at least another decade.

Finally, and even worse, it could lose in a referendum because, as James Graham asked in a Guardian Comment is Free commentary on Wednesday, who is going to campaign enthusiastically for AV?

Returning to the Lib Dems' STV amendment, this proposes a scheme based on Local Authority areas, mostly electing 4 or 5 MPs.

Overall, it gives 118 constituencies returning 512 MPs (20% fewer than at present).

Constituencies, such as Glasgow, Cumbria, Teesside, Croydon & Bromley, West Sussex, Oxfordshire, have simple familiar boundaries; and when necessary the scheme can be reviewed using the current year's electoral roll - in contrast to the present process of Boundary Commissions, which is so cumbersome that this year's new constituencies in England are based on the electoral roll of 10 years ago.

The key simplicity for the voter is that you can express your real preferences: there is minimal scope under STV for tactical voting.

And all votes are equally influential, because all seats are equally marginal.

(Tactical voting and wasted votes are not only faults of FPTP, they are a significant problem for proportional representation systems that use top-ups, such as the system used for the Scottish Parliament and the AV+ scheme recommended by the Jenkins

That 90% of voters will end up with an MP of their first preference party, compared with 48% under the present system in 2005, is not just good for voters but for democracy.

Conservatives would have MPs in deprived city areas, and Labour MPs in rural counties; both parties would have to have policies attractive to voters everywhere, rather than concentrating on the small minority of swing voters in marginal constituencies.

The common argument by supporters of FPTP, attempting to justify a party having 55% of the seats on 37% of the votes, is that a one-party government is desirable. Surely this puts the cart before the horse. If a party wants to govern on its own, it should have to set out a manifesto that attracts a majority of voters.

The Rowntree Trust's regular "State of the nation" opinion polls have consistently shown majority support for a voting system that would give parties seats in Parliament in proportion to their share of votes, the latest June 2009 by a majority of 63% to 22%.

So please, Mr Brown, let's have a referendum offering genuine proportional

If you really want to include the option of AV, it needs to be a 3-way referendum.

There are obvious reasons why that would be unsatisfactory, though you could at least please those who like self-referential jokes by using AV to count the referendum result.

I'll bet you a hundred pounds to ten that AV will come last, though it will get my second preference.

* Guest post from Professor Denis Mollison of Heriot-Watt University.


Sunder Katwala said...

Thanks to Denis for the post.

As I mentioned in an earlier post STV or not STV?, a number of comments in response to Neal Lawson and myself discussing electoral reform on Comment is Free, thought we overlooked the merits of STV.

In the subsequent discussion, CIF thread, on the 'are non-urban constituencies too large' point, I asked whether electoral reformers had produced an indicative map to give people a tangible way to assess that. So Denis' map offers a very full response to that question.

Sunder Katwala said...

As it happens, YouGov did carry out a poll on political reform for the Fabians (fieldwork July 1st-3rd 2009), reported in our summer Review, which did ask a multiple choice question of First-Past-the-Post, Alternative Vote and Proportional Representation [unspecified system of PR].

Which of these systems is your first preference?

Present system: FPTP 25%

Alternative Vote: 25%

Proportional Representation 34%

Don't know: 16%

And which is your second preference?

Present system: FPTP 24%

Alternative Vote: 37%

Proportional Representation 25%

Don't know: 14%

sanbikinoraion said...

Unfortunately, reporting 1st and 2nd preferences like that is not particularly useful; we have no way of knowing which 2nd prefs are related to which 1st prefs. Traditionally in preferential voting the 1st prefs and the winner (after all preferences have been allocated) are reported. Any chance of getting the winning totals?

Richard Gadsden said...

This seems to be the only place I can comment on the map.

I can only comment seriously on my own region of southern Lancashire (including Merseyside, Greater Manchester and the northern parts of Cheshire that used to be in Lancashire).

When combining local authorities, Prof. Mollinson has followed the boundaries of the former metropolitan counties rather more closely than is necessary. I suspect the same may apply in Yorkshire; the division between West and South Yorkshire is not one that lives in people's hearts in the way that the division between Yorkshire and Lancashire does.

Perhaps the most striking problems are these:
"Sefton, Knowsley and St. Helens". This is basically "Liverpool Outer"; there's very little connection between outer suburbs like Bootle, Halewood and Kirby and separate towns like St. Helens and Southport.
"Liverpool and Wirral" Oh, look, the Mersey Tunnel seat. Not a good idea; very much opposed on both sides of the Mersey.
The Cheshire seats - Warrington and Halton have far more in common with each other than with the rural parts of Cheshire; the same applies in reverse.
The GM seats aren't bad; I'll probably have to reconstruct them to fix Merseyside and Cheshire, but they can work.

OK, an alternative:
Liverpool: includes Knowsley and the Bootle part of Sefton. 538,916 - 6 MPs
W Lancs: Mollinson's SW Lancs plus Southport/Crosby from Sefton - 526,857 - 6 MPs
[rename SE Lancs to E Lancs; SE is Manchester, not Blackburn]
St. Helens, Wigan, Halton and Warrington "South Lancs". 621,194 - 7 MPs
Wirral/West Cheshire 493,691 - 5 MPs
{alternatively, move Halton from South Lancs to West Cheshire and make both constituencies 6 MPs; or we could split Halton along the Mersey between the two, which would match the traditional boundary)
East Cheshire/Stockport 506,443 - 6 MPs
Bolton/Bury 343,092 - 4 MPs
Oldham, Rochdale and Tameside 484,198 - 5 MPs
and retain the Manchester (4) and Salford & Trafford (4) seats.

Denis said...

Richard, thanks very much.

I'll be delighted to receive other suggestions ref. the map.

While many of the constituencies seemed (given the rules of the scheme) the only natural choice (e.g. Cornwall, Oxfordshire, Edinburgh), others were more problematic, and I'll welcome suggestions for revision by those with better local knowledge.

Denis Mollison
(only one "n" in each name, please)

Anthony Z said...

While West and East Sussex are the historic counties, the demographic and cultural differences are between South and North Sussex, with South Sussex being Brighton and the urban coastal area from Littlehampton through Brighton and Eastbourne to Hastings, and North Sussex being the rural areas and commuter towns, plus Gatwick and Crawley.

I'd make the two constituencies Sussex Coast (Arun, Adur, Worthing, Brighton & Hove, Eastbourne and Hastings) and Sussex Rural (Chichester, Mid Sussex, Crawley, Lewes, Wealden, Rother).