Log in

No account? Create an account

Previous Entry | Next Entry

This is an assessment of all the main possible methods for electing national governments, written by a lecturer at the University of Reading with the explicit aim of making "the findings of research on electoral systems available to a wider audience ahead of the referendum in the UK planned for May 2011." A fellow Yes campaigner recommended it to me at the beginning of this year as a balanced, rigorous and accessible guide to the main strengths and weaknesses of both First Past the Post and the Alternative Vote, so that I'd be equipped to argue about both intelligently in the course of the referendum campaign - and it has certainly provided that extremely effectively. Not only that, but the timing of the publication means that the author wrote it in the full knowledge that a referendum on the issue was going to be taking place, so that he was able to draw on up-to-date material such as the outcome of the 2010 General Election and comment on up-to-date issues such as the planned constituency boundary changes. So although the book actually goes beyond simply FPTP vs. AV, it is particularly well-geared to the current debate between the two, and examines them with the specific circumstances of the referendum in mind.

The book begins by setting out what sort of criteria it is sensible to judge electoral systems by. Here, Renwick formulates a template consisting of issues such as rewarding popularity, achieving fair representation, enabling effective government, offering voters real choices, encouraging turnout, providing meaningful local representation and preventing corruption. He then goes on to apply these criteria successively to First Past The Post, the Alternative Vote, simple methods of Proportional Representation, Mixed electoral systems (i.e. any system where some MPs are elected by one method and some by another) and the Single Transferable Vote. Another two chapters consider factors beyond the voting system per se which also affect the fairness of the system, such as the ways in which candidates are selected by parties and systems of recall. And the final conclusion weighs all of the competing systems up together, points out that no electoral system is perfect, and fundamentally advises readers to consider the facts calmly and rationally in the run-up to the AV referendum on May 5th.

Throughout, Renwick strikes what was for me the perfect balance between being accessible and readable for non-experts, while at the same time tackling technical issues thoroughly and without trying to gloss over the more complicated details. His arguments are very much grounded in serious academic research - presumably conducted mainly in the course of writing a previous publication entitled The Politics of Electoral Reform, but here presented in a more accessible manner for the interested general reader (why hello there, impact agenda!). Further reading and complicated technical calculations are provided at the end of the book if you want to know more (which I generally did), but kept out of the main text in order to maintain a clear focus on the key issues.

Of the electoral systems covered, I've known that First Past The Post produces horribly distorted results ever since seeing John Cleese's classic video on the subject at some point in my teens. But Renwick brought home to me just how distorted it is. He reports (p. 27) that according to a BBC pre-election model (which I can't now track down online), if Labour, the Tories and the LibDems had each won 31% of the total national vote in 2010, Labour would have got 314 seats in parliament, the Tories 207 and the LibDems 100 - an effect produced by factors such as the unequal geographical spread of their voters, unequal sized constituencies, poor turnout and more. Whatever you may think of any of the three main parties, that clearly isn't fair.

Even worse (p. 28), roughly 1 in 10 elections under FPTP produce the 'wrong' winner - i.e. the party which wins the most votes nationwide does not secure the most seats in parliament. This happened in the UK in 1929, 1951 and February 1974, and also occurs in the same proportion across comparable countries elsewhere using the same system. As Renwick says, "On such an important issue, this is surely a worrying error rate." One hell of an understatement, I think. It quite shocked me to learn that the system was really that distorted - and it certainly gives the lie to the No to AV camp's claims that FPTP delivers decisive results and ensures a clear winner. Not in one in ten elections, it doesn't.

Unfortunately, the Alternative Vote alone cannot fix these problems with disproportionality, because it isn't a proportional system either. Indeed, as Renwick concludes, "the reform that we are being offered in the coming referendum is not very radical". But as he also shows, it is better (though not perfect) at identifying the Condorcet winner in each constituency. Understanding that was what really solidified my own support for AV, and enabled me to write this post all about it. It seems to me pretty important that each constituency should be electing an MP who has demonstrated a real mandate of support within their own seat - to say nothing of the beneficial knock-on effects of ensuring that which I covered in my previous post on the subject.

Meanwhile, as for proportionality at the national level, Renwick uses the results of this research into how voters might have used preferential voting in the last seven elections (now updated to include 2010) in order to investigate the issue. He uses something called the Gallagher Index to measure disproportionality in election results - which is quite complicated but basically compares the difference between vote share and seat share for each party, and then combines the results into a single overall figure encapsulating the degree of mismatch between the actual election result and a perfectly proportional result. He then produces a lovely graph on p. 68 comparing the degree of disproportionality under the two systems. And thankfully, I can share his findings with you without breaching his copyright, because a graph based directly on Renwick's research is also included on p. 17 of the IPPR report on AV, which is publicly available online and can be downloaded here:

To me, what that basically shows is that over time, the two systems are pretty much the same. Sometimes FPTP is more disproportionate; sometimes AV is. Well, at least that's assuming that polling data on how people might have used preferential voting has anything to do with how they actually would in real life - which, to be fair, is a pretty huge leap. Still, you have to work with what you've got, and my take on that is that AV isn't any worse than FPTP from the point of view of proportionality, but it does do a better job of identifying the majority consensus in individual constituencies. Ergo, it's worth having.

The other systems which Renwick investigates are Simple Proportional Representation, Mixed electoral systems and the Single Transferable Vote. All of these either are or can be proportional systems (depending on exactly how they're set up), which certainly makes them superior to FPTP in my book. But I can't say I'm very keen on the ones which dispense entirely with constituencies and / or remove the capacity of voters to vote on the basis of individual candidates rather than just parties. I have a German colleague who is used to a mixed system, whereby voters have two votes at election time: one for an individual local constituency MP, and one which is used to elect a second set of MPs allocated on a nationwide proportional basis. She thinks it is a great system for Germany - but she did also comment that Germany is a federal nation with much stronger local government than we have in the UK, so that if she had a local problem, she would go to her local town council (or equivalent) rather than her local MP. That's not the case in the UK, and I think that having two separate 'streams' of MPs, half of whom were busy with local constituency casework and half of whom were not, probably wouldn't work for us.

All in all, I think Renwick is right to say that the Single Transferable Vote is probably the best possible system for a country like ours. It's preferential, in that voters use numbers to rank candidates, but it differs from AV in that the constituencies are larger, with each one usually returning between three and five MPs (each requiring between 1/4 and 1/6 of the highest-ranking preferences in the constituency to win). The larger constituencies thus contain a more diverse set of voters, helping to overcome the distortions currently caused by geographical concentrations of party support, while the proportions of votes in each constituency translate more directly into a proportional set of local MPs. The nationwide result isn't perfectly proportional, but it strikes a good compromise between being much more reflective of the national will than FPTP or AV, while still making it difficult for extremist parties to win seats. It also maintains a constituency link for all MPs, and indeed arguably offers better local representation for individual voters, since there is a greater chance that at least one of the your local MPs will be from the party you consider most representative of your own views. Thus Tory voters can call on their local Tory MP, Labour voters the Labour one, etc.

No system is perfect, though, and Renwick also points out that STV can make MPs all too responsive to the will of their local constituents. This happens in particular if parties put up more candidates per constituency than there are winnable seats in it, meaning that candidates within the same party are competing against one another for the available seats, rather than just against representatives of other parties. This can be a good thing for voters, because it allows them to select the particular candidates they like from their favoured party, rather than only having options which a party selection system has presented to them. But it also makes all seats inherently marginal, and gives MPs a very strong motivation to make sure that they please their own contituents above all else - including party whips trying to get them to support unpopular measures. In Ireland, which uses STV, it's arguable that these sorts of factors recently caused MPs to avoid tackling their national debt for all too long, thus leading to the country requiring a bailout package. Clearly, that wasn't the only factor, since Ireland is far from the only country which has suffered from a banking crisis in recent years. But it's a sobering example of how that sort of problem can be exacerbated by an electoral system which mitigates against making decisions which are necessary in the long term but which will be unpopular in the short term.

Obviously, it would probably have been helpful if I'd got round to reviewing this book a little earlier than 10 days before the referendum, so that anyone interested in reading Renwick's views for themselves could have had time to buy their own copy and read it before putting their cross on the ballot paper. But thankfully, Renwick's views on AV specifically are readily available online in the form of this briefing paper produced for the Political Studies Association. I can recommend it very highly to anyone wanting a properly balanced account of the arguments. Like the book, his assessment of the various strengths and weaknesses of AV as compared to FPTP in the paper is balanced, nuanced and objective. But the fairly clear conclusion to me is that AV is a small but measurable improvement on FPTP - and therefore worth having.

Click here if you would like view this entry in light text on a dark background.


( 32 comments — Leave a comment )
Apr. 25th, 2011 10:15 pm (UTC)
I'm afraid I'm not too keen on STV -- I'm not convinced that "proportionality" is the be-all-and-end-all. I think strong and appropriate local representation is also important.

When constituencies return from a number of candidates on a party list system you have the drawback that you are voting for a party and no longer for a party. This encourages homogeneity in party members -- the party can decide that it will put forward its most obedient members in the constituencies where they can win. This is already done to a great extent. However, we can still have "characters" and "locally popular" MPs who are great at campaigning for their local area. Under the STV this is massively diluted as you're no longer voting for the MP but for the party. If you would like Labour MP A to get in but can't stand that bugger Labour MP B then there's no way to express that.

On the other hand if you vote for MPs by name then there's simply too much information. I can just about recall things about my current labour and libdem candidates (labour won). I can't recall the name of the tory candidate. If there were 15 candidates then I'd have not a chance and I am someone who cares a little about politics. For most people you would just end up voting in the order the party says and you're back to party list.

I don't know... is it really better than FPTP? I think it has many disadvantages.
Apr. 25th, 2011 10:26 pm (UTC)
STV has you voting by name, not by party. AMS has you voting by party.

See here for the example for Edinburgh Council, where we had a choice of multiple members of some of the parties, and could vote for the representative we wanted.

AMS (as she is done in Scotland) gives you a local representative (via FPTP) and then list representatives to top you up, giving you the best of both worlds (in my opinion). But the lists are party controlled, so most political wonks I know who dislike party control prefer STV, because then you can vote for the representatives you actually like.
Apr. 25th, 2011 10:28 pm (UTC)
Sorry you are correct yes. The same voting system has different names in this circumstance.
Apr. 25th, 2011 10:27 pm (UTC)
Also, you don't have to know about all of them. In my case I'd look at the Lib-Dems, Labour and SNP members on offer. I wouldn't look at the Conservative, UKIP, etc. to see if I wanted any of them.
Apr. 25th, 2011 10:28 pm (UTC)
It's still more information than I could deal with I'm afraid.
Apr. 25th, 2011 10:30 pm (UTC)
You say you don't want the party to control who gets put forward - but you also say you don't want to find out information on a few people and make an informed decision.

I'm left wondering what you do want, as your two claims seem, to me, to be contradictory.
Apr. 25th, 2011 10:32 pm (UTC)
I want a system with only a single representative for each party in each constituency. Sorry, I thought that was clear.

I don't believe an informed decision is an informed decision if there's more choice than I can process. In the case of a system when I'm picking between four MPs from each major party, that's over my cognitive threshold and I'm just choosing at random.
Apr. 25th, 2011 10:34 pm (UTC)
But it hasn't _reduced_ your choice - you've gone from having no choice to having no choice.

And at the same time it has raised the choice for other people who are willing to take on the cognitive load.

So you lose nothing, and others gain.
(no subject) - steer - Apr. 25th, 2011 10:37 pm (UTC) - Expand
(no subject) - andrewducker - Apr. 26th, 2011 06:37 am (UTC) - Expand
(no subject) - steer - Apr. 26th, 2011 08:43 am (UTC) - Expand
(no subject) - andrewducker - Apr. 26th, 2011 10:51 am (UTC) - Expand
(no subject) - steer - Apr. 26th, 2011 10:57 am (UTC) - Expand
(no subject) - andrewducker - Apr. 26th, 2011 11:00 am (UTC) - Expand
(no subject) - steer - Apr. 26th, 2011 11:29 am (UTC) - Expand
(no subject) - andrewducker - Apr. 26th, 2011 11:55 am (UTC) - Expand
(no subject) - steer - Apr. 26th, 2011 12:10 pm (UTC) - Expand
(no subject) - andrewducker - Apr. 26th, 2011 12:43 pm (UTC) - Expand
(no subject) - steer - Apr. 26th, 2011 01:03 pm (UTC) - Expand
(no subject) - andrewducker - Apr. 26th, 2011 01:12 pm (UTC) - Expand
(no subject) - steer - Apr. 26th, 2011 01:16 pm (UTC) - Expand
(no subject) - andrewducker - Apr. 26th, 2011 01:14 pm (UTC) - Expand
(no subject) - steer - Apr. 26th, 2011 01:15 pm (UTC) - Expand
(no subject) - andrewducker - Apr. 26th, 2011 01:23 pm (UTC) - Expand
Apr. 25th, 2011 10:35 pm (UTC)
As far as I understand it, the beauty of STV is that it allows you to vote on the basis of party or candidate (or gender, or age, or ethnic identity, or anything else you prefer), without that choice compromising any of your other preferences. E.g. you could choose to give all of your highest preferences to the women on the ballot paper, even if some of them weren't from your favourite party, without that actively harming the chances of your favourite party winning seats. This means that if you know you like Labour MP A but hate Labour MP B (as you say), you can express that without it hurting the Labour party as a whole nationwide. But if you don't feel you know enough about the individual candidates while still generally liking the Labour party, you can simply rank all the Labour candidates highest (presumably in a fairly random order).

I'm not aware of any version of STV which elects from party lists (i.e. doesn't let voters choose between individual candidates) - certainly, Renwick doesn't mention it. Surely that would defeat the whole object of allowing voters to choose on the basis of whatever criteria they see fit - which he mentions quite a lot?
Apr. 25th, 2011 10:40 pm (UTC)
As Andrew points out, I'm incorrect to refer to that system as STV, it's AMS. (I find the terminology confusing as they're the same VOTING but with named candidates or not).

Yes, the "rank all labour candidates highest in a random order" is what I fear it would come down to. In this case the voting system has become:
A) MPs from parties elected randomly -- but more likely
B) MPs from parties elected in order party likes (if STV variants allow party to advise a prefernece) in which case we're back to party list.

To me it's the SKY TV of voting systems -- you've got acres of "choice" of what to do but it's a rather meaningless choice because it's choice between indistinguishable sludge.
Apr. 26th, 2011 06:38 am (UTC)
STV and AMS are not the same voting system at all.

AMS has a single FPTP vote and a separate regional vote. In both cases a single X is placed in a box. STV has a single ranked vote, whereby you number your preferences in order.
Apr. 26th, 2011 08:47 am (UTC)
Hold on, it was *you* who corrected me and said the system I meant was called AMS. *sigh*
Apr. 26th, 2011 09:05 am (UTC)
No, I said that STV doesn't have you voting by party, it has you voting by person, and that AMS is the one that has you voting by party.

I have a post planned to explain STV and AMS, but I'm not posting it until after the AV referendum, because the waters are muddied enough.
(no subject) - steer - Apr. 26th, 2011 10:40 am (UTC) - Expand
Apr. 25th, 2011 10:36 pm (UTC)
Thank you for a calm and intelligent view of a subject which seems to be treated either with total apathy or with increasingly heated and unreliable assertions. Reading your comments, and the PSA paper to which you linked, has not changed my mind - I will be voting for AV (as I know you will be), though I do wonder if it is worth the effort to achieve such a small difference. It's really only worthwhile if it represents the first step towards a properly proportional system, and I agree that STV with multi-member constituencies is probably the best option - but I strongly suspect that is too far a step to ever be achievable.
Apr. 25th, 2011 10:44 pm (UTC)
You're welcome - but thank Dr. Renwick rather than me, whose ideas I'm just reporting, really. :-)

It shouldn't be that much effort to make the small step to AV, really - though I'm aware that the No campaign in particular is making it into a massive fuss, which (the way it's going) could have destabilising effects well beyond the issue itself.

As for STV - I would like it myself too, but having looked back at Renwick's discussion of it while writing up this review, and having seen the strength of entrenched feelings currently arising over the tiny issue of AV, I have to say that I agree with you, and am more doubtful now about whether it will ever be achieved here. Still, one thing I am sure of is that the chances of getting it will be massively reduced if we don't at least start by saying Yes to AV. :-)
( 32 comments — Leave a comment )

Latest Month

November 2018


Powered by LiveJournal.com
Designed by chasethestars