Some of the British papers are giving publicity to an analysis by the Electoral Reform Society purporting to show that the Alternative Vote system (which Labour is offering to the Lib Dems now, and the Tories are offering a referendum on) would have made little difference. Specifically, the claim is that actual result of 307/258/57/28 (C/L/LD/Oth) would translate as 281/262/79/28 (with STV giving you 246/207/162/35). But this is a completely static analysis, since it presupposes that matters like candidate selection would stay the same under both systems. This is surely wrong: under AV, the main parties would have an incentive to select candidates who would appeal as the second choice of the eliminated parties. This would often mean a convergence to the centre (on the assumption that the Lib Dems stay third, which they might not) but it might mean the selection of Tory candidates who would get the votes of eliminated UKIPers.
{ 58 comments }
jay bee 05.11.10 at 10:46 am
I don’t see why they describe it as “minimal” difference when both AV and STV give the LibDems the numbers to make a clear majority with both Labour and the Conservatives. Seems like quite a significant difference to what they’ve got now?
alex 05.11.10 at 11:01 am
I’d love to know how they guessed what the unexpressed second preferences of all those fringe-party voters that would be the first to be eliminated were.
Gareth Rees 05.11.10 at 11:15 am
I’d love to know how they guessed
They didn’t guess, they polled. Do read the Electoral Reform Society’s explanation of their simulated results where they acknowledge the obvious problems with the simulation.
“In looking at AV and STV we have assumed that votes cast on 6 May would have been ‘first preferences’. That, of course, is not the case […]
“AV and STV use preference voting. In our models we have used the second preference data of a ComRes poll of 26 April 2010 (fieldwork on 24/25 April 2010). This data, however, has a number of shortcomings […]”
Farah Mendlesohn 05.11.10 at 11:41 am
It also assumes that the parties will stay the same, but there are plenty of Lib Dems who would peel off to the Greens if they thought they had a chance, and enough isolationist Tories to boost UKIP. AV might well lead to a redistribution of people in the voting booth across four or five parties and change the face of those parties dramatically. I’d expect the cores to remain, but the fuzzy edges to reconfigure.
alex 05.11.10 at 11:41 am
Ah, they guessed.
alex 05.11.10 at 11:42 am
Blimey, Farah, fancy seeing you here.
chris y 05.11.10 at 12:04 pm
You never know, there might even be a splinter group which would represent the interests of the organised working class.
ajay 05.11.10 at 12:06 pm
there are plenty of Lib Dems who would peel off to the Greens if they thought they had a chance, and enough isolationist Tories to boost UKIP
Wouldn’t matter. The point about AV is that if no one gets a majority of first preferences, the votes of the last-placed candidate are redistributed according to second preferences. If still no one has a majority, the same happens to the second-last candidate and so on up.
So it doesn’t really matter if there are (for example) lots of Lib Dems who put Green as their second preference, because it’s unlikely that Lib Dem votes will be redistributed while the Greens are still in play (because the Lib Dems will probably get more first preferences, most of the time). What’s more important is whether there are lots of Greens who would put Lib Dem as their second preference.
ajay 05.11.10 at 12:09 pm
Unless I’ve misunderstood you and you mean there are lots of current Lib Dems who would vote Green for first place if they thought they had a chance.
This is also not a factor: AV isn’t a proportional system, it works at existing constituency level only. So your assessment of whether the Greens have a chance isn’t likely to change under an AV system.
Neil 05.11.10 at 12:13 pm
jay bee, I think they maintain this isn’t a significant difference because in the long run, we’re still all dead.
Ian Whitchurch 05.11.10 at 12:23 pm
307/258/57/28
Hung parliament. Any of the big 3 win, assuming factional discipline
281/262/79/28
Hung parliament. Any of the big 3 win, assuming factional discipline\
246/207/162/35
Hung parliament. Any of the big 3 win, assuming factional discipline
alex 05.11.10 at 12:31 pm
“…the organised working class.”
If only there were such a thing, imagine what it might do…
Ban immigration, for a start, I, alas, suspect…
Gareth Rees 05.11.10 at 12:37 pm
Unless I’ve misunderstood you and you mean there are lots of current Lib Dems who would vote Green for first place if they thought they had a chance.
Why do you add that last caveat? Surely under single-member-constituency AV or STV you no longer need to take into account which of the candidates “have a chance”?
Ian Whitchurch 05.11.10 at 12:44 pm
Gareth,
Yes, you do. Australia has had STV for some time, and it’s a real consideration whether a certain candidate can come third or second after preferences.
This has allegedly led to major parties “running dead” so their voters preferences can flow through to the possible-to-win Independant, thus f.cking over the other major party (cf Cleary, Phil).
Pete 05.11.10 at 12:59 pm
“the organised working class”
British Airways cabin crew?
Gareth Rees 05.11.10 at 1:05 pm
Right, but that must be a much rarer kind of dilemma than in FPTP. At the moment almost every Green supporter in the UK is in a dilemma: do they vote their real preference, or do they vote tactically? In AV, the dilemma would apply mainly (only?) in constituencies with Conservatives likely to be ahead on first preference votes, with Greens and another party close to a tie for second. That dilemma is not what I understood by ajay’s caveat in #9.
Ben Alpers 05.11.10 at 1:24 pm
Semi-OT: Interesting how quickly things seem to have fallen apart between the Lib Dems and Labour. The hostility of Labour backbenchers to voting reform is apparently the biggest stumbling block. But there’s also reportedly an unwillingness to give the Lib Dems what they want on civil liberties.
Zephyrus 05.11.10 at 1:26 pm
You never know, there might even be a splinter group which would represent the interests of the organised working class.
You might even get a couple dozen purporting to!
Ian Whitchurch 05.11.10 at 2:13 pm
Ben,
It will be a lot easier to be the opposition to a LDP-supported government than to be a LDP-supported government, and this is a lot easier to see as a backbencher than as a minister with a chauffeur and so on.
That said, Labour should pass STV in order to dish the Tories …
Tom 05.11.10 at 2:21 pm
It surprises me that it’s so rarely understood, including by the Electoral Reform Society, that AV supports the two-party system, just like FPTP.
http://rangevoting.org/TarrIrv.html
(There AV is called IRV)
Ben Alpers 05.11.10 at 2:28 pm
@Tom: Isn’t IRV what is being called “AV+” in the British political discussion? “AV,” as used in British politics, only allows voters to rank two candidates.
Current 05.11.10 at 2:32 pm
AV is a way of systematizing tactical voting.
STV with multiple member constituencies is the best for of “real PR” as far as I can see. It seems to work well in Ireland.
Tim Wilkinson 05.11.10 at 3:02 pm
I should have thought that only a national party list system would be “real PR” in the most plausible sense of that term, given that the premise of PR is that you are essentially voting for a party. If you have constituencies, there is always going to be the possibility of wasted majorities.
Perhaps not a very big probability though. Unsurprisingly, I haven’t come across any simple quantitative assessments of the relative merits of voting systems: i.e. ones which look at how likely various systems are to violate the various voting criteria. Arrow style impossibility results, based on the bare possibility of failure under not necessarily very plausible conditions aren’t very useful really, at least given that AFAICT no system avoids all such failures.
The use of constituencies is also I suppose based on some idea of geographical (or other, e.g. Lebanon?) group representation at work, too: local preferences not to be drowned out by national ones sort of thing.
Ben: Isn’t IRV what is being called “AV+†in the British political discussion? I had the impression that ‘AV+’ incorporates a list-based top-up element.
Tom 05.11.10 at 3:05 pm
@Ben Alpers:
No, AV+ has regional party lists too:
http://en.wikipedia.org/wiki/Alternative_Vote_Top-up
Ben Alpers 05.11.10 at 3:07 pm
@ Tim and Tom: I stand corrected.
chris 05.11.10 at 3:16 pm
The example in Tom’s link involves a small third party of… radical centrists? And then goes on to demonstrate that the smaller of the two major parties (the one with 35% first-preference support) can tactically vote to throw the election to the compromise third party, which is implicitly claimed to be bad for some unspecified reason. (Funny, I thought that the ability for parties that can’t win by themselves to give effective support to someone near them in political space without completely closing up shop and endorsing another party’s candidate was supposed to be a *feature* of non-FPTP systems.)
If you assume that small parties tend to be at the edges of the electorate’s political space rather than at the center (which seems reasonable, since the two behemoths are generally wrestling over the median voter to begin with; and if you don’t like the implicit unidimensionality in that concept, Tom’s linked example is just as bad in that regard, with two preference orders left off the table because *obviously* nobody would put the centrist last in their honest preferences), then that example won’t work. Eliminating the small parties early will move their votes to the major-party candidate closest to them; the result is still that a major-party candidate is elected (it’s unavoidable in single-member constituencies that someone close to the political center of gravity will be elected) but the small parties still show up in the election results because their supporters aren’t forced to tactically vote for a major party.
There are several other examples linked from that page, but I think they all have the “radical centrist splinter” problem or similarly wildly unrealistic voter preferences ex ante. And all they manage to prove is that tactical voting is *sometimes* useful under IRV (although they can’t seriously dispute that the situations where it arises are much less common than FPTP, and especially single-member-district-based FPTP with regional political variations, which is what the UK actually has).
So it seems like “don’t support that thing that is better than the status quo, because *this* thing (that has no chance of actually passing, but never mind that) is EVEN BETTER!”, which is of dubious wisdom even if true.
chris 05.11.10 at 3:27 pm
Unsurprisingly, I haven’t come across any simple quantitative assessments of the relative merits of voting systems: i.e. ones which look at how likely various systems are to violate the various voting criteria.
Doesn’t this require complex and dubious assumptions about the preference-state of the electorate and how the parties position themselves on the political landscape? Some configurations are impossible in practice (a majority of sozialists* choosing red-baiters as their second choice, tiny parties in the center, parties that are a long way from any voter, etc.) but ISTM that modeling all of those constraints would be rather difficult and you could never be completely sure that you had done it accurately.
*Misspelling intentional because of that thing with the spam filter.
James Conran 05.11.10 at 4:04 pm
“there’s also reportedly an unwillingness [on the part of Labour backbenchers] to give the Lib Dems what they want on civil liberties.”
That would be a little surprising given that 90 day detention provoked what I think was the biggest backbench rebellion of the Blair years:
http://www.guardian.co.uk/politics/2005/nov/09/uksecurity.terrorism
thomas 05.11.10 at 4:14 pm
@Ian Whitchurch, @Gareth:
STV leaves the parties with plenty of room for tactical manoeuvering on which seats to fight, but it makes it relatively unusual for the voter to have an incentive to vote other than genuine preference. In order to do useful tactical voting you not only need a close result, you need to know quite a lot about other people’s intended votes, right the way down the preference list.
Tim Wilkinson 05.11.10 at 4:17 pm
chris @26: something like that, yes; hence ‘unsurprisingly’. But it must be possible to make some progress in that direction, at least specifying some boundary conditions etc (and no doubt the experts have done so) – in fact your #26 is along those lines innit.
James Conran @28 – I was a bit aghast to hear GB mention sodding terrorism in (IIRC) his opening offer speech all those days ago. Depressingly this probably means Labour intend to play that grubby little card in attacking any LibCon alliance.
James Conran 05.11.10 at 4:17 pm
To illustrate how tactical voting could occur under AV, imagine a seat with only the big 3 running, with their support as below:
Tory 40%
Labour 30%
Lib Dem 30%
Then assume 90% of Labour voters give their no. 2 to the Lib Dems whereas Lib Dem voters go 60/40 Labour/Tory.
In this case who won the seat would depend on who get eliminated first – if Labour comes last the Lib Dems take the seat, if the Lib Dems come last the Tories do.
But it’s all a bit complicated (and indeed speculative) so I wonder whether many voters are sophisticated/informed enough to actually vote tactically under AV.
Tim Wilkinson 05.11.10 at 4:21 pm
JC @30 – voters would have help though. I understand parties often give quite detailed instructions – to the point of providing a specimen ballot paper – under AV regimes.
skidmarx 05.11.10 at 4:33 pm
Tory MP Daniel Kawczynski was on the TV this afternoon saying that AV would force him to appeal to BNP supporters for second preferences.
Brett Dunbar 05.11.10 at 4:38 pm
@30
The table under your scenario
40 Con
27 Lab LD
3 Lab Con
20 LD Lab
10 LD Con
The condorcet (compare each possible pair and see if one candidate can beat all others head to head) comparisons are:
43 Con
57 LD
40 Con
60 Lab
30 Lab
30 LD
If the conservative voters have an expressed preference then there is a condorcet winner which is invisible to AV. The Conservatives are condorcet losers as if facing any single opponent they lose.
James Conran 05.11.10 at 4:52 pm
@ Skidmarx,
Politicians should already be appealing to BNP voters for “1st preferences”. That would be the way to defeat the BNP. Admittedly AV might cause politicians to do this with a slightly altered rhetorical strategy. In any case I suspect BNP transfers would go all over the place, with a large proportion probably being non-transferable (i.e. no 2nd preference, or going to already-eliminated candidates).
Stuart 05.11.10 at 4:59 pm
Tory MP Daniel Kawczynski was on the TV this afternoon saying that AV would force him to appeal to BNP supporters for second preferences.
That would seem a strange strategy, in the vast majority of seats aiming for second preference of right of centre lib dem voters and non-left wing green voters (and even centrist ex-blairite new labour supporters) would seem a better idea. Trying to get second preference votes from the extreme wings of the political spectrum is rarely a good idea in such voting setups as it alienates more than it attracts.
alex 05.11.10 at 5:19 pm
As 34 notes, implicitly [‘Condorcet’], since the C18 it has been understood that complex voting situations can produce counter-intuitive results. But counter-intuitive results are better than systematically unfair ones.
Martin Bento 05.11.10 at 5:54 pm
BBC says it looks like Tory/Lib. It doesn’t mention civil liberties issues, which is something I would care about much, but don’t imagine the Tories would be better on than Labour, though possibly no worse.
Stuart 05.11.10 at 6:06 pm
I guess as any bill put forward in the area of civil liberties the Lib Dems will be able to help vote down if they feel necessary, as long as the Conservatives don’t decide to make it a confidence issue and dare the Lib Dems to pull down the government. Of course this might make the Conservatives rush to try and implement anything that might be controversial early on, as there might be a strong negative reaction in the public if the Lib Dems pull down the government quite soon, whereas after a year or two (if it lasts that long) it might not have as much impact.
scathew 05.11.10 at 6:15 pm
Man, I thought our American system was unfathomable…
alex 05.11.10 at 6:24 pm
GB resigns! First!!
Martin Bento 05.11.10 at 6:52 pm
I don’t understand British politics well, but doesn’t it seem like the Lib Dems are shooting themselves here? With serious cuts coming, Tories will, I assume, sacrifice not just the poor but the middle-class too to shore up the rich. That doesn’t look to me like a government that will retain popularity for long, and it has a weak mandate now. Is it certain that electoral reform would pass in a referendum anyway? If I were Lib Dem, I would play nice till I got my electoral reform, then look for a popular excuse to bring down the government. Ultimately, the LD’s have to look to be replacing one of the majors, right? There is a core Tory constituency that I do not imagine is up for grabs, save perhaps by the far right. Core constituency on Labour’s side has been much betrayed and may be gettable. But maybe I’m not seeing clearly because of my distance from the situation.
Lemuel Pitkin 05.11.10 at 7:02 pm
Am I correct in understanding that what happened here is that Labour MPs rejected the idea of a “rainbow coalition” and/or electoral reform as the price for it? And that therefore the LDs went back to the Conservatives with less leverage than they had less week? And does this mean that even the weak-tea AV referendum offer is presumably no longer operative?
Philip 05.11.10 at 7:34 pm
Lemuel, my understanding is that labour weren’t able to make a deal with the Lib Dems to take back to the PLP. It looks like the Lib Dems will go into coalition with the Tories though this hasn’t been confirmed. If there is a Con/Lib coalition the Conservatives have offered a referendum on AV.
Ben Alpers 05.11.10 at 7:35 pm
@ Lemuel Pitkin:
I think we don’t know what deal the Lib Dems ultimately made with the Tories. I’m not even sure if we know whether it will be a coalition or a deal for Lib Dems to support a minority Tory government on budgetary votes and the Queen’s Speech.
My sense of what happened with the Labour negotiations is that Labour was effectively unable to budge on anything due to the threat of a backbench revolt (and they needed every single vote). As one former Lib Dem MP I heard interviewed put it, Labour was only able to offer their manifesto (and in the case of AV, perhaps not even that), while the Tories gave real concessions.
matsig 05.11.10 at 7:57 pm
Just out of curiosity: How come the alternative electoral systems most commonly discussed in the UK seem to be AV, STV or “real PR”? Wouldn’t the least drastic, but still significant, reform be from FPTP to a modified version of the French system (“real majoritarian elections in single-member constituencies”). Modified since I’ve always found the 12% threshold to participate in the second round to be rather low, but put it at 25-30% and you will still have the geographical link from FPTP, but give far more chances for third parties and geographically strong fourth and fifth parties. Is there a reason why I haven’t even seen this suggested?
Gareth Rees 05.11.10 at 9:30 pm
We’re discussing single-member AV/STV because that’s apparently what the new government is going to hold a referendum on.
Other electoral systems that currently exist in the UK include the proportional additional member system of the Scottish Parliament, the parallel additional member system of the Welsh Assembly, the multi-member STV of the Northern Ireland Assembly and the closed party list system of the European Parliament. But none of these appears to be on offer for the House of Commons.
matsig 05.11.10 at 9:50 pm
Hmmm… On the one hand, that is of course a sufficient explanation. On the other hand, it also increases my confusion. If there are so many additional systems already ready at hand in the UK, how come they are NOT on offer (or even being suggested) for the House of Commons? Mind you, I’m not interested in a normative discussion about what would be the ideal electoral system, but rather the reasons as to why it is AV/STV that is up for referendum, whereas a system that is closer to the existing one (and thus logically more palatable to Labour and Tory alike) and (at least as far as my limited understanding allows me to determine) no less beneficial to the LibDems is not.
djr 05.11.10 at 10:08 pm
matsig@46 – AV is “real majoritarian elections in single-member constituencies”. The French system is essentially a less-good variation on AV (i.e. AV where only each voter’s first preference and relative ranking of the candidates who come first and second on first preferences is taken into account). The disadvantages of this system over AV were demonstrated in the 2002 French presidential election. The French system also requires two rounds of voting, which would be a more significant change to the UK system than having the option to number candidates on the ballot paper rather than just using a single X.
matsig 05.11.10 at 10:26 pm
Ok, I will grant that AV is (or could be) “real majoritarian elections in single-member constituenciesâ€. I will also grant that there are problems with the French system, primarily if people don’t bother turning up for the first round (being from a “real PR”-country myself, that is what I see as the “natural” system, not FPTP, AV or STV). However, AV, or for that matter STV, is far more complicated for the voters. The French system, for all its faults, is the easiest way to get a somewhat proportional effect while maintaining single-member constituencies. Vote once – If there is a REAL winner (i.e. with a majority of the vote behind him/her), you’re done. If no-one gets a majority, you go to second round. I fail to see how that would make a bigger difference to the existing FPTP than any system that requires voters to rank candidates by order of preference, regardless the benefits of such a system from a normative point of view.
And just to be clear about my normative standpoint, I consider any electoral system that has voters choosing between individual representatives who make decisions for them to be inferior to one that has voters choosing between competing political programmes (i.e. parties), it’s just that I’m aware that I am to some extent socialised into the latter system out of habit. Consequently, I find the British negligence of an electoral system that seems to me, as an outside observer, to be far more culturally acceptable, surprising, to say the least.
Gareth Rees 05.11.10 at 10:30 pm
how come they are NOT on offer (or even being suggested) for the House of Commons?
Who knows what was suggested?
The two major parties stand to lose seats as a result of electoral reform, so they won’t do it unless they have to, and then only in the most grudging way. AV is about the smallest possible step in the direction of proportionality, so that’s what’s on offer.
Current 05.11.10 at 10:54 pm
One possibility that folks haven’t discussed is retaining FPTP but going back to multi-member constituencies. For example, each ~5 constituencies could be merged together and a FPTP rule applied where the candidates with the top 5 shares of the vote won. It wouldn’t be PR, but it would allow small parties a better shot.
Modicum 05.11.10 at 11:28 pm
@Current:
The system you’re describing is called the Single Non-Transferable Vote. It has been used in Japan, Taiwan and some other places.
Actually it is capable of achieving roughly proportional representation. (For the results to be proportional both the parties and voters have to think act tactically). But I’ve read at least one political scientist who argues that SNTV should be categorised as a form of PR.
So I think it would be a good system to propose in the UK or the U.S. It is very similar to the current system and very simple to explain to voters.
g 05.11.10 at 11:51 pm
What advantage would multi-member FPTP have over the other options that have been considered? It seems to me like every objection that gets applied to multi-member STV would apply to it too, as would every political reason for the Red and Blue parties to dislike it; but it would be less proportional and extract less information about voter preferences than multi-member STV would.
I’m probably missing something. Current, what are the advantages of multi-member FPTP?
Current 05.12.10 at 12:24 am
Well, I think multi-member STV is better. But, I think multi-member FPTP would be an easier sell to Tories. It was used in Britain in the 19th century.
Brett Dunbar 05.12.10 at 2:09 am
@47
Both Scotland and Wales use a proportional additional member system. The Welsh system uses a smaller top up so is less proportional. 20 top up AMs with 40 constituency AMs as opposed to 56 top up MSPs with 73 constituency MSPs.
Alex 05.12.10 at 2:39 am
Actually, AV would’ve made a difference. It would’ve led to a near Tory wipeout in 1997:
http://news.bbc.co.uk/1/hi/uk_politics/8506306.stm
Harald Korneliussen 05.13.10 at 9:54 pm
One Ka-Ping Yee, currently working for Google.org, made some nice images and a flash application to illustrate the problems with AV/IRV (really it compares several different single-winner systems, but IRV is the one that jumps out at you). It’s a nice visual explanation of the problems ranted at by the rangevoting.org article.
Also, chris, “radical splinter centrist” parties isn’t so unlikely as you think. Here in Norway the centrist vote is split between three small parties.
Comments on this entry are closed.