Talk:Single transferable vote/Archive 1

Archive 1

Archive 2

Archive 3

→

Archive 5

Question of capitalisation

SIngle Transferable Vote is ALWAYS capitalised and treated as a proper noun. Similarly, when linked with proportional representation and spoken of as 'Proportional Representation through a Single Transferable Vote, the term is used as a proper noun. JTD 20:25 Feb 12, 2003 (UTC)

I'm sceptical, it's not a proper noun in the usual meaning of the term. It's similar to party-list proportional representation and similar articles, which are not capitalised according to Wikipedia naming conventions. ( 23:59 Mar 22, 2003 (UTC)

I agree. A Google search also indicates that this term is often not capitalized and our rule is that unless a term is either a proper noun or always capitalized then the term should be in the lower case. I'll move the article myself. --mav

It is capitalised when referring to the system. It is not capitalised when referring to the methodology. It is appreviated as PR.STV never pr.stv. (If that is done in an essay by a student of political science, they invariably get a right bollocking and are told to learn the basics of political science.)

A google search is absolutely useless because you cannot tell in which context it is being used in each article on it. (Not that google searches are a particularly reliable source of anything. It doesn't rely on trustworthy sources but any sources, from the best available to sheer garbage. All it takes is for a lot of sources to feed some wrong fact off each other and hey presto we have the Gospel of Google. Google suggests that we should have an article on William Gladstone. But do a search of proper academic sources and you find he should be (as every student of British politics and history knows) William E Gladstone or William Ewart Gladstone, never William Gladstone. When that sort of standard was applied, and not merely a google search, the page was renamed with a full consensus.

Another example: According to a google search:

Charles Windsor: 393,000
Charles Mountbatten-Windsor: 352

Yet his name is not Charles Windsor and 393,000 references are wrong. His name is Charles Mountbatten-Windsor - sources: his own office, Buckingham Palace press office, banns posted at his marriage. So on this and so many other issues, google is complete and utter crap.

So please Mav, stop relying on google. It is in many many many cases absolutely unreliable rubbish. STÓD/ÉÍRE 00:27 Mar 23, 2003 (UTC)

Thus I said "indicated." We are also not talking about a fact here but usage by English speakers. Therefore Google is a useful external tool to consider that doesn't rely on our own POV. The fact does still remain that the term in question is not a proper noun and is not always capitalized (and this article should cover both the system and methodology). I wasn't the one who moved the article BTW; paren beat me to it. ---mav

If it is describing the system, it generally is. If it is describing merely the methodology, it isn't. The article on wiki is describing the system known and referred to in political science textbooks as Single Transferable Vote (STV). Or when used with proportional representation, it is called PT-STV or PR.STV. Only where textbooks describe a vote outcome through a single transferable vote is it not capitalised. Usage depends on context, and in this context, it is generally capitalised. In fact failure to capitalise it in an exam context where it was being used as in the article would see student docked marks because in political science faculties it would be seen as simply and factually wrong, as wrong as writing the 'princess of Wales' when you are talking about a specific Princess of Wales and not all princesses of Wales. I know I sound a bit petantic here, but I know if university students read a page about the specific voting system written as 'single transferable vote' they are not going to take wiki as a serious sourcebook, thinking that if wiki can't even get the name of the system correctly, how on earth can they trust the article's contents?

Simply look at the opening line of the article. It talks about STV being a voting system, so it is explicitly and unambiguously talking about the system, not merely a description of a methodology. It is patently absurd, grammatically ludicrous and amateurish in the extreme to get something as basic as a voting system's name wrong. I know my professor would be roaring with laughter at seeing an encyclopædia getting the name of an internationally used electoral system wrong. I certainly hopes he never looks at that page with its almighty clanger in the title. STÓD/ÉÍRE 01:35 Mar 23, 2003 (UTC)

Then can we at least agree to disagree and move on? This was a borderline case to begin with due to the majority capitalization usage so it isn't something worth arguing much about - I have more important things to do with my time like adding content to Wikipedia. Do what you think is best for this article. --mav

I have treble-checked in political science textbooks, websites and election coverage books. The vast majority capitalise Single Transferable Vote. Indeed wikipedia stands out in having lower case for some references to STV. I've put together a list of just some of the titles, using the exact words used by them.

• BBC|Proportional Representation|Single Transferable Vote
• Model Election by Single Transferable Vote
• STV - Single Transferable Vote
• Preference Voting - Single Transferable Vote
• Axioms and STV (Single Transferable Vote)
• How to conduct an election by the Single Transferable Vote
• Proportional Representation and the Single Transferable Vote
• The Single Transferable Vote
• Yahoo Groups : single-transferable-vote The Single Transferable Vote method
• Single Transferable Vote hand count

Constitution Debate: Single Transferable Vote and electoral . . .

As the evidence is so utterly overwhelming, with wikipedia standing out on the page on the net as one of a small minority not to capitalise a page dealing with the voting system (as opposed to methodology, where lower is sometimes used) I am changing the title back to the capitalised version. STÓD/ÉÍRE 02:26 Mar 23, 2003 (UTC)

So is this another American English versus British English issue? Martin

Those links don't indicate so, some are British and some not. But could someone please hand Stod some valium? It's really not that important, man. Chill out. :) --AW

---

(to Martin) The problem with wiki articles on voting procedure is that, when a number of articles were placed by a group of political scientists (one Aussie, two Brits, one Irish and a couple of Americans) put on a series of pages on voting systems a handful of people swept in, and without knowing what they were doing, merged totally different systems into the one article thinking they were the same, broke up single articles into bits that contradicted each other, mucked up names, changes references, mucked up formulæ and then just when it looked that things couldn't get each other, a particular user who specialises in capitalisation muck ups (ie changing capitalisation in topics he doesn't know the first thing about because some dodgy google search based on a number of nutty websites told him too) swept in and lowercased proper nouns because ludicrously inaccurate web pages told him to.

As a result there are glaring inaccuracies in terminology, in voting methodology, patently absurd links. One page makes Instant Run-Off sound the same as PR.STV. The PR.STV page links to STV which is ludicrous; PR.STV uses STV, but so do other systems which have nothing to do with PR. And many systems of PR don't use STV at all. That was once clear but the edits and rewrites and redirects, and capitalisation changes and chop-ups now mean that everything is now blurred into one almightly heap of garbage that is utterly unreliable. But having pleaded with those responsible to stop and been met with "but google says . . ." bullshit none of those to did the original work are going to touch the pages. FearÉIREANN 03:33 10 Jul 2003 (UTC)

Moving this here, because it seems like potentially useful criticism of this page (and related pages), though not knowing the history myself, I can't really say. Delete if pointless. Martin 00:12, 2 Aug 2003 (UTC)~

Inclusion of CPO-STV

How can we include CPO-STV? Perhaps we should separate this entry into SSTV and STV, which would also link to CPO-STV. Any other ideas? -- ShadowDragon 06:02, 13 Aug 2003 (UTC)

Linking to it will be easy, getting it written is where the work lies. I'd like to see a CPO-STV page. I've seen descriptions of it (google will find one from the EM list) but I've never discovered what the "CPO" stands for. Pm67nz 10:24, 5 Feb 2004 (UTC)

CPO = Comparison of Pairs of Outcomes -- the whole thing is usually called something like "Comparison of Pairs of Outcomes by the Single Transferable Vote" -- ShadowDragon 23:56, 20 Apr 2004 (UTC)

Non-proportional results

This seems to be Henrygb's personal space and no one is allowed to contribute even fully referenced material without him accusing them of baing POV, whereas Henry's blatant POV and illogical reasoning is prominent. I give up.--Alun 15:14, 9 Apr 2005 (UTC)

"It can often give non-proportional results in close elections such as the 1981 election in Malta where the Maltese Labour Party won a majority of seats despite the Nationalist Party winning a majority of votes. This caused a constitutional crisis, leading to provision for the possibilty of bonus seats to ensure proportionality, as proved necessary in 1987 and 1996. Similarly, the Northern Ireland elections in 1998 led to the Ulster Unionists winning more seats than the SDLP, despite winning a smaller share of the vote."

This is not a STV problem? It's a PR in small constituency problem. --Braunbaer 11:50, 8 Dec 2003 (UTC)

It is an STV problem if you want to describe STV as PR. One essential test of whether a voting system is PR is whether a party which wins more votes than another also wins more (or at least the same) number of seats. In Malta this failed three times, not because the constituencies were particularly small (how large is not small?) but because there are constituencies, combined with a tight two-party system: I accept this could happen with other regional PR-type systems, but the fact is it happened regularly with STV. It can happen whenever there are more than two constituencies; perhaps only national systems should count as PR. In the Northern Ireland elections it happened not because the constituencies were particularly small but because one party attracted transfers which the other did not, and that is directly an STV effect. STV is not PR, though it tends to produce more proportional results than many other non-PR systems. --Henrygb 15:08, 17 May 2004 (UTC)

You seem to be suffering from 'absolutist' reasoning. There's really no such thing as true proportionality. I mean it's imposible as the number of representatives elected is an integer (you can't have half an MP, for example in a 100 seat chamber a party having 3.5% of the vote it will only ever have 3 or 4 representatives, neither of which numbers represents proportionality). The fact that constituencies will never have perfect equality of representation (for example one MP per 20 000 of the population would again lead to 2.5 MPs in a constituency of 50 000). In truth there can only be greater or lesser deviation from proportionality. To get the closest possible to proportionality one would need to have large numbers of representatives (say one per 10 000 electors) elected with a single very large constituency (say the whole country). It's also erroneous to only consider first preference votes when considering deviation from proportionality. This reasoning misses the whole point of STV, which is that electors know that their first preference may not get elected, but also know that their vote will usually count. By only taking into account first preferences when discussing deviation from proportinality the point is missed that a voter may feel equally as strongly for their first and second preference. The other thing to note is that the party system is not particularly important in STV. It is irrelevant that say the MLP won more seats in the Maltese election but less first preference votes because STV is designed so people are electing representatives not parties. In the STV system it is notable that elected representatives find it is beneficial for job security to represent their electors over their party, there are no 'safe' seats as there are in SMP (FPTP). Politicians from the same party are also in competition with each other for the same seats, so this again decreases party power. I am of a mind to modify this section considerably as it makes some highly spurious assumptions. I removed the sentence about ranking all candidates ensuring no wasted vote in Australia. Surely this can't be the case, is one expected to rank what must be at least 30-40 candidates? And your vote is effectively wasted if for example none of your top 10 candidates are elected, especially if your top candidates are all obscure and your later candidates are popular and already elected. --Alun 07:41, 3 Apr 2005 (UTC)

Obviously you cannot get exact proportionality unless you elect fractional numbers of candidates. But you can point to significant failures of proportionality: achieving monotonicity (parties which win more votes win at least as many seats) should be an easy criterion for a proportional voting system. If you look at actual Maltese vote counts (look at the links I gave below on 23 February), you cannot say that the three STV proportionality failures in Malta were because people "voted for candidates not parties", because almost everyone voted for all the candidates from one party and none from any other - when there were more than ten candidates from a single party (for five seats) then they would not see their eleventh vote as wasted if it meant their favoured party won a third seat. STV theoreticians only see non-transfered votes and votes on non-elected candidates in the final round as wasted, and so on this definition having to vote a full list does indeed minimise the number of wasted votes to be less than a quota. --Henrygb 13:05, 3 Apr 2005 (UTC)

I don't quite follow what you mean by 'parties which win more votes win at least as many seats'. This means that a party will get the same number of seats as votes. Do you mean 'parties which win a greater proportion of the vote should win at least that proportion of the seats?' What does the at least mean? In PR it should be exactly, not at least. And I don't follow the logic in only counting first preference votes. You are absolutely right when you say (PR-)STV is not necessarily proportional (and was not devised to be necessarily proportional). But the whole concept of proportionality seems to me to be irrelevant when it comes to (PR-)STV. How can votes which have been sub-divided several times and used to elect two or three people be 'put back together again' and only the first preference counted, in order to prove it was not proportional? A person acheiving a high level of first preference votes (but not enough to get elected) and then not enough other preference votes to get elected surely lacks support from the wider electorate, and only receives support from a loyal minority. The big advantage in (PR-)STV is the strength of the relationship it allows between the elected representative and the electorate, and the fact that it considerably reduces party power. People are truly electing the representatives of their choice, it is erroneous to only look at the votes a party gets and complain if the proportion of representatives doesn't match the proportion of the party's first preference vote (people are not voting for a party but individuals), the party should choose candidates with a wider voter appeal in that case. PR-STV in Ireland, has lead to about a 3.4% deviation from proportionality since 1948 [1]. That's really quite good by anyones reckoning.--Alun 17:34, 3 Apr 2005 (UTC)

When I said parties which win more votes win at least as many seats I meant that if Party A gets more votes than Party B then Party A should win at least as many seats as Party B in a system claiming to be proportional; if it fails this test in practice, voters may complain that the system is unfair and non-proportional. Examples of non-proportional systems which controversially failed this test were the electoral college in the U.S. presidential election, 2000, and the UK general election, 1974 (February). My claim is that in practice STV often fails this test, and the article has a few examples; the Maltese examples are particularly good because "wider appeal" beyond a party's loyal supporters does not seem to happen in Malta. Given that in most STV elections most voters seem to choose the order of parties to vote for and the rank of candidates within each party, it is legitimate to ask if STV is a proportional system. My answer is that it is semi-proportional, as is SNTV, but not as proportional as party-list proportional representation which can also have close voter/candidate contact with an open list, especially on the Finnish model.--Henrygb 20:59, 3 Apr 2005 (UTC)

But you're talking only about first preferences. First preferences are not the only preferences in STV. —Christiaan 21:14, 3 Apr 2005 (UTC)

Believe it or not, I do know that. That is why Malta is such an instructive example, since almost all voters cast preferences for all the candidates from the party they support and for none of the candidates from the other parties. But if you are saying that there is some kind of hidden proportionality shown in late preferences and transfers to other parties (the opposite of the point Wobble/Alun was making about 11th preferences in Australia) then you hit the theoretical point that someone may be elected in a district with a small number of seats but not with a large number of seats, raising the question of what their real proportion of support is. If you want to claim STV is fully proportional then you have to define what you mean by proportional; saying that it is what STV produces is a little circular. --Henrygb 21:46, 3 Apr 2005 (UTC)

But of course Henrygb no one is claiming that STV is fully proportional (see my comment above about absolutist reasoning). There are only deviations from proportionality. I think we all accept this point. This also addresses the point about failing the test of one party getting less seats than another while getting an equal share of the vote, deviations from proportionality do occur in all systems (even party list systems). No system can ever be fully proportional. To generate better proportionality at least three things need to be addresses above and beyond the system of voting and vote counting. Firstly the level of representation needs to be high. That is; the less people needed to elect a representative then the more proportional the result. If there is one elected representative per 10 000 people then the result will be more proportional than if there is one representative elected for every 50 000 people. Secondly representation needs to be equal, with the number of people required to elect one representative being relatively equal on an inter-constituency basis (see the Australian example in the main article). The third requirement to increase proportionality is large constituencies. One constituency of 60 000 electors electing 6 representatives will always produce better proportionality than two constituencies of 30 000 people each electing 3 representatives. These factors are all true even in party list systems. Party list systems also have other areas where they deviate from proportionality. Here in Finland the re-distribution of excess or reminder votes favours the larger parties while descriminating against the smaller ones (the d'Hondt system). This system produces deviations from proportionality which can also lead to two parties with equal vote share getting different numbers of seats, and in this system it is the parties which get the vote and not the candidates. If proportionality is ones sole criterion for a fair election then one can have a single huge constituency (the whole country), with a party list system, with seats distributed to the parties in as exact a proportion to their vote as possible. Of course then there are no local constituency representatives, just a huge mish-mash of people all representing the whole electoral area. In list systems the party has so much power that it detracts significantly from the relationship a representative has with their constituency anyway. No system is without it's flaws and certainly PR-STV can deviate from proportionality, personally I believe that the other benefits it brings far outweigh the small deviations (as I say above 3.4% in Ireland since 1948) in proportionality which can occur. Open list systems are in my oppinion a good second best solution to democracy, but PR-STV, when optimised for proportionality is the best.--Alun 05:48, 4 Apr 2005 (UTC)

You are showing some POV here - don't let it go into your edits. The article states some of the claimed advantages of STV (person before party, few wasted votes etc.) but to have the article saying that this outweighs occaisonal failures to reflect popular will would be going too far. I accept the Australia point - indeed I put it into the main article [2]. You seem to accept that there are electoral systems which do ensure that a more popular party gets more seats than a less popular party through a national list (e.g. Israel); I would be interested to see an example of a party winning a majority of seats in a national or regional party-list PR system which did not come first in the popular vote. You accept STV can deviate from proportionality, so I think it is reasonable for the article to say so; this has not been a universal view among editors. --Henrygb 08:50, 4 Apr 2005 (UTC)

You really seem to be missing the point in a big way Henry. We're not talking absolutes here, it's all relative. There are no "occassional failures" of STV, there are only deviations as there is in any system, and as Alun points out the least are to found in PR-STV. —Christiaan 09:00, 4 Apr 2005 (UTC)

Henry, when it comes to electoral systems it's always a question of POV when it comes to determining which is the best. What I'm saying is that PR-STV displays good proportionality when coupled with other factors which enhance proportionality. All things being equal PR-STV gives generally good proportionality (3.4% deviation) accross all preferences. It also allows the electorate to hold individual representatives accountable in a way which list systems can't. If your only criterion for good democracy is PR based on a party's share of the vote, then there are systems which can generate better apparent proportionality. These can, however lead to elected representatives being less accountable to the electorate. PR-STV therefore couples good general proportionality while keeping election of representatives in the hands of the voters (which is much weaker in list systems). This significantly strengthens the link between electors and elected. It keeps representatives aware that their primary responsibility is to their constituency and not their party. It is your prerogative to believe that as close as possible to proportionality on a party basis is the most democratic form of electing representatives (this is also a POV). I'm just saying that PR-STV cannot be discounted as not proportional just because it doesn't acheive perfect proportionality every time (no system does), or because other systems might produce better proportionality on a count purely of the party vote. Most PR systems are designed to be generally proportional. How much deviation from proportionality is acceptable is a different question and of course will always be POV. Your real position seems to be that certain applications of STV have not produced enough proportionality for your liking. If this is not POV I don't know what is. We all accept that PR-STV can deviate from proportionality Henry, but we also all accept that all electoral systems deviate from proportionality. If you want to point out that PR-STV can deviate from proportionalit then it's only fair to make the same point for all proportional electoral systems. If you want pure proportionality then you're back to the half a representative arguement again. I maintain that to talk of proportionality when a vote has been sub-divided several times and may have been used to elect several different people from different parties is erroneous. Theoretically a person's vote could have contributed to the election of all representatives in a constituency, who might all be from different parties, and their first preference still might not have got elected. You are fixated by first preferences and an absolutist view of proportionality. You seem to be obsessed with proving that PR-STV is not proportional, when everyone is telling you that it has a proportionality at least as good as most other PR systems. Understand that there are no perfectly proportional systems, and that we all agree with you. Also understand that PR-STV is as proportional (in that proportionality can be understood in such a system) as any other system.

Here's the results of the Finnish open list elections held in 2003 [3]. As you can see the major parties benefit from a higher share of parliamentary seats than they should get in a fully proportional system. The smaller parties are discriminated against, due to the d'Hondt method of re-allocating surplus votes. There is good proportionality, like most PR systems offer, but none are fully proportional.--Alun 17:45, 4 Apr 2005 (UTC)

So at least Finland had the most popular party winning the most seats - an interesting example given the reversal between the Swedish People's Party and the Christian Democrats - possibly affected by Åland. Here is another more recent example using Finland, the European Parliament election, 2004: in Finland's results with party lists the two biggest parties got the same number of seats; in the Republic of Ireland's results with STV the most popular party won fewer seats than the second most popular party. This seems to be becoming a habit of STV. --Henrygb 21:40, 4 Apr 2005 (UTC)

Getting the most first preferences does not equate to "most popular party". Why do you insist on this? —Christiaan 21:44, 4 Apr 2005 (UTC)

It does in Malta, and I would assume elsewhere. Assuming of course that 1st preferences are given to voters' prefered parties. Can you give me a definition of popularity or proportional which can actually be measured without looking at 1st preferences? Look at this (continuing my example from the Euroelections) --Henrygb 22:09, 4 Apr 2005 (UTC)

Finland	% vote	% seats	difference
National Rally (KOK)	23.7	28.6	4.9
Finnish Centre Party (KESK)	23.3	28.6	5.3
Finnish Social Democratic Party	21.1	21.4	0.3
Greens Independent (VIHR)	10.4	7.1	-3.3
Left Alliance (VAS)	9.1	7.1	-2.0
Swedish People's Party (SFP)	5.7	7.1	1.4
Others	6.5	0.0	-6.5
Ireland	% vote	% seats	difference
Fianna Fail	29.5	36.4	6.9
Fine Gael	27.8	45.5	17.7
Sinn Fein	11.1	9.1	-2.0
Labour	10.6	9.1	-1.5
Greens	4.3	0.0	-4.3
Independents	16.8	18.2	1.4

The entire rational behind a preference-based voting system such as STV is that summing 1st preference votes is not mathematically nor statistically valid as a means of measuring the popularity of a party. If it was, there would be no reason to use anything other than a single, non-preference system such as First past the post. Malta is generally accepted to be an anomaly that demonstrates one of STV's corner cases, and which demonstrates that STV is the weakest of the preference-voting systems. (See Condorcet method for a stronger preference system.) If one asserts that 1st preference votes on ballots validly determines the most popular party, then one is not using a preference voting system.  — Saxifrage | ☎ 02:07, Apr 5, 2005 (UTC)

I think there are four points which need to be made here Henry.

1) If one want's to find out if STV produces proportionality then counting only first preferences is wrong. I would suggest something like counting the number of first preference, second preference and third preference votes for each candidate in a three member constituency, the number of top four preferences for a four member constituency etc. If one wanted to do even more one could rank the preferences thus; in a four member constituency 4 points for first preference, three for second etc. In a five member constituency give 5, 4 ,3 points etc. Compare these results with the election results. (By the way is this actually an electoral system (giving 5 to the first candidate, 4 to the next etc)? Of course one only gets as many preferences as there are seats available. Could it be varied to give 15 votes per voter (in a five member constituency, based on 5 for top preference, four for second etc) but allowing the votes to be split in any way the voter likes, so all 15 votes could go to one candidate if she was one's one and only choice?)).

The 5,4,3,2,1 version is called the Borda count; allowing you to shift the numbers around is cumulative voting --Henrygb 09:15, 5 Apr 2005 (UTC)

2) You are trying to second guess voting intentions. You are assuming that people are voting on a party basis. In doing this you are in actual fact saying that the voter does not understand the electoral system. If one votes purely by party then one is voting as if it were a list system. In all electoral systems it is essential to have some sort of understanding of how your vote will be counted, in order to be fully enfranchised. It is not a fault of the system if it is being applied incorrectly. By saying it does in Malta you are in effect saying that you know the minds of all Maltese. Also don't forget that you are making an assumption (Assuming of course that 1st preferences are given to voters' prefered parties). So your analysis is contingent upon that assumption being correct. Many might give first preferences to prefered candidates (of whatever party) and later preferences to prefered parties, it's what I'd do.

Actually I am looking at the revealed preferences of Maltese voters. Although they are free to do so under STV, almost none of them vote for candidates of another party before exhausting all the candidates of their own party; almost none of them fail to vote for all candidates from their prefered party; almost none of them vote for candidates from more than one party. So if you were in Malta, you would be unusual. --Henrygb 09:15, 5 Apr 2005 (UTC)

Which just goes to proove my point that the fault is not with the electoral system but in it's application. The best way to fix this is for the electoral authorities to have an education campaign aimed at educating voters on how to get the best results from the system. All voting systems need the electorate to fully understand how their votes are apportioned. Alternatively a change to an open list system might be better in a country where party loyalty is so strong. But remember that counting only first perferences is not the same as counting proportionality. Your list from the Irish election is erroneous if the proportions you have given for percentage of votes cast only includes first preferences, as it ignores all transfered votes, which is what acctually makes it proportional (as Saxifrage says above). I think that by voting exclusively for one party, a voter is far more likely to see a larger proportion of her vote wasted. So you have not shown that STV is not proportional by counting only first preference votes. As I say I really can't see how it is valid to stitch sub-divided votes back together and claim that the intention of the voter was to exclusively vote for their first preference, you are still trying to second guess the voters intentions by doing this. Also almost none of them vote for candidates of another party before exhausting all the candidates of their own party, votes are still being transfered intra party, and how do you know this is true. The results might imply that a larger proportion of voters than you realise actually are voting for more than one party. Do you have any actual figures as to what percentages of people vote exclusively for one party and what proportion distribute their votes between parties? Maybe voters are being more sophisticated than you realise.--Alun 09:48, 5 Apr 2005 (UTC)

OK I will define my terms, which can be calculated from the record of counts:

1. People are shown to fail to exhaust a party's candidates if candidates for that party remain in the count, but surplus transfers for elected candidates for that party or votes for eliminated candidates from that party do not go to the remaining candidates for that party, whether they either go to another party or they are not transfered)
2. People are shown to vote for more than one party if there are transfers from the candidates of one party (surpluses or after elimination) to candidates from another party, whether or not there are remaining candidates from the first party.

My claim is that in Malta these numbers are both very low, indicating that these events are relatively rare; as a consequence parties often have many more candidates than there are seats. Now I have provided a definition which can be checked, would you care to define proportionality in a way which can be checked? --Henrygb 18:12, 5 Apr 2005 (UTC)

I think I have found a way to show you are wrong, Henry, but forgive me if my reasoning is incorrect. I think it's down to the transfer of reminder votes.

(a) In your table for the Irish elections the differences for elected representatives and first preference votes are the differences between transfers, not between what's proportional and what's not. In your reasoning all votes above and beyond the threshold are not counted. The equivalent reasoning in an open party list system would be to count a candidates list votes and compare them to the election result as a whole to see if it was a fair election (ie not counting the vote as a party vote. I think this is annaloguos to SMP (FPTP)).

(b) STV can be made to give the same result as a closed list system. If voters for a specific party all only choose only that party and all put their preferences in the same order then it is exactly the same as a closed list (given equal threshholds and reminder calculation methods). Like this: Once the first preference has been elected (the person at the top of the list) all transfers then go to the same second preference or the equivalent of the second person on the list etc. It should be possible to model whether this is true relatively easily, but it seems self evident to me.

(c) If voters don't put the party candidates in the same order (still assuming that only one party is being voted for) on the ballot then many intra-part transfers will take place, possibly denying a candidate from the party the votes required to be elected in a subsequent round (as the votes have been distributed throughout the party's candidates. This may be the case in your Malt example, especially if the party is putting up many more candidates than there are seats to be won, as the transfers are spread more thinly). It may also be true that even a relatively small proportion of people mixing parties in their ballot (say a couple of percent) could lead to a deviation from the closed list analogy. Again I guess this could be modelled by producing an STV election which resembles a closed list, then introduce an alternative where 2% of voters mix their preferences, one where 5% mix their preferences and one where 10% do.

(d) The only real differences between a list system and STV are: (i) The re-allocation of reminder votes, list systems assume the intention of the voter is to vote for the party, and so the party retains the reminder. STV gives the voter the privelege of transfering her reminder to a different candidate (in the same or a different party). (ii) In a list system if one votes for an obscure party which never acheives a threshold, then one's vote is lost, in STV the second preference is counted as a full vote for the candidate. I guess this could be rectified in the list system by allowing a second preference party.

(e) I think you can model the proportionality of the Maltese election like this.

(i) Count all first preference votes for all parties in each constituency, all second preference votes, all third etc. Produce a total number of votes for each preference per party per constituency.

(ii) Work out your thresholds for election in the same way as for the Maltese election.

(iii) Procede to elect representatives as if it were a closed list system, ie when a representative gets elected from the list work out the reminder. Now you divide the reminder by the total number of votes from the first round and multiply that fraction by the total second preference votes for the party in the election and continue to do this for each round.

(iv) Compare your result to the actual Maltese election. I think it might give a closer approximation to proportionality than just counting first preferences. This is the best I could come up with.

If these observations are correct then STV should be at least as proportional as a list system as votes are transfered in both, it's just that in STV the decision as to where the votes are transfered is left to the voter, but retained by the party in list systems. There may be faults in my reasoning, apologies if there are. Anyway it's always good to be made to think about these things, thanks, Henry.--Alun 06:04, 6 Apr 2005 (UTC)

What you have written is pretty confusing, and seems to ignore many of the point I have made before (see my comment of 17 May 2004 above). (a) On the Euroelections, it is clear that something odd happened in the Republic of Ireland which did not happen in Finland, enough to say that STV is often less proportional. Whether this was constituencies, transfers, differential turnout or what I don't know, but there was a big difference. (b) What you say is true (assuming the same way of calculating the quota) but only within a single constituency; but STV tends to have smaller constituencies than party list systems (which can be nationwide) if only to keep filling in the ballot paper credible - as a result it tends to produce less proportional results overall and throws up more curiosities. (c) I think you are wrong - so long as all voters for a party exhaust that party's candidates before voting for some other party or not transfering, it will make little difference to the STV result if they use different orders within the party. (d) What you say is true, but in party list systems smaller constituencies are less necessary and less common than with STV making the result more proportional. (e) Because of the way individual Maltese voters actually vote, there is usually no practical difference to the party results whether each five seat constituency has STV or party list with a suitable method (highest averages with d'Hondt or largest remainder with Droop); but you definition is pretty circular - you seem to be saying let's base a definition of proportionality on whether it corresponds to an STV result - but you would laugh if I suggested that "party list PR in 1 seat constituencies" would produce the same result as first-past the post, showing the latter as a proportional system. The point I have tried to make throughout this is that in practice STV has produced results which many observers would not describe as proportional at the national level, and the article needs to explain this. If you still don't accept this then you might think about whether your positive view of STV is affecting what you write. The set of voting systems topics is full of people pushing their pet system (the Condorcet fans have been the most biased in my view) and using their limited personal experience (usually their country and its neighbour) to inform their statements: for example you were wrong about Australia on 3 April - instead they allow you to let a party choose your full preference for you if you prefer (a sort of party list STV) instead filling in a long list yourself; see preference voting. I haven't edited the article recently, discussing the topic here. But I will now switch method, and edit the article with summaries rather than failing to persuade you here. --Henrygb 18:30, 6 Apr 2005 (UTC)

In point (a) in the Irish elections I am right if the results you have given are for first preferences only. (b) you are agreeing that it's not an STV propbem but a small constituency problem? Point (c) can be tested easily by running a model election. In conclusion I do not claim that we define proportionality as STV. I claim that it does not make it less proportional just because the voter decides where her excess or reminder goes rather than the party. I think you are the one with the bias against STV and are ignoring the facts, using irrelevant data (like first preferences) to draw erroneous conclusions. We appear to be going around in circles as you do not appear to be interested in the facts, just in being contrary. Why not try modelling my suggestions. I still think STV works, but I a agree that large constituencies are required. In a small place like Malta for example one or two, electing a parliament of say 50 people would suffice.--Alun 05:33, 7 Apr 2005 (UTC)

3) I'm not a statistician, but it seems to me that specific cases do not disprove general trends. I would guess that outlier results occur in all electoral systems, it doesn't disprove proportionality.

4) Why do you assume that the party with the most votes should always get the most seats? All PR systems deviate from proportionality. In the Finnish Election above the two largest parties should have had an equal share of the seats, but in actual fact Keskusta gained two more seats than the SDP, this can be seen as a 400% deviation from proportionality (0.2% of the vote for 1% of the seats). Variation in proportionality does occur, but why should it be discriminatory in favour of the party with the largest share of the vote and against the smaller parties. It's not correct to say that proportionalality is the same as the largest party always getting the most seats (proportionally, the two largest parties in 2003 in Finland should have had the same number of seats), as deviations will always occur (in all systems). In a close election it might hinge on a fraction of a percentage point and whether the fraction is rounded up or down.

I don't claim that more votes leads to at least as many sets is the sole requirement for proportionality, but I do think it is a reasonable expectation for public acceptance of a PR system (the criticism of the US electoral college in 2000 was based on a similar expectation, and that was designed to be disproportional). The frequent failure of STV to deliver this needs to be noted if you want to argue STV is a form of PR. STV has many good things about it; guaranteeing a proportional result is not one of them. --Henrygb 09:15, 5 Apr 2005 (UTC)

But it won't always happen in any truly proportional system due to rounding. In Finland in 2003 Keskusta should have had 49 seats and SDP 49 seats in a fully proportional system, even though Keskusta won more votes. It's the unproportional d'Hondt system which has given them a majority of seats. You should be arguing that no system is proportional if your acceptance of a deviation from proportionality is so strict. Why, for example is it OK for Keskusta to get significantly more seats than is proportional just because they got fractionally more of the vote? Is it not equally unproportional for other parties to be denied their rightful seats, just so the larger parties can get an unproportionally better representation.--Alun 10:01, 5 Apr 2005 (UTC)

Thanks Henry, you have made me think about STV a lot recently and I have to confess that the more I think about it the more I like it. By the way Kokomus is usually refered to as the National Coalition Party [4]. --Alun 08:28, 5 Apr 2005 (UTC)

Feel free to edit the link I gave to European Parliament election, 2004 (Finland)

STV in British Columbia

Although STV was not used in 1949 in British Columbia, it was used in 1952 and 1953: [5] and [6]. 142.179.125.20 02:22, 3 Mar 2004 (UTC) (aka Vancouverguy)

---

STV was never used in British Columbia. In '52 and '53, BC used the Alternative Vote, aka Instant Runoff Voting, which gave voters as many ballots as candidates to be elected (most districts were single-member, but some were multi-member), effectively dividing each district into several single-member districts and allowing residents to vote in all of them. This system is not proportional, as a majority bloc vote can control all the seats in a district, instead of only receiving their proper share, as in STV, which elects multiple members from only a single ballot. Reference

Un-eliminated candiates

I removed this sentence:

A vote is wasted if it ends up on the last candidate to be eliminated.

This is not true. There can be more than one un-eliminated candiate at the end of the voting process, as long as their total votes are less than the quota.

Ben Arnold 02:23, 20 Apr 2004 (UTC)

STV in Northern Ireland

The STV is used for Assembly elections, and European Elections (differing from the rest of the UK) - I beleive STV is also used for local elections. What about the Westminister elections - these don't use STV - do they? I don't know.

STV is such fun to vote with :o) You get such a satisfaction of putting the unlikely people first :o) My own local area had over twice as many candidates as seats - I greatly amused myself weighing up party alignment, and as a final resort, job descriptions!!!

Zoney 23:30, 12 Jun 2004 (UTC)

There are also tactical consideration for parties standing more than one candidate in the election. Standing too many candidates might result in first-preference votes being spread amongst them, and several being eliminated before any are elected and their second-preference votes distributed. Standing too few may result in all the candidates being elected in the early stages, and votes being transferred to candidates of other parties.

But surely once the candidates are eliminated, their second preferences would mostly flow to the other candidates of the SAME party? Thus so long as the combined party vote is good enough, they'll get members elected. I feel like I'm missing something here. Dlw22 16:49, 22 Aug 2004 (UTC)

First past the post is used for Westminster elections in North of Ireland. This is because it is part of the UK system and it uniformly uses first past the post (plurality system).(Frank Crummey)

The reason why it can be risky to run too many candidates in a constituency is that you can spread the votes too thinly causing your candidates to be eliminated before any of them get elected. You must remember that surpluses are also distributed. The ideal situation is to have two candidates hovering just below the quota after the first count allowing htme to pick up transfers from other parties and independents before finally electing both candidates. this does not always happen as planned. It is a fine art form that has been perfected by many parties in the Republic of Ireland. Particularly Fianna Fail in the last general election.(Frank Crummey)

Potential for tactical voting

"However, in older STV systems there is a loophole: candidates who have already been elected do not receive any more votes, so there is incentive to avoid voting for your top-ranked candidate until after they have already been elected. For example, a voter might make a tactical decision to rank their top-place candidate beneath a candidate they know will lose (perhaps a fictional candidate). If the voter's true top-place candidate has not been elected by the time their fake top candidate loses, the voter's full vote will count for their true top-place candidate. Otherwise, the voter will have avoided having had their ballot in the lottery to be "wasted" on their top-ranked candidate, and will continue on to lower-ranked candidates."

...in more modern STV systems, this loophole has been fixed: just out of curiosity, does anybody know which of the Republic of Ireland, Malta, the Australian Senate, Northern Ireland, use this more modern system? My guess is none of them, in which case it is theory (except for some minor private elections) and the tactical voting issue remains. --Henrygb 11:50, 17 Sep 2004 (UTC)

It's been fixed in Ireland. Votes beyond a candidate's required quota are also transferred. The random selection of such votes (from the total votes for that candidate) adds a non-deterministic outcome though. zoney ♣ talk 12:02, 17 Sep 2004 (UTC)

It's also been fixed in New Zealand, where a deterministic version of STV is used for District Health Board and some local government elections. The system relies on a "keep value" which is the fraction of each vote that a candidate needs to keep in order to remain over the quota. If your first choice keeps 60% of your vote at any point in the count, then there is 40% of your vote left for the candidates further down your ranking. A candidate that has not reached the quota keeps 100% of their vote, and a candidate that has been eliminated keeps 0%. If you put a candidate that is likely to be eliminated towards the top of your ranking, they have no effect on how much your vote counts toward remaining candidates. Your vote will be subject to the same keep value as that of those people who didn't try to tactically vote.Ben Arnold 12:38, 17 Sep 2004 (UTC)

Thanks for that, but neither of those comments quite state what the article says. The article implies that if you vote A 1st, B 2nd, C 3rd and the count goes B elected and surplus transfered (your vote stays with A), then in the next round A is eliminated and votes transfered, then only part of your vote will go to C, rather than all of it. Taking the example of the Northern Ireland European Elections 2004 BBC report, somebody who voted Gilliland 1; Allister 2; Nicholson 3; would have had a whole vote certain to be transfered directly to Nicholson at stage 3 under traditional STV, but the article sugests that under a modern system only part of a vote (or a fractional probability) would apply, since it would have to pass via Allister and votes via Allister are scaled down. --Henrygb 21:10, 17 Sep 2004 (UTC)

I agree with your summary of what the article says. The article is correct with respect to New Zealand. I'm not sure if it's true for Northern Ireland. Here's a worked example using NZ STV with 4 candidates going for 2 seats. 300 votes cast, quota of 100. My vote is Alice, Brian, Chris.

First round

• Alice: 20 votes (keep value 100% because she's not elected and not eliminated)
• Brian: 200 votes (keep value 100%)
• Chris: 50 votes (keep value 100%)
• David: 30 votes (keep value 100%)
• at this stage my vote is for 1.0 vote for Alice, even though I really support Brian and Chris, I think Brian is a shoe-in, and I'm trying to vote tactically

Second round (Brian is elected)

• Alice: 50 votes (keep value 100%)
• Brian: 100 votes (keep value 50%)
• Chris: 70 votes (keep value 100%)
• David: 80 votes (keep value 100%)
• my vote is still 1.0 vote for Alice

Third round (Alice is eliminated)

• Alice: 0 votes (keep value 0%)
• Brian: 120 votes (keep value 50%, 40 of Alice's second preferences went to Brian, but he only keeps half)
• Chris: 90 votes (keep value 100%)
• David: 90 votes (keep value 100%)
• my vote is 0.0 vote for Alice, 0.5 vote for Brian, 0.5 vote for Chris

Fourth round (Brian's keep value is adjusted down)

• Alice: 0 votes (keep value 0%)
• Brian: 100 votes (keep value 42.67%)
• Chris: 95 votes (keep value 100%)
• David: 105 votes (keep value 100%)
• my vote is 0.0 vote for Alice, 0.4267 vote for Brian, 0.5733 vote for Chris

My tactical voting didn't work, because some of my vote still goes to Brian, even though he'd already been elected by the time my vote was transferred from Alice.

Ben Arnold 08:32, 19 Sep 2004 (UTC)

What about this tactical voting opportunity? I want Candidate A and Candidate B to get elected. Of the two, I prefer Candidate A. However, I believe that Candidate A will exceed the quota by far on the first count, and I also believe that most of the voters who give their first preferences to Candidate A will give their second preferences to candidates other than Candidate B. Therefore, I give my first preference to Candidate B such that she or he gets my full vote instead of a fraction thereof. I've actually done this before and provided Candidate B's margin of victory (and Candidate A, as I expected, was easily elected). Has it been corrected in any extant STV method? (Unsigned 18:20, 16 Jun 2005 User:66.7.37.54)

No, and it cannot be, if you are certain your preferred candidate will be elected and so do not express a preference for that individual. --Henrygb 01:14, 17 Jun 2005 (UTC)

Software?

STV seems to contemplate the system of paper ballots filled out by the voter, allowing him or her to rank preferences manually. The voting machine systems employed by most U.S. jurisidictions contemplates solely a first past the post system. Should I assume that all or nearly all STV is done where paper ballots are still used? Occasional media coverage of the use of STV in Cambridge, (and, it seems to me, a few other places like Berkley, California, I'm not sure that the article is right when it says that STV is used in the U.S. solely in Cambridge currently) always seems to emphasize how complicated it all is, how slow to canvass, etc., the subtext always seeming to be "Why can't they just vote like the rest of us, instead of doing something like this? Why do they think that they're so much smarter than the rest of us?" Perhaps because the U.S. media are for the most part so notoriously lazy, they seem to have a hidden or perhaps even a somewhat open agenda which seems dedicated to preventing the spread of STV or indeed any other system other than first past the post, which is relatively easily explained and easily covered from the "horse race" aspect, which is what most U.S. media seem to obsess about, "Who's winnng and by how much?" as opposed to "What are the real issues and how will they be addressed?"

I've said most of that to ask, I guess, is there a good software system that will allow this to be done at popular prices? (American politicians don't put much stock in spending lost of money on elections; after all, the current system was good enough to put them into place, so it can't be that flawed!) If it has to be done everywhere the way that the media has portrayed it as being done in Cambridge, then the odds of it ever being done anywhere else in the U.S. to any great extent are almost none. Rlquall 16:20, 13 Nov 2004 (UTC)

http://vote.dsu.org.uk

Electronic voting using STV has been trialled in a real election in Ireland (in four constituencies). Purpose built units were used (as opposed to a PC with voting software), and the HMI interface consisted of a touchscreen (laid out similarly to the ordinary Irish ballot paper). One ticked off the choices in order (using the touchscreen), and the machine then displayed what would be submitted as your vote (offering a confirm or re-input option).

As regards the calculation of results, it's quite simple to write a program to calculate the result of an STV election - it's a common Computer Science project for undergrads at my local Uni. After all, people carry out STV by hand even for minor polls here (like Student Union elections). One can do this with just ballot papers, boxes and people who can count and can follow the procedure for transfers. It's relatively simple to substitute a computer for the person manually doing the set process. The only slight issue is the random selection of ballots making up a "surplus" (if a candidate reaches the required no. of votes to be elected, all further votes for that candidate are transferred - in order that they are not "wasted").

Currently, the e-voting scheme in Ireland is on the back burner. It was intended that it be rolled out nationwide for the local and European elections this summer, but a committee established by the govt. (intended to rubberstamp the project and quash public dissent) decided they could not condone the immediate use of the equipment. Concerns included the security of the software (new patches were being applied only months before the election, without having been tested or checked), a lack of paper or other audit trail, and miscellaneous other concerns (the standard e-voting ones really). So the govt. backed down and we have millions of euro worth of equipment sitting in storage!

zoney ♣ talk 20:17, 27 Nov 2004 (UTC)

I imagine an additional concern would be the sheer number of candidates in some electorates. If you have a hundred candidates running for twenty seats there's no way you'll fit them all on the one screen. I imagine the computerised version of the Robson Rotation would fix the donkey vote effect, but I can assure you I'd be royally pissed off trying to find all the candidates I wanted in the order I wanted if I had to keep scrolling up and down and down and up again. Much better a piece of paper where I can go straight where I want whenever I want. Felix the Cassowary 07:09, 30 Apr 2005 (UTC)

STV is not just a voting system for people or parties

The opening sentence currently says, "The Single Transferable Vote, or STV, is a preference voting system designed to minimise wasted votes in multi-candidate elections while ensuring that votes are explicitly for individuals rather than party lists." However STV is not restricted to the use of choosing people. It could, for instance, be used to choose a flag, or a myriad of other things so I think the language in this article should be changed to reflect this. What do you think? Christiaan 14:13, 30 Dec 2004 (UTC)

I think the whole article is unduly focused on party politics and really needs to be generalised in a lot of areas. —Christiaan 19:01, 12 Feb 2005 (UTC)

Problems electing large bodies?

Can someone address typical solutions for electing prohibitively large bodies? If you were to elect, say, 100 legislators out of a pool of three or four hundred candidates, I assume the ballot and voter information materials would be so long and complicated as to diminish voter participation. What have been the experiences and/or solutions?

First of all, when do people ever vote for such insanely large numbers of people (and I mean each voter voting for all)? Any such voting system will have a very long ballot sheet! So usually if a body of such a size is being elected out of such a large pool, then each voter usually only votes for a small handful (or even just one) of the positions. Often this is done by breaking things up geographically. Or indeed, voters may not vote directly for the candidates, and just express party preferences. Or a number of other schemes exist.

In any case, STV is combined with some system of addressing this problem. It's not a problem that STV itself deals with. In Ireland we have constituencies, and use what's called STV with proportional representation to elect our parliament (an example of such a problem as you postulate).

zoney ♣ talk 11:08, 6 Jan 2005 (UTC)

The largest body elected by STV is probably the New South Wales Legislative Council. The whole state is one constituency and 21 MLCs are elected every 4 years to a council of 42. There are problems with massive ballot papers but there are also group tickets to deal with them. The parties register a list of candidates and you can vote for that list, or give preferences across different lists, or vote for candidates individually. I'll try and work up a section on group voting tickets.Alan

The New York City elections 1937-47 had just one district per borough, with up to 99 candidates on the (4-foot-long) ballot, leading to many exhausted ballots (complete preferences not required) and alphabetical voting. This was one reason STV was abolished there (another being the election of Communist candidates). Joestynes 01:13, 2 May 2005 (UTC)

Do you have any sources on New York STV - I've tried to trace accounts without much success...--Red Deathy 11:20, 2005 May 3 (UTC)

If the reduction in proportionality and the risk of Gerrymandering caused by using destricts seems like a too big tradeoff for manageability, there is the alternative of a Delegative Election method: Before election, every candidate publishes a list with his own preferences and every normal voter gets the most simple ballot where you can only make one mark. Yes, I do understand that it shifts quite some power away from the voter, but it would be trivially to implement even on national level. It also avoids the headache to choose which surplus votes get transferred.84.144.102.225 21:39, 27 May 2005 (UTC)

"Is STV a proportional voting system?"

This section in the article is misleading. It uses an extremely narrow definition of the word vote to mean singular. It even has to make the qualification that ""votes" here meaning first-preference votes". But vote is not singular by definition so why are we using this definition? STV is clearly proportional in that the outcome is proportional to voter's preferences. Why mislead people in this regard? —Christiaan 09:51, 9 Feb 2005 (UTC)

Okay I've re-written this overly complicated section. Below I've quoted the bits I've removed and explained why and what I've replaced them with, if anything.

STV is not a proportional system in the strict sense. STV does not guarantee that a party will get the same percentage of seats as it gets as a percentage of votes ("votes" here meaning first-preference votes). In fact the notion of a vote "for a party" is less meaningful for STV because votes are not necessarily for a single party. A vote can list candidates from an assortment of political parties, in any order. The candidates that are elected reflect the combined preferences of all votes cast.

STV is a preference voting system so the outcome of a vote is actually more proportional than systems that aportion first-preference-only votes as a percentage. It's like comparing apples with oranges. I've replaced it with, "The outcome of voting under STV is proportional to the collective preference of voters, assuming voters have ranked their real preferences." —Christiaan 21:18, 12 Feb 2005 (UTC)

STV does not guarantee that a party will get the same percentage of seats as it gets as a percentage of votes ("votes" here meaning first-preference votes). In fact the notion of a vote "for a party" is less meaningful for STV because votes are not necessarily for a single party. A vote can list candidates from an assortment of political parties, in any order. The candidates that are elected reflect the combined preferences of all votes cast.

Again this is mixing oranges with apples and it borders on being a non-sequitur. It also focuses unduly on party politics. I've replaced it, and part of the first paragraph with, "Because STV is a preference voting system, whereby voting is done by ranking a list of candidates, the type of proportionality contrasts with many other proportional voting systems for which proportionality is apportioned as a percentage of first-preference-only votes for each candidate."—Christiaan 21:18, 12 Feb 2005 (UTC)

Another complication with proportionality under STV is the constituency system, where a set of candidates is elected in each electoral district. There is no explicit process in STV for balancing the votes between constituencies, so the overall electoral result is merely the sum of the constituency results.

The same can be said of any electorate-based voting. There's nothing complicated about district-based voting. How can you balance a vote where there are no votes to balance? I've removed this paragraph. —Christiaan 21:18, 12 Feb 2005 (UTC)

Within a constituency, however, STV can be said to be proportional for whatever characteristics the voters valued. For example, if 60% of voters put all the female candidates first, and 40% put all the male candidates first, 60% of the winners would be female and 40% would be male. (Assuming there are sufficient candidates of each gender to make up the numbers.)

This is a follow on from the paragraph above. I've removed it. —Christiaan 21:18, 12 Feb 2005 (UTC)

STV provides this proportionality simply by wasting as few votes as possible. A vote is "wasted" if it does not elect anyone; it is partially wasted if it elects someone who gets more votes than is necessary to be elected. STV transfers votes that would otherwise be wasted, and it only transfers such votes.

I found this confusing. I've replaced it with, "STV provides proportionality by transferring votes to minimise their waste. Votes are wasted when they have no affect on the outcome of a vote. Under STV each voter has a single (transferable) vote, regardless of whether there is one vacancy or several. When the votes are first counted, all the first preference votes are allocated (i.e. the candidates with 1 alongside their name). To be elected, a candidate must reach a 'quota' of votes. The quota is needed purely to ensure that only the required number of candidates is elected, and is based upon the number of votes and the number of vacancies. A candidate that receives the quota of votes is elected. When there is more than one vacancy, and a candidate gets more votes than the quota (i.e. is successful), two activities can occur;

1. the successful candidate’s votes are reduced to the quota of votes (since that is all the votes the candidate needs). This leaves a surplus of votes to be transferred. Consequently a proportion of each of the successful candidate’s total votes is transferred to the voter’s second preference. (This proportion of votes is calculated so that it totals the quantity of the surplus votes the candidate doesn’t need.)
2. if insufficient candidates reach the quota after the first preferences are allocated, and after any surplus votes are transferred (as in 1 above), then the candidate who has received the fewest votes is eliminated from the election, and each vote for that candidate is transferred to the voter’s second preference." —Christiaan 21:18, 12 Feb 2005 (UTC)

The degree of proportionality nationwide is strongly related to the number of seats to be filled in each constituency. In a three-seat constituency, using the Droop quota, about a quarter of the vote is "wasted". These votes may be for minor candidates that were not eliminated, or elected candidates' surplus votes that did not get redistributed. In a nine-seat constituency, only a tenth of the vote is wasted, and a party needs only 10% of the vote in a constituency to win a seat. Consequently the best proportionality is achieved when there are a large number of representatives per constituency.

Again I found this confusing and overly focused on party politics. I've replaced it with, "The more candidates there are the less votes that are wasted and therefore the more proportionate the outcome, assuming voters are appropriately familiar with all candidates." Maybe someone would like to elaborate on why, in a general sense rather than as part of a party election.—Christiaan 21:18, 12 Feb 2005 (UTC)

The proportionality of STV can be controversial, especially in close elections such as the 1981 election in Malta. In this election the Maltese Labour Party won a majority of seats despite the Nationalist Party winning a majority of first preference votes. This caused a constitutional crisis, leading to provision for the possibility of bonus seats. These bonus seats were used in 1987 and again in 1996. Similarly, the Northern Ireland elections in 1998 led to the Ulster Unionists winning more seats than the Social Democratic and Labour Party, despite winning a smaller share of the vote.

They didn't win a smaller share of the vote, they won a smaller share of first-preferences. —Christiaan 21:18, 12 Feb 2005 (UTC)

Advocates of STV argue that the apparent disproportionality in STV is indicative of poor support for the party's candidates in second and third preferences. They argue that the STV result is actually a more accurate estimate of the party's support than a simple tally of first-preference votes.

This is not simply argued. This is the reason. The outcome of a vote under STV is proportionate to the voter's collective preferences. I've replace it with, "These controversies are the result of a lack of understanding about how votes are apportioned in STV. The above outcomes are indicative of poor support for candidates in second and third preferences, i.e. overall support." —Christiaan 21:18, 12 Feb 2005 (UTC)

I am reverting you back because you are wrong - particularly in the case of Malta, where almost all voters vote for all the candidates of one of the two parties. Look at http://www.maltadata.com/ and the excel files [7] [8] [9] and tell me that you still think it was the voters' preferences that cause disproportionality. STV is more proportional overall than some other systems, but it is not proportional because it is not designed to be. --Henrygb 22:53, 23 Feb 2005 (UTC)

This blanket statement that STV is not proportional is entirely misleading. In most cases, it is more proportional. In most cases, seemingly odd results arise from vote transfers - a situation specifically designed to ensure fair representation. As regards Malta - well, a contest between two parties using STV would, I imagine, produce somewhat less obvious results.

I am not against clarifying the exact circumstances under which STV does produce less straight-forward results, but stating it is not proportional is not true.

Here in Ireland in fact, it is usually referred to as PR-STV (proportional representation by single transferable vote).

I will be reverting back to the less misleading version for the time being. zoney ♣ talk 00:47, 24 Feb 2005 (UTC)

Christiaan's argument does not address the Maltese anomaly. This was caused by small district size. Gerrymandering is easier with FPTP than with STV, but it is also easier in smaller districts than larger. If the electorate is divided 50:50 (or 51:49), and each district has only 5 seats, then each district splits 60:40 (i.e. 3:2). This only produces proportionality if the same number of districts give the final seat to party X as to party Y; judicious drawing of boundaries can tilt the balance, using the same considerations used by gerrymanderers in single-seat voting systems. STV will indeed always be proportional if there is a single district, but this is impractical in a parliament with 65 seats to fill. BTW if you don't like downloading excel files, the counts are also in HTML. Also, since Malta responded to the 1981 results by guaranteeing top-up seats for the party with most first-preferences nationwide, it is no longer pure STV. Joestynes 01:16, 24 Feb 2005 (UTC)

1. STV is not proportional is entirely misleading. It is not entirely misleading. If STV regularly fails to give a more popular party at least as many seats than a less popular party then it is cannot be said to be proportional even on a basic test.
2. it is more proportional. More proportional than what? Obviously it is usually more proportional than first past the post, but nobody claims the latter is proportional. STV is less proportional than national party-list proportional representation systems, because the latter are proportional by design while STV is not. You do not actually have evidence for your assertions.
3. a contest between two parties using STV would, I imagine, produce somewhat less obvious results. Why? That should be the simplest and most obvious test as transfers (at least in the disciplined way the Maltese vote voluntarily) do not come into it. If it does not work with two parties, there is no hope it will work with more. Imagining does not come into such an analysis.
4. In the 2002 Republic of Ireland election [10], then Fianna Fail and the Progressive Democrats would have won fewer seats and not have a majority with a more proportionate system, while Fine Gael and Sinn Fein would have won more (and Sinn Fein would have won more seats than the Progressive Democrats or the Green Party reflecting their higher vote). The Progressive Democrats even managed to double their seats on a reduced vote compared with 1997. Simply calling it "proportional representation by single transferable vote" does not make it true.
5. STV's main claims are first that it minimises wasted votes, and second it allows voters to choose between individuals rather than between parties (though relatively few do in practice). If you want a genuinely proportional system look elsewhere. If you like STV, then defend it on its actual merits. But do produce evidence. --Henrygb 01:58, 24 Feb 2005 (UTC)

The Sinn Fein discrepancy reflects the difference between first-preference votes and the total of all preferences of all voters. Most voters who don't put Sinn Fein first would put it last: 5% of first-preferences + 0% of second preferences may not be as valuable as 2% of first preferences + 10% of second preferences. So, yes, STV seats are not proportial to first-preference votes, but they are proportional to thoroughgoing preferences, within a single district. Districts with small numbers of seats offer a cruder range of possible seat distributions; no voting system can get around that. Summing the results of multiple districts will therefore be cruder than having a single large district. That is not inherently an STV-issue (small districts, each with a list system, have the same problem), it's just that it's easier to have large districts with list systems than with STV. The Progressive Democrats discrepancy reflects the fact that, unlike 1997, they ran candidates only in districts where they had some chance of winning, so lost the scraps of votes they had previously got in other areas. Again, this is a small-districts issue, not STV. Joestynes 03:19, 24 Feb 2005 (UTC)

Even if I haven't dealt with the Maltese anomaly correctly, which I'm currently having a more indepth look at, you'd be right, it is just that: an anomaly. To then make a blanket statement that STV is not proportional and impose dictrict implications on it is to confuse the matter of what STV is. STV is not simply a voting system for electing political dictrict-based candidates. It's far broader than that and this article should reflect that. —Christiaan 10:31, 24 Feb 2005 (UTC)

STV is certainly proportional within voting constituencies. Adding constituencies into the mix shakes things up, *a bit*, but it's a damn sight better than first past the post. The constituencies in Ireland, as far as I know, are to ensure decent geographic proportionality.

Constituencies aren't specific to STV though, and to use that to say STV does not return proportional representation is misleading.

Again, I repeat that I am not against going through the factors that impact on PR, but I am against outright stating that STV not proportional. It would nearly be better to omit the whole section, as it seems that the PR issue depends entirely on voting factors other than the use or not of STV. zoney ♣ talk 11:18, 24 Feb 2005 (UTC)

Proportional representation is not absolutist. It is imposible to acheive a completely proportional result in any electoral system, if a party receives 3.5% of the vote for a 100 seat chamber then it receives 3 or 4 members, neither figure is truly proportional and one can't have half a representative. In truth there are only deviations from proportionality. PR-STV may deviate slightly from proportionality in certain situations but it is still more proportionate than say SMP (FPTP). Proportionality is not always desirable. Closed list systems can offer good proportionality, but electors can only vote by party. 'Top-up' systems such as the alternative vote as in Germany/Scotland/Wales have the flaw that the top-up lists are often closed lists and they create two clases of representatives, in the case of Wales and Scotland the top-up element is too small to generate true proportionality (it should be 50% of the whole). It seems to me that in some regards the argument about proportionality in STV is somewhat redundant. How does one measure proportionality when votes can be divided and sub divided (half a vote here, a third there etc)? And is it correct to compare the result exclusively with the proportion of first preference votes? Why not first and second, or first, second and third? Proportionality is acheived by having equal representation so that all elected representatives represent the same number of constituents. If two electoral constituencies have the same number of elected representatives, but one has twice the population than the other, then due to the unequal representation PR is not acheived, this would be true of STV or even the open list system used here in Finland. To blame STV for faults like unequal representation or under representation is just plain wrong. To get good proportionality I would think that constituencies would need to be of about 5-8 representatives each and for constituencies to be of between 125 000 to 200 000 people. Therefore there would be about 25 000 people per representative. I believe in Ireland there is a rule that TDs represent between 20 000 and 30 000 people each and that the number of TDs per constituency is varied according to the population size of each constituency. What STV does do is make a strong link between the elected representative and her electors. There are No safe seats in STV and parties have a relatively weak hold on their members. This is because people from the same party will be contesting seats in the same constituency. There are also no party lists, the electors decide who is elected unlike in a closed list system. The open list system offers similar benefits to the electors, but the link between the constituent and the representative is not so strong and an elector can vote only for one candidate, the vote going to party for the election and the candidate for the party list, there is no posibility of splitting the vote between candidates from different parties. In PR-STV the party offers little protection for the politician, if you don't deliver then your constituents can easily vote you out while retaining members of parliament from the same party. It is also true that Independent candidates that are very popular can get elected relatively easily--Alun 10:08, 3 Apr 2005 (UTC)

Thanks for your contribution Alun (Wobble). There's some good text in the above passage which might be worth plugging into to an article like proportional representation somehow, or even this article. By the way, you can sign and date your contributions automatically by putting ~~~~ at the end. —Christiaan 10:28, 3 Apr 2005 (UTC)

Just looking at the data from Malta, I think I can see a way of resolving the debate. Within each constituency, it seems the results are fair, and proportionate.

Assuming that first preferences will transfer within each parties candidates only, we can divide total first preferences by the quota - in each case the outcome turns out to be correct to the number of seats gained. The problem appears to have been that STV - unlike other PR systems, assumes at least one quota of wasted votes (the absolute number is greater with the droop as compared with the hare). The Maltese national party in '81 seemed to have had a disproportionately higher number of partial quotas - 10 of 13 seats had a remainder of more than 0.5, of the MLP's three seats with a wasted quoient of greater than 0.5, only one was above 0.6 as compared with seven of the NP's.

pauses for breath - so, perhaps the article should be altered to state it is tendentially proportional, but doesn't deal with remainders. That's my suggestion. --Red Deathy 08:24, 2005 Apr 15 (UTC)

It does deal with remainders reasonably, in many people's view possibly better than any other system, but only within a single constituency. But since the constituencies are usually not so big as to make the ballot papers impractical (unlike party list systems, where the size makes less difference to the burden on the voter), the combined effect across many of is what so often leads to the examples. --Henrygb 16:57, 15 Apr 2005 (UTC)

Actually, having a fit of insomnia last night, I was wondering whether this distortion is due to the use of a Droop rather than a Hare quota. It is the Droop quota that sets the target at the minimum required to take the seat, and thus leaves the near one quota remainder behind. A hare quota would see all the votes eventually ascribed to a winner - the only distortion from proportionality then would be non-continuing votes and constituency sizes. Perhaps the quota section should have mention of this sort of thing.--81.134.165.178 10:24, 16 Apr 2005 (UTC)

That would be a strange way of looking at it. First, it would require the count to continue even when the number of remaining candidates is equal to the number of places remaining; a fairly pointless exercise. Second, given that receiving a Droop quota of first preferences would guarantee election even if Hare quotas are used, it is reasonable to argue that all votes received between the Droop quota and the higher Hare quota are unnecessary and so should be counted as wasted. --Henrygb 19:37, 29 Apr 2005 (UTC)

Well, I was talking theoretically, since we are discussing in the general, the theory of STV is that were all votes counted, they would find a home, i.e. that the appearance of wasted votes is an effect of the convenience of not counting them - for Instance, Scotland's new local council STV counts have a requirement to not bother transfering surpluses when the surplus is not enough to save a candidate from being eliminated - perfectly reasonable, practically...--Red Deathy 08:35, 2005 Apr 30 (UTC)

Brian Meek has a fairly elegant definition of the quota — An essential feature of an STV election is the 'quota'. If there are s vacancies to be filled, the quota q is the smallest number such that, if s candidates have q votes each, it is not possible for an (s+1)th candidate to have as many as q votes. Thus if the total votes are T, then T-sq < q, but T-s(q-1) >= q-1, whence q = [1+T/(s+1)], where the square brackets denote 'integer part of'.' Alan

The definition.

"The Single Transferable Vote, or STV, is a preference voting system designed to minimise wasted votes in multi-candidate elections while ensuring that votes are explicitly for candidates rather than party lists."

That doesn't quite cover it. I think an aditional sentence, which I'd propose go something like this needs adding:

"In it's basic form, it works by allocating an elector's whole vote to their highest ranked candidate who has not been removed from contention through either election or elimination in sucessive rounds of counting."

I think that is needed to explain the singleness of the single transferable vote.--Red Deathy 09:49, 2005 Apr 7 (UTC)

Watch out for it's/its. But a more serious problem with what you have said is that if a voter's prefered candidate is elected, then on a fraction of the voter's vote gets transfered to the next choice, or there is only a fractional chance that the vote will be transfered (depending on system). --Henrygb 11:41, 7 Apr 2005 (UTC)

Henry, yes - perhaps it needs to read "already been removed" - I'm trying to get across the idea of the vote finally coming to rest somewhere. Perhaps thus: "In it's most basic form, it works by allocating an elector's whole vote to their highest ranked candidate who has not already been removed from contention through either election or elimination in sucessive rounds of counting. There are various different ways of carrying this out in practise" --Red Deathy 13:17, 2005 Apr 7 (UTC)

I note you've removed 'Whole' - fairy snuff, though I still think something needs including definitionwise to indicate a difference between Borda and Cumulative style preferential voting. i.e. that preferences indicated are not consequtive votes or of cumulative value, but indications of the order for which you would vote for the candidate based on their availability.

The vote always remains of a value of one - even under Senatorial rules/Gregory method that value remains because the fractional transfers can be understood to add up to the value of the whole votes that would have been transfered. All that happens then is that the destination of those transfers is effectively communalised. The logical basis remains the integral vote.

Where this variant is not used, the fractional chance of being transfered is external to the vote - and differs in no way from the situation under first past the post or other systems where you find your choice of candidate unavailable, and so must decide whether to vote for an alternative. Finally (*breathes*) You can ensure that your vote goes to your highest preference candidate, by only marking one preference.--Red Deathy 08:05, 2005 Apr 11 (UTC)

My points would be:

1. In a fractional vote transfer system, your vote can end up in bits all over the place, with different parts helping elect several people (or failing to. Indeed, with Ben Arnold's example on a modern mechanism to discourage tactical voting, even whole votes from eliminated candidates can be split up when transfered.
2. In a random vote-selection transfer system, the vote may stay whole through the process but you cannot be certain what will happen to it in the counting process, even if you can see everyone else's vote; that is rather different to most other voting systems, though not random ballot.

So I would prefer to say "one" rather than "whole" as the meaning of "single", in that your vote is worth a total of "one" whatever happens to it. But I accept that the transferable part means that your vote (or a bit of it) can (-able) only move on to other candidates when your earlier choices are not longer active in the count. --Henrygb 18:35, 11 Apr 2005 (UTC)

The example.

I think it would be useful to concretise the example a little. There is a useful example in Lambert and Lakeman - [11], which runs something like:

"If a class of children were choosing representatives, say, they could stand behind the candidate of their choice. Since they all know that each candidate only needed a certain number of classmates to vote for them to win, those arriving last in line for a candidate who already has enough votes would choose to not waste their vote, and instead move to another line to help someone else to win. Likewise, those children whose candidate obviously could not win, would move to another line, and so on, until all the positions are filled.

When using an STV ballot, these preferences are set out in advance, as instructions to the counters."

This could go at the begining of the Voting section.

Again, it emphasises the singleness and the transferability of the vote, and illustrates the logical underpinnings of it.--Red Deathy 09:49, 2005 Apr 7 (UTC)

Renaming

Good move on the renaming. STV is used both as a proper noun (the name of the system, Single Transferable Vote) and a general description (the working of the system, ie, votes used in an electoral system by means of a single transferable vote). As the system is formally known as Single Transferable Vote (and written STV, not stv) it clearly belongs in the title in the capitalised form. It was rather hurting Wikipedia's credibility that someone like STV was written in the wrong case. It made us look as though we did not know what we were doing, which was a deeply unfair impression of the work that people have been doing in this article. It is a lesson of how Wikipedia needs to be careful to get the name and capitalisation correct, so as not to undermine the strength of its work in the articles. FearÉIREANN 01:48, 30 Apr 2005 (UTC)

Not criticising the capitalisation of 'Single Transferable Vote' (on which I defer to others who know better), but whether the abbreviation is STV or stv is totally irrelevant to whether the expansion is Single Transferrable Vote or single transferable vote. English (nowadays) almost always capitalises the first letters in mostly-initialisms (except if the abbreviation is pronounced as a word); CDs are compact discs even though the abbreviation is CD.

Actually it is very relevant (as is capitalisation in general which is why it exists). STV refers to a specific, defined, methodology of voting. Lowercased, single transferable vote can be used to describe a process which may replicate the idea but not necessarily follow the methodology of STV. So one for example writes about PR.STV (Proportional Representation using the Single Transferable Vote [written like that]) which means a strictly defined form of proportional representation, using the specific concept and (crucially) the methodology of STV. It is the same as understanding the difference between the generic word president and the specific title President of the United States, the latter of whom has a specific meaning, concept and powers that the general word president may not, as it could be referring to any president, anywhere, possessing any powers or none at all. Capitalisation is crucial to know how general of specific a term is being used as. FearÉIREANN 04:23, 30 Apr 2005 (UTC)

Sorry, I think you missed my point. I was talking about the abbreviation and that particular argument, not the general case of 'Single Transferable Vote' vs 'single transferable vote'; I'm talking about when you say (and written STV, not stv), not when you say STV refers to a specific, defined, methodology of voting. Actually, I suspect I might've misunderstood you, in which case ignore me. Felix the Cassowary 06:41, 30 Apr 2005 (UTC)

Cleaning up the article

• Can I ask people to focus on the article and not the particular form or virtue of STV, PR or whatever that will solve all known problems in the universe? The article needs serious attention. Throughout the text:

• • Headings are misleading.

• • Terminology is inconsistent.

• • A number of minor topics that should have their own articles are crowded into this one.

• • Use of sub-topics, numbering etc is wildly inconsistent.

• • I also feel this article lacks a simplistic and easy to understand introduction. It jumps straight into mathematics and complex terminology. Alan

• • • I've taken a stab at this, and have moved a lot of the math into a new counting article since most of it isn't quite needed for the main article here. I'll also add more to the introduction later to make the topic more approachable for novices. Scott Ritchie 10:19, 17 May 2005 (UTC)

• Is there any good reason for the subtopics 'In practice' and 'STV in Australia' not to be merged under a descriptive title? Alan

• Can we set up a new monotonicity sub-topic under issues?

• Are Countback and Recount actually separate ways to deal with a vacancy?

• • I read the two electoral acts. The methods are almost exactly the same, so I merged the subtopics. I also deleted an inaccurate sentence: "Because this election is conducted by using the votes that specifically elected the now removed representative, as well as the remaining votes that were wasted, the replacement candidate is most likely to represent the preferences of the voting population that are not represented in the government." The wasted votes are not used. Logically a system almost guaranteed to return a clone cannot favour or disfavour anyone. See Schedule 6, Tasmanan electoral act

• • • By "remaining votes that were wasted" I believe the correct term is instead "exhausted and wasted". Consider an election where candidate A has resigned and we use the countback method to calculate who to take his place. Suppose 2/3 of a quota-filling number of ballots read A-C-B and 1/3 read A-B-C. In addition, half a quota of ballots simply read B, with no mention of A. In the original election, B was eliminated early, and this half of a quota of ballots were "exhausted" - effectively volunteering themselves for the waste pile. If we eliminate candidate A from the ballots and use the countback method to compute a replacement, however, we still need to reconsider the ballots that had B removed - since more people (among the uncounted ballots) prefer B to C here, B should be the replacement, even though C has 2/3s of A's original voters. The way this is done for our student elections here (UC Davis) is to configure the software to automatically mark the other members as winners and immediately drop the resigning member, then recompute the election. A great page on it is here:

http://www.daviswiki.org/Countback - Also, we need a section on exhausted votes. I'll get around to adding it tomorrow. Scott Ritchie 09:23, 16 May 2005 (UTC)

• • • • The worked example is inaccurate. In a countback you cannot have more than a quota of votes in play, so an example combining 2/3+1/3+1/2 does not add up. The legislative term seems to be exhausted votes.

• • • • • Why not? There can be nearly a quota of votes still uncounted (using Droop quota) at the end of an election, and then nearly two quota's worth after the votes from the resigning candidate return into play. All that's needed is for that 1/2 to come from the pool of votes that ended up wasted in the original election - ie they didn't elect a candidate. Scott Ritchie 15:43, 17 May 2005 (UTC)

• • • • I've finished reading the Tasmanian and ACT electoral laws. The {http://www.elections.act.gov.au/casual.html ACT fact sheet] is very clear. Neither uses wasted votes. A rationale is hard to find because it's basically not an idea I've seen before. Essentially, it would disfranchise the people who elected the outgoing candidate because you're partially rerunning the first election rather than an election among those people who elected the outgoing member. Incidentally, we need to revise the defintion of wasted votes. They do not include 'those that go to... candidates that have already won'.

• • • • • They do include "candidates that have already won" in non-STV systems, however. This is the "Wasted vote effect," which is probably worthy of its own article. Waste is minimized in STV systems, and this particular aspect of waste happens to be reduced to zero. Also, including wasted and exhausted ballots is done for some elections using computer programs. See this countback: http://davischoicevoting.org/countback/ackerman/Vacancy%20Round.htm - The 350 ballots that went to elect James Ackerman, along with the 327 exhausted ballots, would have elected replacement Leticia Miller. Both were from the same party. Note that the 327 "exhausted" ballots includes the wasted ballots - in this case those listing James and the other winners at the very bottom and those portions that transfered in surplus from them to end up in the exhausted pile after their candidates didn't win. Note the total votes counted here is 677.61073, although 224.38833 become exhausted again after the initial winners are excluded and candidates start getting defeated since they don't list Leticia or her closest competitor. Note that some systems don't allow for early exhausted ballots, and exhausted ballots are by necessity wasted. To some degree, there really isn't a meaningful distinction between wasted and exhausted ballots in STV, since STV systems can't waste a ballot by putting more than a quota's worth of votes into a particular candidate - therefore the only wasted votes left are those in the exhaust pile after quotas are filled. It is interesting that the Australians intentionally don't count the exhausted pile when computing a replacement candidate - this has the effect of ensuring the same votes will remain wasted (and the same people unrepresented), however it also has the effect of making it more certain that the replacement candidate will resemble the original. Note that the Australian method isn't possible in systems that allow incomplete preference listings if you wish the replacement candidate to represent a full quota - it's virtually guaranteed that at least one of the votes during the countback will exhaust itself before a new candidate is chosen, and if you're only using a quota's worth of votes (those of the original winner), you will by necessity have less than a quota. Scott Ritchie 13:15, 18 May 2005 (UTC)

Malta 1981

I have removed the following section:

Critics of the constitutional change, however, argue that the electoral results reflected a strength of STV rather than a deficiency as the system was able to recognise a preference among voters for some charismatic individuals of one party but the party platform of another. Furthermore, proponents of STV argue that the outcome of an election should not change if the only difference in the ballots is the partisan affiliation of some of the popular candidates, rather than an actual difference in the preferences expressed by the voter.

The problem with this is that there is no evidence that the problems in Malta were caused by a candidate being more popular than their party. This Even those who were very popular saw almost all of their surpluses transfer to other candidates from the same party. Excel table with counts provides the data. --Henrygb 11:03, 11 May 2005 (UTC)

Imperiali quota?

Can anyone out there provide any citations that imperiali has ever been used in an STV election, or the rationale for so doing? Droop and Hare both make a sort of sense, but Imperiali is a) arbitrary, and b) I can't see how it could work in STV since it would gaurantee thatan excess of quotas would be filled....--Red Deathy 12:38, 2005 May 11 (UTC)

• Nope. It should be deleted or outlinked to its own article like the Hagenbach-Bischoff quota. Alan 13:32, 11 May 2005 (UTC)

• • Should we bite the bullet and delete it? Alan 05:06, 14 May 2005 (UTC)

Ordering candidates in a list system

Is there any actual example of a party using STV to order candidates in List PR systems? This sections strikes me as another case, like the Imperiali quota, where a minor topic gets undue prominence.

• Agreed, it doesn't seem to fit or make much sense. I'm going to remove it now until someone can provide a source for this actually happening, as it seems unlikely that List PR systems would use STV internally for parties rather than just using STV in the electoral system. Scott Ritchie 11:58, 10 July 2005 (UTC)
  • Here is the text I removed, feel free to re-add it if you can find a relevant source:

Ordering candidates for List PR

STV can be used used as a social preference function to produce an overall ranking of candidates. For example, closed party-list PR allots seats proportionally to lists and chooses candidates from that list in the order set down on the list. Some parties decide this centrally but others give the choice to the membership. STV has been used in such elections; in such an election the party members mark their ballots in just the same way as in a normal STV election.

When the votes come to be counted the same ballots are used first for an STV election to choose one winner, as an instant runoff election; then the same ballots are used to see which two would win when the votes are counted for a two winner election and so on. The head of the list is the party member who would be elected in the one winner count. The second place is taken by the next party member who succeeds in the two winner race (in the unusual event that the winner of the single winner count did not succeed in the two winner count, then the first to be elected in the two winner count), and so on.

Scott Ritchie 12:21, 10 July 2005 (UTC)

Warren STV?

The webpage http://stv.sourceforge.net/ makes a reference to "Meek and Warren STV". I've not had any luck digging up what "Warren STV" is. Can someone explain? Should the answer go into the article? Thanks. TreyHarris 10:05, 14 May 2005 (UTC)

• See Incorporating X-voting into Preference voting by STV. We already mention Meek. Alan 11:39, 14 May 2005 (UTC)

• Ah, thanks. So, if all voters use preference voting, "Meek and Warren STV" ≡ "Meek STV"? I read that webpage as saying "Meek and Warren STV" meaning "(Meek STV) + (Warren STV)", not "(Meek + Warren) STV", but it's ambiguous. Given how much moaning I've heard lately on policy pages about how preference voting is a pain, and approval or plurality voting is worth the unfairness just to reduce that pain, perhaps Warren's X-voting method should be mentioned. TreyHarris 21:16, 14 May 2005 (UTC)

Subdistricts method of replacements

Is this method used anywhere? It's not even clear how it's supposed to work - how do members elected from a multi-member constituency claim "sub-districts" anyway? I'm nearly certain the countback method (with or without counting exhausted votes) is used instead, as the votes going for a candidate represent "perfect sub-district" exactly. Scott Ritchie 18:04, 18 May 2005 (UTC)

I went ahead and removed the following text: "====Subdistricts==== If each legislative district is divided into as many sub-districts as there are representatives, each representative can be elected in the single-member sub-district where they perform best. Vacancies can then be filled by an instant runoff ballot within that sub-district." - feel free to add it back in if you can find some sort of reference for use of the subdistricts method, or can explain how it works. Scott Ritchie 02:34, 24 May 2005 (UTC)

It was proposed by Hoag and Hallett. The only use I'm aware of is in some Australian NGOs, mainly student associations and that's my doing (or fault). The way to elect members in the subdistricts was to run a second count with only the winning candidates eligible, and elect the candidate who 'won' the subdistrict if they had not done better in another subdistrict. It's essentially the same as a countback. Both methods are designed to identify the electors who chose the outgoing candidate. The advantage of subdistricts is that it allows for changes of opinion since the general election. I have not read Hoag and Hallett (and don't have a copy to hand) for a while, but I think they said it had been used in some US elections.Alan 00:35, 28 May 2005 (UTC)

Proposal to merge STV and Instant-runoff voting articles

As irv is a single-winner case of stv, I propose to merge the two articles into one, with a redirect page for irv. I request any comments and volunteers to help if deemed to be constructive.--Fahrenheit451 21:34, 25 May 2005 (UTC)

I don't mind, but STV fans might, saying IRV is not PR, or IRV is simpler because there is no transfer of surpluses, or that some IRV systems eliminate all but the top two candidates after the first round of counting, or that when Australia has AV/IRV for the lower house and STV for the upper house they are very different. --Henrygb 00:09, 26 May 2005 (UTC)

I strongly disagree. The two methods are quite different, both in terms of exlpanation needed, results, and encylopedic impementation. There's more, or should be more, to the articles than just how the methods work. Moreover, the STV page is already approaching too large as it is - we had to section out the particular details of counting since it was too long, and if IRV's content were here we'd have to do that as well. Scott Ritchie 00:24, 26 May 2005 (UTC)

I disagree as well. They may be fundamentally related, but in practice and implementation they're quite different. RSpeer 15:08, May 26, 2005 (UTC)

I disagree. The fact that one thing is a degenerate case of the other doesn't mean they should be dealt with in one article, any more than we should collapse Chemistry into a discussion of a special case of Physics. It makes most sense to describe IRV in its own terms, and describe STV in its own terms, and then point out that STV's degenerate case is identical to IRV. If one used STV terminology to describe IRV, it would be a disservice to people trying to learn about IRV; the reverse, a disservice to people trying to learn about STV. If you're going to use two different terminologies, then there doesn't seem much point in merging in the first place. TreyHarris 16:11, 26 May 2005 (UTC)

Absolutely not, no way under any circumstances. They are different and used in different contexts. FearÉIREANN^(talk) 23:26, 26 May 2005 (UTC)

Looks to me that it should not be done then.--Fahrenheit451 23:21, 26 May 2005 (UTC)

I think it actually should be done. The math is identical. The Irish constitution actually specifies therir president is elected by 'the system of proportional representation by means of the single transferable vote' (S13.2.3) It would need a careful paragraph about IRV and AV being single district forms of STV. The ACT electoral act provides variable quotas becuase not all their districts have the same magnitude.The alternative is a third article discussing what the STV and IRV have in common. Alan 00:28, 28 May 2005 (UTC)

Well, the comments given above are a super-majority opposed to the idea, which is compelling not to. The irv article already has references at the beginning about its genesis from stv. I think we should leave well enough alone.--Fahrenheit451 22:38, 30 May 2005 (UTC)