Talk:Single transferable vote/Archive 3

Archive 1

Archive 2

Archive 3

Archive 4

Archive 5

→

Archive 7

Choosing Later Candidates

Are candidates who were eliminated in early rounds restored when choosing later candidates? I had thought so, but the description suggests they are not. Example:

99 people favor A > B > C > D > E > F > G 1 person favors E > F > G > A > B > C > D 1 person favors F > G > E > A > B > C > D 1 person favors G > E > F > A > B > C > D

B, C, D get no first place votes, so are immediately eliminated. It ends up choosing A, E, F, G, far from a proportional representation. Had B, C, D been restored after A's election, they would have been picked, even with the downscaled votes. Or am I misundertanding? 71.8.222.86 19:33, 13 February 2007 (UTC)

You are misunderstanding. (Assuming there is more than one winner) A's surplus is distributed before any eliminations, since the surplus is big enough to affect the final result. I once saw an election where someone started with 0 votes, received 0.01 of a vote in a surplus transfer, and then was eliminated. --Henrygb 23:55, 13 February 2007 (UTC)

There are a number of problems listed which are actually about the limits of any form of proportional system - for example "malapportionment". These are all reasonable discussions - including the fact that most people that use STV deliberately sacrifice strict proportionality to keep districts (constituencies) etc small.

Oscar nominees

The Academy of Motion Picture Arts and Sciences employs a preferential ballot for the selection of the five nominees in each of the various award categories. I believe that the counting system is a form of STV although they do not actually call it that. Here is an article I found that describes the process:

• Eight things every voter (and fan) should know about Oscar’s decidedly unique nomination process.

Perhaps somebody more knowledgeable than I could confirm that the process described in the article is STV? The article describes the setting of a Droop quota, and any film that gets enough votes to meet that quota is considered ‘nominated’. Then, the article says, "The stacks of votes belonging to films that have clinched a nomination are immediately set aside. Then the remaining ballots are resorted into the existing piles based on the voters' second-choice films." --Citefixer1965 00:27, 27 January 2007 (UTC)

Not quite. Apparently, you are only allowed to rank five choices (maybe at most five, maybe exactly five, the article is unclear). True STV should allow (if not require) you to rank all candidates. Since there are thousands eligible for, say, best supporting actor, that won't really be feasible until the ballots are cast online. The limit is set at five presumably is because that's the number of winners [winners, that is, of the "honour of just being nominated"]. The use of five as the limit is essentially arbitrary; there's no particular reason they don't limit it to 8 or 10 or 15 rankings, which would more closely approach true STV. jnestorius^(talk) 01:44, 27 January 2007 (UTC)

Reading on, I see you also need to get at least one first-preference to be successful. In true STV, you can get zero first preferences and still be elected simply by receiving transferred votes from the surplus of previously-elected candidates. jnestorius^(talk) 01:49, 27 January 2007 (UTC)

Still interesting enough to add to History and use of the Single Transferable Vote#NGOs. Thanks! jnestorius^(talk) 02:45, 27 January 2007 (UTC)

It looks STV without transfers of surpluses (but with transfers from those eliminated jumping over those already elected). It would be interesting to know if quotas reduce in later rounds. Without transfers of surpluses, those with no first preferences will be eliminated anyway. --Henrygb 00:02, 14 February 2007 (UTC)

I think it does redistribute surpluses: point six of the exposition says "We haven't even gotten into the scenario in which a film receives considerably more votes than it needs, in which case it secures a nomination and all of its ballots are redistributed into other piles, with the second or third or fourth choices counting as a fraction of a vote." jnestorius^(talk) 15:41, 14 February 2007 (UTC)

You are probably right - I only got as far as: and it'll move to "A History of Violence" only if "Cinderella Man" is either out of the running or it has already secured a nomination and no longer needs his help --Henrygb 23:16, 14 February 2007 (UTC)

Photograph of a best-actor nomination ballot here: [1]. --Citefixer1965 08:42, 19 February 2007 (UTC)

I found an article in Variety from 1993 that gives more information. The article says that fractional surpluses are redistibuted if the candidate’s tally is more than 20% over the quota. --Citefixer1965 17:49, 19 February 2007 (UTC)

That bit is very odd.

Another check kicks in when a nominee receives more than 20% of the quota from first-place votes. A system had to be instituted to keep from discarding all of those ballots, which would skew the count on the remaining potential slate. The proportion over 20% is then calculated to a percentage that reflects the required quota. So, if Clint Eastwood were to receive 75 top choices -- 50% above the necessary level -- the individual votes would be revalued to 0.7 and the second-place votes, which would have otherwise not been counted, are given a value of 0.3.

The trigger criterion seems arbitrary but the example gives a correct transfer ration for ordinary STV. Perhaps the journalist has simply misunderstood the process? Really we need a quote from the horse's mouth, rather than a hack with a deadline who's never seen the system before. jnestorius^(talk) 01:10, 20 February 2007 (UTC)

Capitalization

Why is the title of this article capitalized? →bjornthegreat ^t|_c 04:51, 1 June 2006 (UTC)

Why are so many instances of the phrase "single transferable vote" capitalized? That is like capitalizing "runoff voting" or "underground irrigation system". It is incorrect, according to Wp:manual_of_style#Capital_letters. It is not a proper noun. -Pgan002 02:42, 3 February 2007 (UTC)

I just came to this talk page with the same point. Can anyone make the case that each word in this article's title should remain capitalized? -Phoenixrod 19:46, 20 February 2007 (UTC)

Finding the winners section

This section was marked as "repetition" but I think I have found a good solution. Essentially, the same information was being presented three times. The ordered list of steps is clearest so I have retained it. Hence I have removed the tag. Endorphin78 11:16, 27 May 2007 (UTC)

Creating a new sub-page

Lo All, well, I noticed taht the article was too long and all but unreadable - the issues section was a rambling rag-bag thatstill needs serious editting - but my main thought is that we need to keep the main page simple, we've already forked off a couple of times for history and counting - the topic is so big we need a lot of pages, and users can navigate them by category if need be. Basically, I picked out key sentences from the paragraphs I moved so as to form the basis of a precised section here, while moving the bulk of the text elsewhere. The struggle is on to regain featured article status!--Red Deathy 07:28, 31 May 2007 (UTC)

Should the main article be about, what it is (i.e. the basic idea of selecting multiple candidates in order of preference), the history, and where it is used. Things like implications & counting etc should be (is?) as separate article. --Nate1481(^talk/_contribs) 13:54, 31 May 2007 (UTC)

Unclear

This section is unclear: "two tables below (the first is a numerical representation, the second is pictorial)". Ask someone who has never seen this page what this means, regarding the tables shown. Also, why does chocolate appear twice in the first table? DBrnstn 14:46, 18 August 2007 (UTC)

Unrelated: are "Write-in candidates" typically allowed? If not, this blocks write-in campaigns. If so, this can change (and possibly thwart) the dynamics, if voters write-in their first choice as their second, third, and fourth choice also. (Orange, orange, orange, orange.) Perhaps "write-ins may not duplicate prior choices." DBrnstn 14:46, 18 August 2007 (UTC)

STV Action

I am the Webmaster of STV Action (http://www.STVAction.org.uk). I would STV Action tobe added to the list of proponent groups in this article. I thought it best not to add this link myself after reading the Conflict of Interest Policy WP:COI. John Cross 21:39, 25 August 2007 (UTC)

Getting back to featured article status

So, what needs to be done to get this page back to featured article status? I think we should make that a high priority, and also get Instant Runoff Voting to featured status. Perhaps a peer review is in order for that latter objective. Captain Zyrain 17:41, 9 October 2007 (UTC)

London Mayoral Elections

Can anyone, with more knowledge on this subject than me, determine whether the London Mayoral Elections (2008) used STV. I remember choosing a first and second choice candidate? C.harrison1988 (talk) 03:33, 28 June 2008 (UTC)

London Mayoral Elections use a limited form of preferential voting called the Supplementary Vote, which differs from STV in two key ways: (1) Voters are only allowed to express two preferences. (2) Candidates who do not come first or second in the first round are eliminated even if it is still possible for them to win after second preferences are counted. Duncan Keith (talk) 06:04, 28 June 2008 (UTC)

Current Use

Now it reads "...it remained in approximately two cities, including one in Massachusetts." Approximately two? Could it be one or three? What other(s) retain STV? Irv (talk) 04:58, 26 November 2009 (UTC)

Unclear - allocated value

In comparing "paper-based" to "value-based" counting methods, the definition of "value" is unclear. An example showing the difference would be helpful. -- Beland (talk) 02:52, 5 January 2009 (UTC)

I've removed the text in question entirely, I'll paste it here:

There are two basic forms of counting an STV election: one is based on the number of ballot papers allocated to a candidate; the other, more accurate system, is based on the value of the vote as it progresses throughout the count. This variance is important in the calculation of a candidate's surplus. The paper-based formula (the surplus divided by the number of ballot papers) undermines proportionality and the "one vote, same value" principle due to some votes increasing in value at the expense of other votes. The value-based formula (the value of the vote divided by the surplus) avoids these shortcomings, ensuring proportionality and maintaining the "one vote, same value" principle.

It seems like this is an attempt at explaining the difference between Meek's method, the Gregory method, and the old "pick a few random ballots for the winner out of the hat and count them" method. In any event they're better explained just 2 sections down in the article, after the example at the Differing Counting Methods section. Scott Ritchie (talk) 04:12, 21 January 2010 (UTC)

Failed Definition

A definition of an acronym must be complete. To give the words which the letters of the acronym represent does in no way explain the acronym appropriately. We are left with a three word phrase without a clear meaning. This needs to be addressed first and foremost. - KitchM (talk) 16:24, 31 January 2010 (UTC)

Do you have more specific suggestions for improvement? Gabbe (talk) 23:21, 31 January 2010 (UTC)

Andrae

Did Thomas Hare and Carl Andrae invent STV independently of each other, or did one copy it from the other? Are we certain that Hare was the (only) originator? 86.163.210.187 (talk) 17:35, 26 August 2010 (UTC) Update: it's my understanding now that the two men invented it independently, and that Andrae has a reasonably good claim to have invented it first. 86.163.4.201 (talk) 22:27, 31 August 2010 (UTC)

Edit on the issues section

I am making these comments as Edits because I find the Discussion on this subject wholly off-topic.

It is elementary that STV procedures can be affected by strategic voting. The procedure can give a minority party a majority in the Legislature. (The party is a minority party in the sense that it commands strictly less than 50 percent of first preferences and is beaten in pairwise comparison by another party, which itself is an all round winner (in pairwise comparisons with all other parties)) Further, such an all round winning party can, under any STV procedure, be left with no representation in the Legislature at all.

All this can be shown with 3 member constituencies and 3 parties putting lists forward.

Further every STV procedure can deliver a set of members in the Legislature, which set is weakly Pareto inoptimal (ie there is another set of candidate-members which all voters prefer under sincere voting).

This can be seen very simply by considering a 3 member constituency with 4 party lists. Three of the parties, a, b and c have about the same level of support which is overwhelmingly larger than z's, the fourth party, with it's small amount of first preference support. .

The three largest parties' voters vote cyclically, viz, aPbPc, bPcPa, and cPaPb.

Because of the small amount of support for z, the member set under any STV system must be 1/3 'a' party candidates, 1/3 bs and 1/3 c's.

Now z is the second worst alternative for all these voters. Thus when these voters exhibit risk aversion the set of members all from z is weak Pareto superior to the equal mix of a, b and c members.

This result may need a little explication in view of the weak state of the theory of coalitions to be found elsewhere

The result can be justified in three ways. Either equal 3 way voting in the Legislature results in a lottery (probably the best outcome given our assumptions here) or some coalition emerges which is the result of 3 party bargaining and in which the outcome is an admix of distinctive policies of the parties party to the coalition or z is better than the convex combination of the policies advocated by the parties a b and c. Members always vote 'sincerely' in the sense that they advocate their best alternative throughout. The lottery option is justified by the balance between the parties. In contrast, should the three popular parties comprise a pair with opposed preferences, xPaPy and yPaPx and a middle ground party putting 'a' top (with a fourth party, z, as above) in every constituency, then the member set under any STV procedure is equal numbers from the x, a and y parties. Now despite the FPTP equality 'a' is an all round winner against x and y in the Legislature and the electorate. Should z be in second place in the ordering of all the x, a and y parties supporters then it would be z that would win in the Legislature and the country. (Given, of course, that the Legislature uses pairwise comparison methods rather than the wholly inadequate 'Aye' and 'Nay' method used in the British Parliament(a method producing results worse than some Arrow has described as threatening sanity)!)

No clear pattern emerges. With singleton all round winners Pareto inoptimality can occur. Equally the key is NOT 4 parties nor 3 member constituencies. For if xPaPy and yPaPx for half each of all voters, with little variability in any of the 2 member constituencies, then the member set is equal numbers of x and y members. If all voters are risk averse so that 'a' as a certainty is preferred to the lotteries on x and y and on x, a and y, then the member set is weak Pareto inoptimal under sincere member voting.

The ease with which this last result can be generated should act as a note of caution in respect of even numbered member constituencies. Trivially they are prone to all the problems described in the first two paragraphs above.

The interesting thing is that the above points to a serious difficulty when STV uses constituency - wide rather than jurisdiction - wide voting.

With 6 -member constituencies and two opposed preference parties forming their core and constituency based voting, parties with as much support as 8 percent of the popular vote can find themselves unrepresented, whilst in 100 seat jurisdiction-wide voting two such parties would get....8 seats each. If they are alternatives, having extreme parties talk in Legislatures may be better than suffering their street expressions of opinion.

All the above shows how STV just does not solve the difficulties I raised with voting systems in a series of papers in Theory and Decision and resolved in the context of majority decision making in Synthese a decade ago. Dr. I. D. A. MacIntyre (Do please re-assign this contribution to where you think it best belongs. (Obviously I think here))

will edit this in the near future if anyopne wants to provide suggestgions please do so. — Preceding unsigned comment added by Digmores (talk • contribs) 22:54, 10 January 2011 (UTC)

Referendum

Wasn't this what the referendum in the United Kingdom in May 2011 was about? ACEOREVIVED (talk) 23:43, 5 May 2011 (UTC)

Nope, the referendum is on whether or not to introduce the Alternative vote. It's similar to STV in that you rank candidates, but there's only one candidate per seat in AV. rpeh •T•C•E• 08:57, 6 May 2011 (UTC)

Article could really use some expansion on the Irish bits

This article could benefit a great deal from some review of the history of STV in the Republic of Ireland and Northern Ireland. The articles on Ireland, Gerrymandering, and The Troubles hint that STV might have come about as a solution to some of the serious problems caused by the old voting system that may have lead to the Irish civil rights movement, such as disenfranchisement via gerrymandering. I don't know anything about the history myself, however, but Wikipedia could certainly use some review of this topic (either here or in the related Irish articles). I'm going to tag this article for expansion. Scott Ritchie 06:53, 11 July 2005 (UTC)

The only slight problem with this is that STV there started with the Irish elections, 1921. It continued in the Free State/Eire/Republic despite the desire of leading politicians to end it. It was stopped in Northern Ireland (late 1920s?) because too many Labour or other non-sectarian representatives were being elected. Then came local gerrymandering (most notably in Derry/Londonderry). In Northern Ireland it came back into local government in 1973 where it has remained. It is a matter of opinion whether electoral areas are still carefully drawn: see for example Belfast [2] and compare unionist/loyalist Court with nationalist/republican Lower Falls next door. Meanwhile in the Republic the average number of winners in a constituency is below 4. --Henrygb 19:51, 11 July 2005 (UTC)

And even Derry/Londonderry carries a curious result in 1973: the United Loyalists and the Alliance Party between them won a majority of the first preferences, but the SDLP, Nationalists, and Republican Clubs won a majority of the seats under STV.[3] --Henrygb 00:17, 12 July 2005 (UTC)

How's that curious? Why would STV have preferences if that sort of thing was forbidden, or to be considered odd? It happens all the time, because STV isn't proportional on first preferences, but proportional on the balance of preferences. (A controversial case of it happened in the last Australian Senate election for Victoria, but that was due to an oddity of the Australian system allowing Parties to allocate preferences if the voter lets them—and it probably won't happen again.) Felix the Cassowary 02:13, 12 July 2005 (UTC)

Very late in the day comments, but here's my 2c. It probably looks odd to those familiar with the parties and the likely transfer patterns. STV in practice can throw up anomalies because although it's aiming for reasonable proportionality within a single constituency, multiple constituencies together can create anomalies. In an individual constituency there is always a near quota of votes left over at the end of the count so one party can get a smaller % of seats than its vote even with heavy internal transfers. See for instance this Tasmanian example (scroll down a bit to the diagrams) where the Liberals had 31.4% of first preferences, 31.7% on the final count and 20% of seats. Note also that Labor had 47.2% of first preferences, 50% of the final count and 60% of seats.

(Irish observers will be surprised to see the parties nominating so many candidates which would be electoral suicide in Irish elections but this is a consequence of the specific local rules - a voter must cast at least 5 preferences for their vote to be valid and candidates from the same party are grouped together. Running 5 candidates minimises the leakage as many voters will just vote for a single party - especially voters who wouldn't turn out under non-compulsory voting.)

If this effect is repeated in multiple constituencies then the anomalies start to stack up and magnify. Differential levels of turnout in difference constituencies can also create problems in comparing the overall popular vote to the seat result. One of the most notorious cases was the Maltese general election, 1981. In a country with a very strong two party system and very heavy internal rates of transfers one party got the most votes but another the most seats. This provoked an opposition boycott and a constitutional crisis which was resolved by introducing an emergency top-up method so in future elections the party with the most first preferences is guaranteed a majority of seats. This kicked in to the Nationalists' benefit in 1987 & 2008, and Labour's in 1996. Timrollpickering (talk) 11:05, 17 May 2011 (UTC)

As for the disparity between Court and Lower Falls this isn't due to the original drawing of those areas but rather a combination of the failure to redraw since and differential turnout. The current ward boundaries & DEA combinations were drawn up about twenty years ago and were never intended to last as long as they have; however restructuring of local government in Northern Ireland has been repeatedly deferred by a combination of the problems with getting lasting devolution and more recently the lack of political will (on multiple sides) for it. Consequently the ward boundaries have not yet been subject to a review despite population shifts that have seen Court decline comparative to Lower Falls. There's also the issue of differential turnout with the Falls traditionally having a much higher turnout than Court, but very few electoral systems take this into account and the boundaries are based on the registered electorate only. Timrollpickering (talk) 11:05, 17 May 2011 (UTC)

Hare-Clark

In what way is Hare-Clark different from conventional STV? 203.198.25.249 (talk) 12:39, 17 May 2011 (UTC)

Question about the example

Text from the article:

Round 2: Chocolate's surplus votes transfer in equal proportions to Strawberry and Sweets according to the Chocolate voters' second choice preferences. However, even with the transfer of this surplus no candidate has reached the quota. Therefore Pear, who has the fewest votes, is eliminated.

Why is it split equally? 2/3 of the party that wanted chocolate wanted strawberries. Shouldn't the vote be split 5/3 instead of 4/4?

An explanation would be appreciated. —Preceding unsigned comment added by 76.28.184.209 (talk) 23:12, 16 August 2010 (UTC)

I fixed it, whoever wrote "equal proportions" was simply wrong or confusing. The correct ratio is 4 and 2, since there were 8 strawberry and 4 sweet votes. I'll note the graphic was right the whole time. Scott Ritchie (talk) 20:11, 17 August 2010 (UTC)

The example was useful for me, but it didn't explain how things work when voters declare third or fourth choices. --Shay Guy (talk) 20:50, 27 June 2011 (UTC)

Voting every day, any time you want.

Vote at any time. On any issue. Every voter can vote themselves or delegate their vote to other voter - whether individual or conglomerate (party list etc). Revoke your vote delegation at any time, either permanently or for a specific issue where you want to vote yourself. The ballot is secret but your vote's use is open, for you.

Amount of nation-wide votes delegated to them gets a delegate a proportional weight to their vote they cast through the system themselves or delegate further. This changes constantly, thus the will of voters is constantly represented truly, with no distortions, at any point in time. No more waiting 4 years to overturn the corrupt government/president.

Governments are formed by a majority of votes. Anti-oscillation device is that to overturn a decision/gov't, 80% majority is needed in first 3 months; 65% after a year; 50% + 1 vote after 4 years.

Simple. Just. No vasted votes, no gov't cheating the voters. Instant true representation of the populace's will.

There isn't really a good reason why we should not do this. It's time we have just and true voting system already. WillNess (talk) 09:04, 15 August 2011 (UTC)

If you can find this in a published work, its own Wikipedia article is a possibility. —Quantling (talk | contribs) 20:04, 19 August 2011 (UTC)

Proxy voting#Delegated voting has something very similar, as it turns out - almost exactly the same, sans the "any time" part. I read about this somewhere on John McCarthy's web pages more than a few years ago, but when I later tried to find it there, I couldn't. Don't know exactly what happened.

There's also an idea of each voter earmarking their taxes, to decide in which way they'd be used. Some would "donate" to defense, others to welfare, that's how the budget would get formed automatically. Of course there too it must be made retractable, changeable at any point in time. The really massive use of computers/connectivity can really change the way we do democracy/self-governing.

The immediacy of votes I saw e.g. in one of Stanislaw Lem's humorous Star Diaries, where a President of some planet would get his approval ratings continuously - with a twist where when they'd get lower than a certain limit, he'd be automatically killed. No-one wanted to be a President there, naturally. :) WillNess (talk) 23:33, 19 August 2011 (UTC)

Transparency

An entire article on voting but not one metion of the word transparency??? Single Point of Failure - when the vote gets disconnected from the name, and the magnet for being hi-jacked by the status quo. 100% Transparency is wasted on Facebook where it doesn't matter but not applied where it really matters, voting, i.e. John Alfred Smith votes for Bill# 3402, Elizibeth Jane Thompson votes against Bill# 3402. — Preceding unsigned comment added by 95.132.180.154 (talk) 19:46, 23 February 2012 (UTC)

Are your hypothetical John and Elizabeth members of the legislature (at any level, including Congress) or are they "just" voters? If they are members of a legislature, their voting records should be available. I'm not sure that has anything to do with this article anyway. STV is generally a method by which the public elects officials, not a method for voting on legislation within a legislative body. (There may be exceptions to that, but it is generally true.) Neutron (talk) 21:15, 23 February 2012 (UTC)

Complexity/Transparency/Size 'gerrymandering'

It is a complex system - not to fill in a ballet paper of course but the counting and allocation of seats. Transfers, surplus votes, meek rules. Much the same results could be given in an open-list PR multi-member system.

Well, infact STV may be less proprotional than an open-list as people can vote between parties and the share of seats a party gets may differ from the share of first preference vote. Just look at Ireland's results under STV.

There is also the issues of size. A 3 seater, requires a party/candidate to get 25% of the vote, a 9 seater, 10%. The bigger the connstituency the more proportional - but more remote an MP is to constituents.

If you have STV then surely all seats have to be the same size, or very similar. You can't have some constituencies returning 3 MPs but others 8 or 9. That is 'size gerrymandering'(Coachtripfan (talk) 15:58, 17 July 2013 (UTC))

Not sure what your point is, but much of what you suggest is wrong. For example, you can have unequal constituency sizes, just as is common in other proportional and non-proportional systems that rely on multi-seat districts. Compared to list systems, STV is likely to be more proportional when district magnitude is low because it reduces wasted votes. List systems also can seem "opaque" when you get into the fine points of allocating seats. And so on. 96.255.160.105 (talk) 14:46, 19 July 2013 (UTC)

If constituency size varies from say 3 members to 9 - then that leads to a wide variation ie 25% or 10% needed to get elected. True, other multi-member systems may have this. It is less proportional than Mixed Member Systems that allow minor parties to get elected. A party with 10% spread throughout a country would not get elected under STV. It is also more complex than MMS in allocating seats. Regional Top Up

Then there's also the issue of 'above the line' voting - ie voting for a party ie essentailly closed list or 'below the line' ie voting for individual candidates. Australian State elections allow both. 95% or so vote for the party list! STV versus Open List complexities

(Coachtripfan (talk) 08:26, 15 August 2013 (UTC))

The New Example

I don't want to knock other people's hard work - but I don't find thenew example that easy to follow - IMNSHO the pictures are distracting and the star votes aren't that clear. I do think a simple numerical table suffices. Maybve I'm just being a miserable old git. Sorry. --Red Deathy 12:31, July 29, 2005 (UTC)

• I cleared out some of the unnecessary pictures, so it should look a little smoother now. Also, the stars were just an arbitrary character I picked - perhaps something larger and easier to color-differentiate would be better. I do think the added visibility of seeing which votes are transfering is helpful, however - I did this demonstration live with toothpicks, and found that people grasped it a lot better when there was a visible movement of certain toothpicks from one pile to another. The colored stars are an attempt to illustrate that sort of movement - something we can't get with numbers. Scott Ritchie 10:50, 30 July 2005 (UTC)
• My only complaint is that the stars in the example are too hard to distinguish. All of the reddish-orange colors look about the same. Even if they don't match the pictures, some of them should be changed to different colors like green, purple, or black. I also suggest a more visible vote marker, like ♦, except I'm not sure if entities like that show up in everybody's browser. Maybe even just a bold x. RSpeer 06:59, August 15, 2005 (UTC)
  • You're definately right about this. I went and changed the stars to x's and made the colors more visible. My concern, however, is with people who are color-blind in various ways - perhaps we can find one to comment on the current colors, altering them with dashes of red and blue if need be. Scott Ritchie 21:37, 16 August 2005 (UTC)
    • First impressions: The new one is hard to read even with the x's. At first I thought there was an error, because it said "20 voters" and I only saw 6. It does not come easily that the x's represent multiple voters with the same votes. If numbers after x's and colons were used, such as "4x:", "2x:", "8x:", etc., it would be more clear --207.70.164.210 11:17, 30 September 2005 (UTC)

• I think there's also a problem with it. Look at round 2. Chocolate's been elected and we're looking for 2 more winners. We only have 14 votes that are still not yet locked in. The droop quota at round two is floor((14/3)+1) == 5. This means that strawberry wins in round 2, followed by an elimination of pear in round 3, and a win for orange in round 4. Correct me if I'm wrong please. :) --06:58, 11 December 2009 (UTC) —Preceding unsigned comment added by 24.82.135.189 (talk)

• To avoid ambiguity in Round 1, can we please have a different number of surplus votes for chocolate compared to the quota? Currently they are both six which makes the explanation slightly ambiguous. And in Round 2 could we please have different values for number of surplus votes compared to total number of transferable votes (currently both are six). I know this is not easy to design and still have the equations give neat integer solutions. Sqgl (talk) 10:51, 26 August 2013 (UTC)

"number of surplus votes/total number of transferable votes (that have the second preference)*number of second preferences of the given candidate" wouldn't this be clearer as (number of second preferences of the given candidate * number of surplus votes) /total number of transferable votes [that have the second preference] — Preceding unsigned comment added by 81.108.89.77 (talk) 19:54, 4 June 2014 (UTC)

Is STV really proportional?

It depends on how many members are elected per constituency, if the number is maximum six (this is the situation in most countries adopt this method) it's much less proportional than pary-list or MMP which usually have at most 5% threshold. — Preceding unsigned comment added by 212.235.27.240 (talk) 14:25, 7 June 2012 (UTC)

Look at the Irish General Election of 2011. Fine Fail with 36% of the vote get 46% of the seats. Finna Fail with 17.5% of votes get 12% of seats. This seems to be for two main reasons - constituency sizes range from 3-5 members ie 25-16% needed to get elected - and nature of prefential cross-party voting.

Arguably there is a choice between smaller sized constituencies which are less proportional and larger constituencies which are more proportional - but less 'local'.

Check Wikipedia link to Irish 2011 election (link not working) (Coachtripfan (talk) 08:58, 15 August 2013 (UTC))

=STV is a proportional system. Any system becomes less proportional with a fewer number of seats, but that is separate from the system choice. So while not every application of STV is as proportional as what it could be, the system of STV itself provides proportional representation within the restrictions of that district magnitudeRRichie (talk) 15:12, 17 August 2013 (UTC)

If your focus is on party share of the vote - then STV may not be always proportional as later preferences from other candidates from other parties may be factor. Look at Sligo-North Leitrim in 2011, Sinn Fein got 13% and Finna Fail 22% yet SF got a seat and FF didn't. That is because SF attracted a lot more later preference votes from Indpendent candidates than FF. (Coachtripfan (talk) 10:49, 4 September 2013 (UTC))

@Coachtripfan on the contrary, your example shows that the majority of people preferred SF over FF. Focussing on the number of people who gave their *primary* vote is ignoring the way preferential voting works. The primary vote does not tell you which party people hate the least. Sqgl (talk) 13:42, 4 September 2013 (UTC)

Preferential voting takes account of people's second, third preference - but shouldn't the only prefence that matter be what people really want ie first prerence? 'the primary vote does not tell you which party peopel hate the least'. Exactly, voting should be about which party you positively want not which you hate least down to hate most. Most democracies with PR do not of course have STV. (Coachtripfan (talk) 20:14, 9 September 2013 (UTC))

I would say: the term proportional is not really applicable for preferential voting methods. I.e. STV is proportional to an interpretation of all the votes where you count the first preferences of some voters and the second preferences of some other voters and the third preferences of again other voters and so on. However, this is obviously something very different than the "normal" proportional, because most (arbitrary strange) apportionments can be described as proportional to such an interpretation of n-th preferences, where n varies with every voter. Thus, STV is just something different than a proportional system. It is not really "proportional" or "not proportional", but just different. However, this is my opinion. Most advocates of STV like to call it "proportional". If one of them is here: Could you please explain to what exactly the result of an STV election is proportional? (You're welcome to assume a unique not small (maybe 100 seat) constituency) Thx. Arno Nymus (talk) 23:16, 9 September 2013 (UTC)

As additional question: Does someone has a good and clear definition of a proportional system? I think the intuitive definition ("for each party/group the ratio of the seats is roughly the same as the ratio of the votes"; the article proportional representation does not give a clear definition, too) is not sufficient (What does "roughly the same" mean, where are the limits of that?) to really answer the question "Is the voting system X a proportional voting system?" Arno Nymus (talk) 18:38, 11 September 2013 (UTC)

Proportionality refers to like-minded voters electing seats in proportion to their share of the vote. By this definition, STV is fully proportional. The district magnitude is not tied to the system choice. STV is proportional within whatever district magnitude is provided.RRichie (talk) 18:25, 12 September 2013 (UTC)

• Ok, since the number of seats for a party/candidate have to be an integer (no "half seats"), there can't be a seat distribution that is (exactly) proportional for most vote distributions. Hence, the "roughly the same" above. That was the reason, why I asked what exactly the definition is. It's the question about the limits of "proportional". Maybe something like "Whenever the number of ideal claims of seats (i.e. number of votes divided by Hare quota) is an integer for ALL parties, the seat distribution must allocate these ideal claims. In all other cases, the result must lye within the bounds these ideal results pretend". But, ok, let's skip that definition question.
• Now, for one choice votes ("I vote for X!"), the meaning of "the seat distrubution is proportional to the vote distribution" is clear, since every vote belongs to a single party/candidate.
• On the other hand, for ranked choices votes ("My first preference is X, my second is Y, third is Z...") this is not so easy. So, my question is, to which candidate (or) party belongs a ranked choices vote when checking a ranked voting system for proportionality? Arno Nymus (talk) 21:18, 17 September 2013 (UTC)

There is no right answer to this. Some analysts persist to do based on first choices, but others make a good case for doing it based on the what you have after the transfers are done. The fact that the latter can be true shows just how STV delivers proportionality while giving voters a freer range of choice among candidates than most PR systems.RRichie (talk) 22:41, 19 September 2013 (UTC)

I thought about it and only taking what you have after the transfers are done in account would also give some terribly stupid system the predicate "proportional" (when the transfers are terribly stupid). E. g. propose the "most hated voting system" (MHV) that makes STV on the reversed votes, i.e. it starts with the last choices of each vote and redistributes to the second to last choice if necessary and so on. The result would be as proportionally to "what you have after the (MHV) transfers" as STV is proportionally to "what you have after the (STV) transfers" (the only difference would be that the MHV transfers are much more stupid than the STV ones). I don't think that MHV could be considered "proportional", but I also can't think of a definition for which STV is proportional and MHV is not. Any ideas on this? Arno Nymus (talk) 01:02, 24 September 2013 (UTC)

Suppose N is the number of voters. Suppose M is the number of seats. Suppose there are at least M candidates. Suppose the Droop quota is given by N/(M+1). Then Droop proportionality says that, for every natural number m with 0 < m ≤ M and for every set of (at least m) candidates, the following must be true:

If strictly more than m Droop quotas of voters strictly prefer every candidate of this set to every other candidate, then at least m candidates of this set must be elected.

The single transferable vote is proportional in so far as it satisfies Droop proportionality. Markus Schulze 12:51, 25 September 2013 (UTC)

In this extent AV or even FPTP are proportional too. — Preceding unsigned comment added by 212.235.27.240 (talk) 11:08, 24 February 2014 (UTC)

In single-winner elections, Droop proportionality is identical to mutual majority. AV satisfies this criterion, but FPTP doesn't. Markus Schulze 20:53, 4 October 2014 (UTC)

Small example to show proportionality

Here's a real and small example STV poll I created awhile ago, 20 voters, ranked ballot on favorite fruit, and counted for 3 winners, counting two ways as: 25% droop quota or 33.33% Hare quota. I allowed tied rankings, which were counted as an equally divided vote. The counts below are rounded to 0.1 votes. To simplify the summary below, I hid about 8 candidates who were removed by bottom-up elimination and never gained more than 1.0 votes. I also eliminated a 3-way tie with 1.2 votes in one step in the first round show.

In the final round of the Droop count, there was a last-place tie between Banana and Pineapple, requiring a tie-breaking method. The Hare quota worked out better to avoid ties. In the Droop quota, 74% of voters picked one of the 3-winners (25%,25%,24% support each). In the Hare quota, 82.5% of the voters picked one of the 3 winners (17.5%,30%,32.5% support each). You could say Droop "wasted" 24% of the votes, and Hare wasted 17.5%, on the 4th candidate who can't win.

Anyway, this example shows me STV found the 3 strong candidates among mutually exclusive voter bases, although not necessarily the "strongest candidates". Proportionality assures a WIDE set of voters are represented, and any voter who is a part of a united quota faction is guaranteed at least one winner. I like the Hare quota better, but political elections probably prefer the Droop quota which gives more power to bloc ranking by larger coalitions. Tom Ruen (talk) 00:29, 10 September 2013 (UTC)

Droop quota: with 20 votes, 3 seats, (25%=5 votes)

1. Cherry=4.0 Banana=3.2 Apple=3.0 Mango=2.2 Strawberry=2.2 Pineapple=2.0 (Apricot=1.2 Nectarine=1.2 Raspberry=1.2)
   • Eliminate set {Apricot, Nectarine, Raspberry}
2. Strawberry=4.3 Cherry=4.0 Banana=3.3 Apple=3.0 Pineapple=3.0 (Mango=2.3)
   • Eliminate Mango
3. Strawberry=5.0 Pineapple=4.3 Cherry=4.0 Banana=3.7 (Apple=3.0)
   • Eliminate Apple
4. Cherry=5.0 Strawberry=5.0 (Banana=4.8 Pineapple=4.8) [NOTA=0.3]
   • Winners: Cherry (25% Droop), Strawberry (25% Droop), (Banana or Pineapple)

Hare quota: with 20 votes, 3 seats, (33.33%=6.7 votes)

1. Cherry=4.0 Banana=3.2 Apple=3.0 Mango=2.2 Strawberry=2.2 Pineapple=2.0 (Apricot=1.2 Nectarine=1.2 Raspberry=1.2)
   • Eliminate set {Apricot, Nectarine, Raspberry}
2. Strawberry=4.3 Cherry=4.0 Banana=3.3 Apple=3.0 Pineapple=3.0 (Mango=2.3)
   • Eliminate Mango
3. Strawberry=5.5 Cherry=4.0 Pineapple=4.0 Banana=3.5 (Apple=3.0)
   • Eliminate Apple
4. Strawberry=6.5 (32.5%), Cherry=6.0 (30%), Pineapple=4.0 (20%), (Banana=3.5 (17.5%))
   • Eliminate Banana
   • Winners: Cherry=6.7 (33.3%) Strawberry=6.7 (33.3%) Pineapple=6.1 (30.5%) [NOTA=0.5 (2.5%)]

Tom Ruen (talk) 00:29, 10 September 2013 (UTC)

Thx for presenting an example so that I can try to understand. Without the ballots, it's a little bit difficult to see the values. Also the redistribution of votes that exceed the quote is here somehow hidden. And I think that allowing equally divided votes doesn't make it easier to see. But I'll try.

However, it seems to me that at the last (or prelast) step (of hare) there are the following votes:

• cherry: 6 (4 first preferences and 2 second preferences)
• strawberry: 6.5 (2 first, 4 second, 0.5 from an equally divided 1th through 6th-preference)
• Pineapple: 4 (2 first, 2 second)

So, are these votes (8 first preferences and 8 second preferences and 0.5 equally divided 1h through 6th) to which the seats are proportional? Arno Nymus (talk) 20:44, 10 September 2013 (UTC)