|WikiProject Politics||(Rated C-class, Mid-importance)|
- 1 Discussion
- 2 Requested move
- 3 What happens if one candidate does receive an absolute majority? & other clarifications
- 4 8/9/07 changes
- 5 Monotonicity?
- 6 In France
- 7 Page name is unclear
- 8 Removed section
- 9 Condorcet winner and runoffs
- 10 Weimar Republic --> German Empire
A 45% B 40% C 15%
ELECTION TWO Features candidates A and B in a "Run-off"
Advantages of a Run-Off Election
In a run-off election, the possibility of a "third party spoiler" is eliminated because no voter is forced to cast a vote for a candidate they do not favor. Here is an example:
If the US Presidential election were a run-off system, you might see this scenario. Voter 1 favors Ralph Nader. Voter 2 favors George W. Bush. Voter 3 favors John Kerry. In a regular election, assuming every voter votes their conscience, you see this result: Bush 1 Kerry 1 Nader 1 If voters vote strategically you might see this result: Voter 1 - Kerry Voter 2 - Bush Voter 3 - Kerry
Voter 3 has voted strategically, but in the process has voted against his preference. This is the heart of the spoiler issue in an electoral system. Voters who vote for a third candidate that is weaker than the top two, often prefer one of the top two candidates to the other. If they vote for the third candidate, their vote may be "wasted" because they could vote a candidate they prefer, and actually cause the candidate they least prefer to be elected by doing so. If they vote strategically, they will assure a candidate that they prefer is elected, but not the one they most prefer, which is the assumption in an electoral system.
A run-off system compensates by allowing voters to vote their first preference in the first election. In the second election, only the top two candidates are allowed to participate, so the possibility of a third party spoiler has been eliminated.
Runoff voting → Two-round system – I know, I know, I know; all varieties of English are equally valid. However, the paragraph about terminology -- the last two sentences, complete with self-reference, are the smoking gun -- indicates that we need a move with either a redirect dab at its new location or a dab at the current location (probably unnecessary). Also see the discussion over longcase clock, a case of the inverse: the generic but uncommon name (longcase) instead of the common but non-inclusive one (grandfather). Moulder 06:56, 24 July 2006 (UTC)
My support is for the disambig option now despite previously thinking it unnecessary. (Or maybe a generic article about the concept of "runoff voting" and how it developed, if applicable.) This is mainly to distinguish between the exhaustive system and instant-runoff voting. The trickiness of this question is why I decided to be conservative instead of bold. Moulder 22:25, 24 July 2006 (UTC)
(Re: Vegaswikian) An article about generic runoff voting, i.e. comparing and contrasting the different systems and how the term runoff applies to them, should suffice; the initial paragraph can mention the fact that the US calls two-round voting runoff voting, and then each of the systems that employs runoffs can get a mention. Moulder 22:18, 30 July 2006 (UTC)
What happens if one candidate does receive an absolute majority? & other clarifications
What happens if one candidate does receive an absolute majority? Does he/she/it win without another round of voting? If so, that needs to be stated.
What happens if candidates draw, or if you had an election which went something like this: A - 25%, B - 24%, C - 23%, D, C, E, F... - 1% each? Would C still be eliminated in the second round?
Njál 16:17, 29 April 2007 (UTC)
Most runoff systems currently in use have a rule that a candidate receiving an absolute majority wins without needing a runoff. This is not, however, required by the method.
What happens in the case of an exact tie for second place has to be settled by something external to this method. In the near-tie situation you mentioned, C would indeed be eliminated in the second round, leaving an A v B runoff. (This actually happened in the 2002 Presidential election in France, where Jean-Marie Le Pen finished in second place, less than a percentage point ahead of Lionel Jospin.)Jack Rudd (talk) 00:58, 29 May 2008 (UTC)
I assume that runoff-voting satisfies the monotonicity criterion? Intuitively, it seems obvious that it would, but is there proof? (If the proof is trivial then in my opinion, that's all the more reason to add it to the article.) If, however, it does not in fact satisfy monotonicity, then can somebody please come up with a counter-example? 220.127.116.11 (talk) 02:36, 17 May 2008 (UTC)
It doesn't satisfy it. Here's a counter-example:
39 prefer Edwards to Roemer to Duke
22 prefer Roemer to Edwards to Duke
8 prefer Roemer to Duke to Edwards
31 prefer Duke to Roemer to Edwards
So 39 vote Edwards, 31 Duke and 30 Roemer. The top two vote-getters are Edwards and Duke, who advance to a runoff, which Edwards wins.
Now make two of the Duke voters switch to Edwards. Then Edwards gets 41 votes, Roemer 30 and Duke 29. Edwards and Roemer advance to a runoff, which Roemer wins.
This same example, incidentally, shows that runoff voting fails the favourite-betrayal criterion, although the more usual favourite-betrayal would be for voters to switch from Duke to Roemer in the first round.
Jack Rudd (talk) 00:17, 29 May 2008 (UTC)
Have changed the reference to every French election. Versions of the "two-round system" as decribed here are used, with differing rules, in presidential, legislative (National Assembly), and cantonal (Département general council) elections. For regional elections, however, the system used is
- scrutin de liste selon un système mixte combinant les règles des scrutins majoritaire et proportionnel, en un ou deux tours, et sans panachage,
while for municipal elections there are two systems depending upon the size of the commune:
- scrutin de liste à deux tours avec représentation proportionnelle
- scrutin majoritaire plurinominal à deux tours avec panachage.
The last two systems provide for two rounds, but they do not fit the definition of "two-round system" as given in the lead-in to this article, since neither of them is "used to elect a single winner".
The present article, as it stands, covers only one particular variety of two-round voting: what the French, more precisely, call the
Page name is unclear
Looking at the name of this page, it can only possibly make sense once you are at it. Standalone it lacks clarity or specificity that it relates to VOTING. I would suggest that for clarity that the page should be named Two-round voting system Let me say that I know that this poses some difficulties due to the underlying nature of redirects, however, IMNSHO it needs to be done. -- billinghurst (talk) 03:10, 15 October 2008 (UTC)
- The article first sentence makes it clear enough. Two-round system is the standard name for it. Any links to this article can give a longer description if desired, like Two round voting system. Tom Ruen (talk) 03:52, 15 October 2008 (UTC)
I removed this section. There's no citations, and I have no evidence these systems are used, or even if they exist, whether they belong under a "two-round system" article. Tom Ruen (talk) 00:54, 11 November 2009 (UTC)
Viability Threshold Voting
In some types of runoffs, the initial round of voting is used to determine viable candidates, which can possibly total more than just two.
For example, some precincts in American runoff elections count any candidate receiving at least 15% of the total vote a "viable" candidate, and their name will appear on the subsequent general election ballot (the general election only requires a plurality to win).
In other versions, candidates must receive at least the average percentage for major candidates. For example, in a primary with 5 candidates, the average threshold would be 20%. Each candidate receiving at least 20% of the vote progresses to the second round. Another option in a reiterative count is to implement a bulk exclusion rule where candidates whose sum of all votes are less votes then then candidate(s) under consideration are excluded at the same time.
Condorcet winner and runoffs
Just to clear on why I'm reverting the recent change, if you have forced elimination in any round based on votes cast, you can eliminate the Condorcet winner. I don't see how you can say otherwise.... I glanced at the paper, but can you provide a simple example? Suppose you have three candidates, A, B and C. C is the Condorcet winner, but trails both A and B in core support and finishes third. How can you say the Condorcet winner will win if C is eliminated based on the first round results? RRichie (talk) 20:13, 25 June 2011 (UTC)
- The complex model involves a game-theory equilibrium. In such an equilibrium, every voter votes in such a way that no voter has anything to gain by voting differently. Their actual votes may thus appear inconsistent with their internal preferences. The paper is saying that under that kind of equilibrium, and a limited number of candidates, the CW always wins. If there is a Condorcet winner (according to the voters' internal preferences), then the game-theoretical (robust political, which is basically trembling hand plus coordination) equilibrium is where voters vote in such a manner that the Condorcet winner is elected. (This can be proved through backward induction.)
- In that way, it's like Approval. Approval voting doesn't let people specify relative rank of the candidate beyond simple approve/disapprove. Therefore, you would think Approval can't find a Condorcet winner since it has no way of knowing when a candidate is a Condorcet winner. But the game-theory equilibrium model mentioned in Approval voting is alike what I just said: it's a situation where no voter has anything to gain by voting differently. When voters vote according to that equilibrium, Approval picks the Condorcet winner (according to the voters' internal preferences) when there is one, and so does the two-round system with three candidates.
- As for your example, if you have:
- 46 voters prefer A to C to B
- 44 voters prefer B to C to A
- 10 voters prefer C to A to B
- then the equilibrium has the 44-voter group vote for C in the first round. If they vote for B, then no matter what the C-first group does (besides voting for their most disliked candidate), A wins; but if the 44-voter group votes for C, the C-first group has no reason to vote for any other candidate than C and so C wins outright.
- This may be intellectually interesting, but it has no bearing on what would happen in real-life two-round runoff elections, which is what your point seemingly is trying to address in the article itself. Candidate C would lose and not advance to the runoff. Period. .. Now of there were no forced elimination and a candidate had to win 50% to win, I suspect A and B voters would settle on C. That's the point of my edit. RRichie (talk) 05:01, 26 June 2011 (UTC)
- The model is of what would happen if voters knew the others' preferences. The more they know about the preferences of each other, the closer it gets. Besides, this reasoning, and this explanation of models, is accepted for Approval voting (as one can see in the article on AV), and thus I don't see why one should accept it there but exclude it here.
- I would say it doesn't belong in the approval voting article either, at least not mixed into the main text as if this were a real way to think about the system as it is. Furthermore, even if you know about the preferences of other voters, you need to rely on them all acting sincerely, when there's no guarantee of this being the case..Fankly, this seems like so much gibberish when applied to thinking about how these systems might work if tried. If anything, it should be a separate paragraph, if used at all -- both here and in the approval voting article, where editors have allowed it to be treated on an equal basis with thinking about the system as it really would be if implemented. Not good. RRichie (talk) 12:34, 26 June 2011 (UTC)
- It's not just any modification of the ballots that would elect a given candidate (and thus require honesty). The equilibrium is stable in that there's no incentive for any player, or (for the robust political equilibrium) for any coalition of players to vote otherwise, except if that would lead them to another equilibrium. Therefore, it can arise from selfish cooperation, so insincerity in the act of voting doesn't really enter into it. To deviate from the equilibrium, once reached, would only harm the voter/s.
- What I wrote in my edit is factually right (the paper shows that the equilibrium exists) and there's precedent from the approval voting article. Your edit, though, does not follow from the citation. You say that the game-theory equilibrium is unrealistic and the conditions required to reach it wouldn't happen in an ordinary election. Would you accept a compromise that keeps my original edit, but adds text to the effect that the game-theory result requires idealized conditions and therefore provides an upper bound on what can be achieved with coordination or knowledge of other voters' preferences, but that it may not hold for real elections, particularly not large ones?
That seems sensible, although I must confess that this whole kind of theoretical discussion doesn't seem to be included in the midst of discussion of criteria that affect elections as they are. In that way the whole approval voting presentation is very misleading too and really should be changed. Saying approval voting meets all these criteria under conditions that will never happen is quite questionable to me, and saying that runoffs will always elect the Condorcet winner under certain conditions also is quite confusing, as those conditions would never exist. RRichie (talk) 18:40, 27 June 2011 (UTC)
Weimar Republic --> German Empire
I corrected that. A two-round-system was used in the Kaiserreich, while the Weimar Republic used a method of proportional representation.