In case of tie

What if an election is tied? Apparently it is not that unusual, at least in contests with relatively small electorates. FiveThirtyEight tells of some recent US cases. It also cites a study by Casey B. Mulligan and Charles G. Hunter that says:

between 1898 and 1992 there were no tied elections for [US] federal office and only one (New York’s 36th District in 1910) that was decided by one vote. Only twice has a general election for a state legislative seat resulted in a tie: a Rhode Island state Senate race in 1978 and a 1980 New Mexico state House race. The former was decided by a special election, the latter by a coin flip.

I understand by way of Yusaku Horiuchi that ties for the last seat are fairly common in Japanese municipal elections, where SNTV is used.

What about other countries that readers may have familiarity with? And do most jurisdictions have clear rules of what to do in even of a tie? The FiveThirtyEight post refers to some cases where there were no such rules, and the council or other body had to make up a procedure.

50 thoughts on “In case of tie

  1. I recall that some western State uses poker or some other chance card game to resolve ties. Many of countries award a tied contest to the older candidate.

      • I’ve just realised that my reference to Nunawading (here and at the thread linked) is probably very obscure, not least because Wikipedia has no entry on it, and even its entry on the 1985 Victorian State election gives results only for the Legislative Assembly and not the Council.
        Basically: before PR was introduced for the Victorian Upper House (starting 2006), Labor only won a majority at one election, 2002 when it also won a landslide in the lower house.
        Each of the 22 Council “provinces” (a) was divided into four single-seat Assembly districts and (b) elected one of its two MLCs in rotation. Even when electoral boundaries for both Houses were reformed in the 1970s to make them equal within a 10% margin, Labor was still systematically disadvantaged because its vote was more concentrated and the Council constituencies were four times larger. I believe it was common that, for example, a Liberal or National MLC might cover the same area as only two or three conservative MLAs, whereas most Labor MLCs represented “provinces” with a 4-0 Assembly sweep (and a high popular vote for Labor in the 60% range).
        The closest Labor ever came to winning a Council majority before 2006 was 1985, when the Cain government was re-elected. Labor won 22 Council seats, and the conservatives won 21, with clear majorities after preferences. The 44th seat was in Nunawading Province, where the two main candidates were tied: their total votes were, from memory, in the forty thousands. The Returning Officer had a statutory casting vote and decided to exercise it after voluntarily drawing lots for guidance (ie, a random draw was not the statutorily prescribed procedure for breaking ties). She picked the Labor candidate so the ALP had a 23-21 Council majority, the first time in Victorian history. However, this lasted all of four or five weeks because the Liberal candidate found enough disputed votes to persuade a court to void the result and order a new election the following month. The Liberal won the second round so the Council was formally deadlocked 22-22. Since Labor was the largest party, it was “forced” [*] to provide a presiding officer who could only vote to break a tie, and since the conservatives now had 23 seats to 22, ties rarely arose.

        [*]I confess I find this idea of “forcing” the leading party to designate one of its members as Speaker, Chair, or President – who then loses their voting rights because there is an odd number of members remaining, making “tied votes” unlikely to occur – quite bizarre. It gives a de facto working majority to the second largest party or coalition. Tasmania had the same problem too, when it used STV-PR with six-seat constituencies before 1956, especially after the rise of organised parties. It would seem to make much more sense to advantage – not to penalise – the largest single party when there is a 50-50 tie, the way the US Senate does with the Veep’s casting vote. If the parties had equal numbers and were deadlocked after repeated ballots attempting to fill the Chair, the Head of State should appoint one of the two candidates (subject of course to reversal by later majority vote of the chamber).

  2. Don’t UK local elections have ties relatively commonly? These ties are decided by random events that are chosen by the local returning officer, like choosing an envelope or drawing straws.

    Australian upper house elections have plenty of ties, mostly at the start of the count when the pointless second candidates of the minor parties are excluded. These ties are usually decided by a random event.

    • Random selection is the rule for all UK elections, but is most common in local elections. There’s at least one most years.

      I’ve seen it done by placing one ballot paper favouring each candidate into a policeman’s helmet and the returning officer drawing one and by the flip of a coin. Many other randomizers have been used; those are just the ones I’ve been personally present for.

  3. New York State has separate election rules for Villages (as opposed to Towns and Cities) and Section 15-126 of Villlage Election Law is copied below. See section 2 for the two ways to break a tie in a village election. One of them is of course a run-off election but the other is for a judge to choose a winner ‘by lot’, if both parties agree to such a scenario!

    1. The inspectors of election of each election district shall, immediately upon the closing of the polls at each annual or biennial election, proceed to canvass the votes cast thereat and shall complete such canvass without adjournment. They shall, before nine o’clock in the forenoon of the following day, file with the village clerk their certificates setting forth the holding of the election, the total number of votes cast for each office, the number of votes cast for each person for such office, the total number of votes cast upon each proposition voted upon, and the number cast for and against it. They shall also deliver to the village clerk at the same time and place all ballot boxes, if there be such, and all unused supplies and the American flag furnished for use in the polling place. If the village contains more than one election district, the board of trustees of such village shall meet at its usual place of meeting not later than eight o’clock in the evening of the day after the election. The village clerk shall produce at such meeting the returns of the inspectors of election, at which time the board of trustees shall canvass such returns and file in the office of the village clerk a certificate declaring the result. 2. a. The person or persons eligible and receiving the highest number of votes for an office shall be elected thereto. b. In the event that more eligible persons than the number remaining to be elected receive for the same office or offices an equal number of votes, the board of trustees shall conduct a run-off election. Such run-off election shall be held on the first Tuesday at least ten days after the final certification of such tie result, subject to the provisions of paragraph b of subdivision three of section 15-104 of this article, provided, however, that the only persons who shall be deemed nominated shall be those persons who shall have received such equal number of votes. The order of the candidates names on the ballot shall be determined by a drawing conducted by a village clerk, in the presence of all those persons who received such equal number of votes, or a representative of such persons. c. Such run-off election may be waived and the selection made by lot as otherwise provided by this section if each person who shall have received such equal number of votes shall file with the village clerk, no later than two days after such final certification of such tie result, a written notice of consent that such selection be made by lot. d. If a waiver of such run-off election shall occur, the village clerk, no later than two days after receiving written notice of consent that such selection be made by lot, shall certify such facts in writing to any supreme court justice within the judicial district in which such village is located and shall within three days summon the candidates before him or her and he or she shall by lot determine which of them shall be elected. – See more at: http://codes.lp.findlaw.com/nycode/ELN/15/15-126#sthash.djtry1j4.dpuf

  4. I also noticed something in the article that I thought was funny; Are there such a thing as local dog catcher elections? I wonder who would elect such a person and who would run against someone wanting the position of dog catcher; I am going to catch more dogs than you.

    • It was once quite common to elect a town dog catcher, but as far as I know, only one town continues to do so: Duxbury, Vermont. The election is held at the annual town meeting. The last candidate was unopposed and won nearly every election–only his wife voted in contrary (in what appears to be an obvious joke).

      http://t.co/vzpPzoftV4

      • I’d heard the phrase a lot but had gathered it’s a metaphor for one of many low-level local offices (“s/he couldn’t get elected dog catcher”). Sort of like Patton Oswalt as Constable Bob in “Justified” s4.
        Having said that, I do notice when one reads the biography of one or another US politician that often their first elective office is something like “recorder of deeds for XYZ parish” or (rather startlingly for Senators, Governors, etc) “judge of XYZ local court”. I can’t think of [m]any democracies outside the US that elect judicial officials (does Japan still have retention polls?), or indeed executive officials other than a president/ governor/ mayor and possibly a vice-president/ governor/ mayor, or indeed any officials other than legislators. Did some of the Eastern Bloc states use to hold (tightly managed ) “long ballot” elections during the Communist era?

  5. You have a choice of default rules. Most Australian legislation orders the DRO (local returning officer or equivalent) not to vote except in case of a tie. Poker games in New Mexico may seem random, but that is their advantage.

    At one time Australian federal ballot papers listed candidates in alphabetical order and the first half of the alphabet was somewhat over-represented in the parliament. The order is now determined by lot.

    There is no obvious reason, certainly no fair reason, to have a default rule that favours incumbents, older candidates, or candidates with red hair.

    • Favouring the older candidate seems fair to me, as the younger candidate is more likely to have another opportunity.

      • I’d argue for favouring incumbents over challengers, and older over younger candidates, in cases of repeated deadlocks, using the same logic as Speaker Denison’s rule http://tinyurl.com/ppeotyg: continue the status quo unless there’s a clear majority (or plurality) favouring some definite alternative.
        Like the rule in RISK tm that the defender wins a tied dice roll.

      • Resolving legislative ties is different from resolving electoral ties. Legislatures are continuing bodies and the electorate is not.

        The ‘Denison’ rules, which actually incorporate precedents set by other speakers in the UK, may just, and only just, be reasonable in a mid-nineteenth century legislative assembly.

        in Exercise of the casting vote of the chair The NSW parliamentary library notes:

        In summary, the principles are:
        The Chair should always vote for further discussion where this is possible (Speaker Addington 1796);
        Where no further discussion is possible, decisions should not be taken except by a majority (Speaker Denison 1867); and
        A casting vote on an amendment to a bill should leave the bill in its existing form (Speaker Denison 1860). Note: in the House of Commons bills as amended in Committee may then be considered in the House.

        Bearing in mind that the abovementioned principles were first conceived in the House of Commons between 1796 and 1867 and that, because of the size of the House the Speaker has rarely had to exercise a casting vote, the relevance of these principles to the New South Wales Legislative Assembly today could be questioned.</blockquote.

        It may be more relevant to note that New Zealand abolished the casting vote when MMP was introduced and their speaker now exercises a deliberative vote to ensure that the lectors who voted for these speaker’s party get their representation int he house. The Australian senate has long followed the same practice.

        The ‘Denison’ rules are about continuing discussion. That is not possible with an election which must be declared and is then finished.

        A rule that favours any social group is going to ensure that group is over-represented. Elections are about electors, not whether the candidate has another chance to enter the parliament before the Grim Reaper bangs on their door. It is not immediately obvious why the best way to resolve an electoral tie is to entrench a legislature of longserving gerontocrats.

    • Before the 1980s, most Australian federal and State ballot papers listed candidates alphabetically by surname. (For the Senate, an alphabetical weight was assigned to each team group by giving each candidate a number 1 to 26 depending on their surname, adding the numbers, and averaging them for each team. Drawing lots would have been simpler!).
      Now, all Australian jurisdictions either draw lots or (the ACT and both Tasmanian houses) rotate the positions within each column.
      Coincidentally, a decade ago the six State premiers were Bacon, Beattie, Bracks, Carpenter, and Carr, with Mr Rann (not to be confused with mr Wran) the sole representative of the last 23 letters of the alphabet. .Made COAG conferences look like the Polish Sejm.

      • If ties do occur, they are – as Alan notes – more easily resolved under STV-AV because you can look back one count at a time until you find who was ahead when the tied candidates were not yet tied. If, however, they were tied at every preceding count (or if the tie arises on first preferences), STV-AV still needs a tie-breaker.

        However, countervailing that (and as an STV shill, it pains me to acknowledge this, but then as an STV shill I’m even more pained by dishonesty in electoral system debates) is a point made by critics that ties are more likely to arise under STV-AV. If the votes are, say, 40, 28, 16 and 16 among four candidates, then no tie has arisen if you’re using FPTP. However, a tie does arise if you’re using AV because you need to know which of the two candidates on 16 to exclude first. In this case, any of the four could conceivably win and before you crush a butterfly on a wheel while hunting dinosaurs you need to know what might happen if it were to flap its wings (so to speak).

        Compare the cottage industry of psephologists pointing out how Australian Senate results – one or even two out of six seats per State – could often have been changed by a shift of just a dozen votes here or there between the Sex While Fishing Party and the Palmer One Family United candidate on the first or second count. Multiply that by an order of magnitude if the result really were random.

        In lieu of “favour incumbents over challenger candidates and, subject to that, favour older over younger candidates”, I could accept a different form of random selection – whichever of the tied candidates had drawn the lower position on the ballot-paper, should win. Assuming he or she would have drawn fewer donkey-votes, it is reasonable to assume that he or she attracted more genuine, thought-out, considered, non-donkey [fox? Hedgehog?] votes. We don’t of course know HOW many, but if the numbers of formal legally-marked ballots are equal then we don’t need to quantify.

        Under this system, there’s still a random draw, but it would have been made in the relative calmness of the pre-polling stage, behind a sort of Rawlsian veil of ignorance that should mitigate some of the partisan heat. (No candidate wants to say BEFORE the polls open “I might defeat my opponent by only a handful of votes, or even tie in a dead heat.” Instead they want to talk up their chances.)

        Not sure however how to adapt it to rotated ballot positions as per ACT/ Tasmanian Hare-Clark. I presume electoral officials keep track of how many ballots of each version (ABC, BCA, CBA, etc) are distributed and returned unspoiled, and these are unlikely to be exactly equal. My recollection of various US State provisions for rotating ballots, quoted by Hoag and Hallett in 1920, is that they required that “as far as practicable” equal numbers from each batch be distributed to voters, but add that “inadvertent departures from this requirement shall not invalidate an election”. Tasmania’s ELECTORAL ACT 2004 now requires that Legislative Assembly ballot-papers “are to be collated in an approved manner so that each ballot-paper is followed by a ballot-paper having the ballot-paper names in a different order [within each column]” (s 97(10) http://tinyurl.com/pbcza8z – identical provision, minus the words in square brackets, for the Legislative Council in s 98(3) http://tinyurl.com/orzv9vq. The ACT Electoral Act 1992, Second Schedule http://tinyurl.com/orxd8yt likewise requires that “The [Electoral] Commissioner shall ensure that ballot-papers distributed to a polling place for the purposes of an election are so collated that the ballot-paper immediately following another ballot-paper in the issue is in a form different from that of the other ballot-paper” [clause 3] and “The OIC [officer in charge] of a polling place shall ensure, so far as practicable, that the form of a ballot-paper issued by an officer to a person claiming to vote at that place is different from that of the previous ballot-paper so issued by the officer” [clause 4]. I suppose you could be really strict about this and legislate to the effect that, if there are N versions of the ballot-paper, then any consecutive N voters (not just any two) must receive different versions of the ballot-paper.

      • Here’s how Kevin Bacon explains it (contrasting Tasmania’s Hare-Clark STV system with what I call the “mainland upper houses” or “closed-list” version of STV):

        “… Firstly, because all preferences in Hare-Clark are controlled by the voter, the situation that arose in WA in 2013 (in which the exclusion order of very minor candidates determined two positions) does not apply. In Hare-Clark candidates cannot get elected on preference snowballs from very tiny vote shares. Secondly, because of the way Hare-Clark does surpluses (by last bundle rather than by the elected candidate’s whole vote history), Hare-Clark doesn’t have tipping points based around the exact time of a successful candidate’s election..”

        (“Denison Ballot Papers Damage Report Released”, Tuesday, June 10, 2014 at http://www.tinyurl.com/mss3uwn)

      • Aaaarghhh. Kevin BONHAM. Mixed the good doctor up with the late Tasmanian Premier Jim Bacon.

      • Tom, thanks for giving me this opportunity to clear some things up for myself about STV!
        “In Hare-Clark candidates cannot get elected on preference snowballs from very tiny vote shares” Is this an inherent property or a tendency?
        “the way Hare-Clark does surpluses (by last bundle rather than by the elected candidate’s whole vote history)” Oh, I had hitherto been under the impression that all Australian jurisdictions did the full fractional allocation of preferences. Either way, what exactly is the last bundle? The last votes to be added to a candidate’s tally from preferences?
        Hare-Clark doesn’t have tipping points based around the exact time of a successful candidate’s election..” Could you elaborate on what this means exactly?

      • Hi JD,

        (We here can all grasp how STV counting goes, but working out how to make replies follow the appropriate comments is… what’s that word? – “complicated”

        (1) Re preference snowballs and (3) tipping points… I take Kevin Bonham to be meaning that this doesn’t happen mainly because in both of Australia’s “Hare-Clark” versions of STV, preferences are mainly optional – Tasmanian voters are compelled, and ACT voters are urged, to number at least as many candidates as there are seats. (His post explains this at length, better than I could).

        My anecdotal understanding is that, in most elections in these jurisdictions, there isn’t a concerted effort by party leaders to urge voters to number more candidates than the minimum required, so most voters stay within their first-choice party’s team and stop there. This applies even when, say, Labor and the Greens may have governed in a coalition or at least a working alliance during the outgoing parliament. (Ironically, I think there WAS a strong flow of preferences between Greens and ALP in the 1989 Tasmanian election, the first where the Greens won a balance of power, ie, there hadn’t been a previous coalition/ alliance).

        Under Hare-Clark PR, unlike AV in single-member seats, exhaustion of cross-party votes doesn’t swing a huge number of seats from one column to another. (Contrast the 1998, 2001 and 2004 Queensland State elections. At federal level, and also in Victoria which is now the only State with full-preferential AV and an ongoing coalition between the Liberal and National Parties, the Coalition partners make great efforts to encourage voters to preference each other’s’ candidates. They used to do this in NSW and Queensland also, but optional preferential voting made this harder, and eventually they gave up – in NSW by avoiding almost all “three cornered contests” on each other’s electoral turf, in Queensland by merging into a single party).

        On the other hand, since Hare-Clark is a PR system (and is used in Australia’s two smallest lower houses), a small number of seats may make a big difference to who forms government.

        (2) Hare-Clark (in both cases, I believe) transfers only the last parcel received by the elected candidate, the one that pushed him or her over the quota. It only transfers all his or her votes if the candidate reached quota on first preferences. Thus there are fewer ballot-papers involved and they are all reduced in value identically. We call this “last-parcel Gregory transfer”, after the mathematician who first proposed that (eg) if the quota is 100 votes and a candidate has 120 transferable first preference votes, the surplus should be transferred by reducing each such ballot from a value of 1.0000 to a value of 0.16667.

        By contrast, the Senate system uses “inclusive Gregory transfers”, where all ballots credited to the elected candidate are transferred. On paper this sounds good but it causes distortions because they are all multiplied by the same fractional value when transferred. This ignores the fact they may have widely differing values (eg, first preferences for that candidate had a value of 1.000 ach but second preferences from the elected candidate mentioned last paragraph would have a transfer value of only 0.16667). As a result, some ballot–papers may actually increase in value as the count proceeds, which is a violation of the idea of a single transferable vote. However, Australia (unlike Malta or Ireland) does not have a constitutional requirement of a “ingle transferable vote”, but instead simply uses whatever variant of STV the Parliament has chosen to enact as law. I have heard PR advocates argue that the Senate rules might be unconstitutional as a violation of a different section of the Constitution, “each elector shall vote only nice”, which was intended to prohibit plural voting at federal elections. (Ironically, it has sometimes been invoked by cranks writing letters to the editor to argue that preferential voting is unconstitutional. No word on whether first-past-the-post voting contravened this provision from 1901 to 1918 when voters had to cast three or six separate votes – not just preferences for a single vote, but separate votes – in Senate elections.) But this is still awaiting a willing and able High Court plaintiff (probably a defeated candidate) after 30 years.

        To avoid this defect, Western Australia has added a further refinement, of “weighted inclusive Gregory transfers”. All the elected candidate’s ballots are transferred, but they do not receive the same new transfer value. Instead the existing transfer value of each is multiplied by a uniform fraction so each is proportionately reduced. In my example above, if the second elected candidate has 1.1 quota when elected, all the ballots would be multiplied by 1/11 so the first preferences would go from 1.000 to 0.09191 each in value and the second preferences would go from 0.16667 to 0.01515 in value each. Although more accurate, this is also more complicated and time-consuming and it can lead to more votes being “lost” through rounding off of fractions.

        My own scale of preference among the various options (actual and hypothetical) would be:

        1. What I’d dub “d’Hondt STV” – each ballot has a value of 1/N+1, where N is the number of already-elected candidates ranked, on that ballot, earlier than its most preferred continuing candidate. So once your vote has elected, say, two, candidates, it is marked down to a value of 0.333. The quota would automatically adjust itself. I feel intuitively that this would work, but I have never been able to arrange the computing power to run bulk simulations.

        2. What I’d dub “first-parcel Gregory transfer” – ie, start by transferring the first, not the last, bundle of votes the candidate received. (If so many of them exhaust that some of the surplus remains, move on to the second bundle received, and soon, until either the surplus is gone or all the candidates un-exhausted votes have been transferred). I submit this would remove a slight incentive towards tactical voting, claimed by some observers to sway some voters in Tasmanian elections, where they don’t vote for their real first choice if they think he or she is certain to be elected anyway, thus preserving your ballot from having its value reduced. Electing an MP is a collective good, so if other voters use up their votes to hire the person you secretly want, you can free-ride without contributing.

        3. Last-parcel Gregory transfer as used in the Tasmanian and ACT Hare-Clark systems.

        4. Weighted inclusive Gregory as used in Western Australia.

        5. Common or garden (“unweighted”, “uniform” or perhaps we should say “one size fits all”) inclusive Gregory, as used for the Senate, would be very much my last choice.

      • The day for correcting my own posts: “On paper this sounds good but it causes distortions because they are all multiplied by the same fractional value when transferred” should read “… they are all changed to the same new value”. “Multiplied by the same fraction when transferred” would more accurately describe Weighted Inclusive Gregory, not Simple/ Uniform Inclusive Gregory.

      • Thank you, Tom, I think I get your answers to 1 and 3, although I will have a closer look at Bonham’s article later.

        I am, however, still unclear on the meaning of ‘last parcel’ and how it dictates the distribution of the surplus. If you have 120 votes and you multiply each by 1/6 because you have a surplus of 20, isn’t that looking at all the preferences rather than just the last parcel added?

      • [hoping this comment ends up located under JD’s query, which it’s in reply to…]
        JD, thanks for spotting that. I could have worded it more clearly. In that example, all 120 of the candidate’s ballots were meant to be first preferences, as an ultra-simple illustration of the Gregory fractional transfer formula (120 – 100,/120). If, say, 75 had been first preferences (still retaining their original value of 1.000) and the other 45 votes had been (say) 180 second-preference ballot-papers with a transfer value of 0.25 each, then under the ACT/ Tasmanian rules, only the first 75 would be transferred (at a value of 20/95 each, ie 0.2105).
        (Whereas if you were using the Senate [Uniform] Inclusive Gregory method, all 225 ballot papers would be transferred, and all of them would receive the same new value of 125/225 each, or 0.5555 of a vote. So the 180 second-preference ballot-papers would have _risen_ in transfer value after helping to elect a candidate. A monstrous perversion of the idea of a _single_ transferable vote!)
        The Gregory method was adopted to replace the older method of transferring 20 of the 120 ballots, physically selected at random, with all transferred ballots retaining a value of 1.000 vote throughout the entire count. This method is still used in Republic of Ireland, I believe, and for the NSW Upper House, because it was entrenched in the State Constitution in 1978, which was unfortunate, since STV best practice has moved on since then. The Senate used this method until the 1983-84 reforms, when [Uniform] Inclusive Gregory was adopted.

      • [That should have said “only the first 75 would be transferred (at a value of 20/75 each, ie 0.2667).”
        Ritvars, the Latvian system has another point in common with the Swiss electoral system: limited cumulation of personal votes within a list. Functionally (ie, in terms of finishing order of votes, if not necessarily of the ratios among candidates), giving a candidate either a positive vote, a zero vote, or a negative vote in Liepaja would be equivalent to giving a candidate either 2, 1 or 0 votes in Lucerne.
        Norfolk Island (a very small, isolated, and constitutionally very eccentric external territory of Australia) uses a similar system but without party lists.

      • Indeed, Henry. That’s where I got the idea from. As far as I can work out on the backs of several envelopes, it would (where voters give their preferences by parties) produce a similar variation from standard STV as D’Hondt with List-PR produces from List-PR with Droop quota and highest average.
        Since Hare-Clark uses rotated ballots, it distributes seats more closely to how D’Hondt would since the first-preference votes tend to be more equally divided among the candidates of each party’s team.

    • Thanks Alan. I should’ve been clearer. What I believe in is actually at root a deliberative vote for the Chair but with some resemblances to a casting vote in that the Chair should only vote (a) once all other members have either voted or abstained, and (b) if their vote will change the result. If after the Chair has either voted or abstained, there’s a tie on a motion, the motion should be lost. I don’t agree with the Swedish rule (used frequently from 1973 to 1976, I understand, when the Riksdag was tied 175-175 between left and right coalitions) of drawing lots to break ties *on motions* (including Bills).
      This does however avoid the Victoria 1985/ WA 2002 absurdity of a 22-22 or 18-18 tie morphing into a 22-21 or 18-17 “majority” for the party with fewer popular votes because the chair cannot vote unless a tie already exists.
      Among candidates, I could live with drawing of lots but I actually believe an “older candidate wins a tie” rule would have greater intuitive public support because it is less arbitrary. Institutions use seniority as a tie-breaker in many other contexts (eg, public service promotions when two or more applicants are deemed equal in efficiency; deciding which judge of a court serves as acting chief justice when the normal CJ is absent; deciding which member of the Knesset, Bundestag, etc presides while a Speaker is elected as first order of business; etc); it is easy to determine; and it has no consistent racial, gender or ideological bias in any direction. (Donald Rumsfeld, to pick a topical name, was both the oldest and the youngest DefSec in US history). By contrast, the sheer randomness of drawing lots seems to me to invite a lot of Bayesian regret. If you think Therese LePore has been kicking herself over the Florida 200 butterfly ballot, imagine if it had been a genuine 50.000%-each-way tie, if Katherine Harris had tossed a coin, and if Al Gore had spent the last 14 years telling himself “I should have called heads! I could have been president, but oh no, I had to call tails…!)
      You would have to have a lot of tied elections for it to produce a gerontocracy.

      • Also, there is more leeway for the suspicious to suspect the electoral officer of rigging even a random draw. Whereas birth certificates are, as we all know, taken as final and conclusive proof of all facts averred thereon.

      • Tom

        I take your point about the likelihood of rigging election certificates versus rigging random draws. Barrack Obama may have his doubts.

        On the other hand, the Commonwealth Electoral Act 1918 Australia at Section 213 sets out a voter-verifiable method of random draw.

        Electoral ties are considerably less likely under STV than under other systems (just saying) because you can look back to the last count at which the candidates were unequal.

      • jd

        A parcel is the votes credited to a candidate at any stage of an STV count. The first parcel, obviously is those ballots marked 1 for the candidate. The later parcels comprise the ballots each continuing candidate receives on the election or exclusion of another candidate. The last parcel is the votes which elect the candidate.

        The different methods of calculating transfer values revolve around whether to examine the last parcel or all previous parcels. There is no controversy with excluded candidates, each parcel is distributed at the transfer value at which it was received. There is large controversy with distributing the surplus of elected candidates.

        Personally, and unlike Tom, I am a last parcel conservative.

  6. Latvia uses open list PR. In a case of tie between parties, the last seat goes to the party with more votes. If numbers of votes are the same, the seat goes to the party which submitted candidate list the first. In a case of tie between candidates, the last seat goes to the candidate who was placed in the list highest.

    • Ritvars, I am a bit confused by a couple of points. If two parties are tied, how can the seat go to the one with more votes? Isn’t the point that they have the same votes?

      I do indeed understand the Latvian system to be open list, but that would mean there is no list order. Or is there an order established by the party solely in case of ties?

      • Ritvars, do you mean a “tie” in the sense that the two parties have equally high (D’Hondt or St-Lague) averages, or equally large remainders, but different vote totals? Eg, if Party A has 3.45678 quotas and Party B has 1.45678 quotas, then (assuming you’re using Largest Remainder PR) they would be tied after they are awarded 3 seats and 1 seat respectively?

      • Alright, in the first case, the tie was meant the tie for the last [Sainte-Laguë] divisor.
        As for candidates, parties submit their lists placing candidates in ranked order e.g. http://sv2014.cvk.lv/saraksti/d1c00ea367_V.html. Candidates appear in that order on ballot papers: http://uk.itvnet.lv/gallery/zinas/latvija/art_18576.jpg. Voters may give ‘+’ mark to those they like the most and to strike out whom they dislike: http://cvk.lv/pub/upload_pic/Bals_majas/5foto.jpg Election order is determined by difference of pluses and strike-outs: http://sv2014.cvk.lv/Part-10-Strukt-1.html If that difference is the same for some candidates, they are ranked in a party-given order.

      • In answer to Rob, I might suggest that “Dog Catcher” is and can be a paying job, often with a vehicle, so it’s attractive in that respect. And there certainly are many people who would like to dedicate their lives to the health and welfare of animals, so there’s a big motivation factor there. We have paid dog catchers in my county in upstate New York.

      • Paid, sure, but _elected_ dog-catchers? “Vote for me, I’ll target the yappy little ones!” “My opponent turns a blind eye to surreptitious pooping on the lawn!” etc.

      • Switzerland (like other countries, I suspect) has the same rule. List-order counts only when two candidates of the same party are exactly equal in votes. It’s the most minimalist role for list-ranking possible (short of drawing lots if two team-mates are equal in personal popularity), and the opposite extreme from a closed list such as in Israel.

  7. Well, he opposite extreme from closed lists would be breaking intra-list ties by lot (or drawing straws, etc.). Because even the “most minimalist role for list-ranking” is still a role for list-ranking.

    • MSS, ‘swhat I meant. In Israel or South Africa, list-order determines 100% of election results. In Sweden it determines (say…) 80%, in Belgium (say…) 55%, in pre-1993 Italy it determined (say…) 20%, and in Switzerland it determines 0.001%, ie only when two candidates of the same list are tied in personal votes.I have pulled these percentage figures completely out of my head but you get what I mean

      • I’m pretty sure Belgium uses flexible-list, with at least 90% list order-determined.

      • Thanks, JB, and my overall point still stands. Someone with the time and energy to pore over election returns could measure this quite objectively, ie what % of MPs would still be in the legislature if election were based solely on list ranking. I’d be surprised if this hasn’t been done before.

      • Yes, Tom, I see your point, and it is a good one (leaving aside the specific percentages for non-closed list examples). I’d simply add to it that in a strictly open list, a list order determines 0.000000%. Because there is no list order.

      • Agreed. I suppose an open-list system could have candidates ranked by lot, or alphabetically, or rotated on different versions, and draw lots if they tie.
        Footnote to that, I found out recently that Spain changed the format of its Senate ballots in 2011 so that candidates within each party group are now ranked by the party rather than listed alphabetically. So now the 3 candidates elected from the largest party, and the 1 from the second-largest, will be those chosen by the party rather than those with A, B and C surnames. Limited Vote emulating Australian rather than Irish STV…

  8. Three days before the Martin Place siege, the High Court of Australia refused an application to rehear a matter on which it had previously been tied.

    In 2012, the court divided equally over a criminal appeal by one Man Haron Monis who was seeking the quashing of his convictions for sending offensive items (abusive letters to the families of soldiers killed in Afghanistan) through the mail. The judicial tie had the effect of affirming the decision of the NSW Court of Appeal which had previously upheld the convictions. The Court of Appeal is the appellate division of the Supreme Court of New South Wales.

    The Sydney registry of the High Court and the NSW Supreme Court are within 2 blocks of the site of the siege.

    • An interesting angle, Alan, on the horror that the Lindt Cafe patrons and staff were put through.

      An aside: I immediately recognized that location, as I remember walking through the Eye Hospital grounds, past the NSW Parliament, and into Martin Place where there was a rather tame and lame “Occupy” protest underway. This was in November, 2011.

  9. This would probably be a bad place to mutter that in a preferential count, random draws are much less likely because you can count back to the last stage of the count where the candidates were unequal.

      • At the same time, while ties are easier to resolve under STV/AV (as long as they don’t arise on the very first stage of the count), they are probably more likely to arise and to matter.
        If we define “tie” to mean not just “equal number of votes” but “equal number of votes that makes it not immediately clear which of the equal candidates to eliminate or declare elected first,” then “ties” in the strict sense are more frequent in STV.
        In FPTP, a result of 40, 30, 15, 15 isn’t a tie. Under STV/AV, it is.
        On the other hand, 40, 40, 20 is a tie under FPTP but may not yet be a tie under STV/AV. However, probabilistically, you are more likely to get equal numbers among the “minnows” in single or double figures than among the “whales” with four, five or six figure totals. And in FPTP, only ties among the “whales” count. In STV/AV, ties among the “minnows” can affect the result – see the late Professor Dummett, also Antony Green’s analysis of various Australian Senate races where a few extra votes at a particular “chokepoint” in the count could have caused a cascade in a different direction.
        As may have been mentioned once or twice before, I’m an STV/AV supporter but I’m not one-eyed about its virtues. I concede it has disadvantages: I just think they’re greatly outweighed by its advantages. If one favours a particular electoral system because it encourages more honesty among voters casting their votes, and it counts those votes more honestly, nothing is served by turning oneself into a David Cameron of the STV/AV side.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s