An Alternative vote?

May 4, 2011 14 Comments
By

After the recent Canadian Federal General Election, I was surprised to find out that the Alternative Vote system of election has been tried and discarded in British Columbia, Canada. In the early 1950′s no less. Back in the UK, the Alternative Vote system is being touted as a ‘fairer’ system of apportioning the popular mandate, but is it really?

Taking the 1953 BC election as a yardstick, where AV was the voting system. This being a real world example of the practice in action, not theory or modelling, but real live, in your face numbers.

In 1953 the victorious Social Credit party started out at the first count with just under 38% of the total vote, which increased to just under 46% of the vote after ‘adjustment’. The smaller parties who were knocked out after the first count had their votes allocated to the winning party, not spread amongst the others. See Wikipedia entry here. Er, hold the phone. So how is that ‘fairer’ than the first past the post system? Explain it to me. By that I mean explain it without any flim-flam, in words an ordinary voter can understand. Especially when we’re voting for a specific person to represent us, not merely their party.

The words say ‘fairer’, but the numbers add up to something else. As far as AV is concerned, I’m not convinced. Needs more thought.

Tags: ,

14 Responses to An Alternative vote?

  1. May 4, 2011 at 3:20 pm

    As I understand it, as a least popular candidates are knocked out, their second preferences are then counted, so will be spread among the remaining candidates. Isn’t that how it happened in BC?

  2. May 4, 2011 at 3:58 pm

    Erm, no.

    Let’s start with the 2007 French Presidential elections.

    In the first round, Nicolas Sarkozy got 31% of the vote, Segoline Royal got 26%, and other people got the rest.

    In the second round, all candidates apart from Sarkozy and Royal were eliminated, and the voters solely had to choose between the two most popular candidates. This time, Sarkozy got 53%, and Royal got 47%.

    So Sarkozy got to be President of France.

    AV simply does this in every seat, by allowing voters to make their ballot paper show *not only* their first preference, but *also* who they’d vote for in a run-off. And everyone who wins a seat in AV has won it in the same sense as Sarkozy – that more than half the voters would rather he’d won it than the next most popular alternative.

    • May 4, 2011 at 6:45 pm

      I don’t see the French system and AV as being the same.

      In the French system ALL voters get to change their votes between rounds. In AV only the minority of voters who voted for anyone other than the top 2 gets a chance to vote for another choice.

  3. May 4, 2011 at 3:59 pm

    “erm no” to Bill, not to Longrider, who is correct.

  4. May 4, 2011 at 5:08 pm

    John B

    So effectively disenfranchising minority voters isn’t an issue then? There’s also the observation that people have used the AV system to heavily ‘vote tactical’.

    From what I can see, AV appears to heavily favour Parties rather than Independents. As one who tends to vote for a specific candidate rather than a Party, this is a concern. I’m still not convinced.

    • May 4, 2011 at 8:50 pm

      “There’s also the observation that people have used the AV system to heavily ‘vote tactical’.”

      Unfortunately, there’s no evidence or logic to back it up. The reason people in Queensland voted 11 seats to the One Nation Party is that they didn’t feel they had to vote liberal just to keep the other guys out.

      “From what I can see, AV appears to heavily favour Parties rather than Independents.”

      The evidence in the UK is that non-FPTP delivers better than FPTP in terms of independents. We’ve had 6 independent MPs since 1979. Half of those have been in Northern Ireland, despite them only having 3% of the seats.

      We’ve had around a dozen independents in the mainland in 60 years of elections. Calculated across all parliaments, you have about a 0.1% chance of getting an independent, and most of those were in 1945. If you take that election out of the sums, it’s 0.04%.

      I’d say that when those are your odds of getting an independent, it’s probably worth risking it on another system.

      • Paul
        May 5, 2011 at 12:37 am

        Never mind the fact that for parties like the Lib Dems, UKIP (and the BNP and Greens too), the current system is manifestly unfair.

        In the 2010 election UKIP polled 920,000 votes and got no seats.
        The DUP polled around 18.3% of that number in Northern Ireland (168,000 seats) and got eight seats.

        Something about that doesn’t seem fair to me. Add that onto the fact that Ulster constituencies are smaller anyway than many English constituencies where UKIP pick up votes and it does seem very skewed.

        But even ignoring UKIP, the Lib Dems do very badly out of such a system too.

  5. May 4, 2011 at 5:16 pm

    That’s a pretty confusing table. I can only assume that by ‘Final Count’ they mean the totals of head-to-heads in each seat. It’s not really a useful figure. Just like FPTP, To understand an AV election you have to look at each seat.

  6. May 4, 2011 at 5:52 pm

    The only argument for the No camp that I might accept is that FPTP absolves me from any responsibility for what’s happened to the country, since I’ve always lived in some safe-seat constituency or other. Compromise and coalitions are bad things, eh; but what about the five-star disaster governments we’ve had from both Labour and Conservative. The present mess is being blamed on the last 10 years, but really it’s developed over more like 40. I’ve had enough of the punt swinging left and right into holly bushes all along the river.

  7. Lord T
    May 4, 2011 at 8:44 pm

    Bill,

    Better not think too long.

  8. May 5, 2011 at 6:08 am

    Okay. My very last word on the subject. After due consideration, I don’t think AV is the right democratic solution, yet there’s a body of opinion that seems to think AV is ‘progress’, and any move towards any form of proportional representation is a good idea. My own opinion (lately formed) is that another form of proportional representation may prove more appropriate.

    My major caveat is this; dashing towards the first solution offered by politicians is how we originally got into the mess we’re in. Do we really want any type of voting reform for it’s own sake, just because it’s the only choice we’re given?

    • May 5, 2011 at 7:31 am

      If we don’t, the politicians will be able to point back to “the great referendum of 2011, where FPTP solidly won the citizen’s support” to stop any and all further changes. At least if AV wins, everyone agrees that it’s not ideal, and there’ll be further incentive to change to a better system.

  9. May 5, 2011 at 6:22 am

    I’m with Sackers on this though AV is not the solution to the democratic dilemma we have.

  10. May 5, 2011 at 10:09 am

    I’m going to insert something from my blog here, a commenter, Nigel Sedgwick, who knows his stuff and it’s only fair that both points of view are given.

    I’ll come back later on James’ request to me for further clarification. However, I am a little disappointed in that he seems to be including me in his “therefore wildly speculating as to how it works”. I’ll have you all know that I was returning officer for the elections to the Royal College of Science Union, back in 1973. That used AV (STV with single seat constituencies) for, IIRC, 3 union posts and over 1,000 voters. All the Imperial College Union and constituent college unions used AV/STV and everyone understood how it worked: including especially me and my counting team.

    More up to date, Armando Iannucci, creator of “The Thick of It” and other good things, writes today in support of AV in the Telegraph and, more extensively, in the Independent.

    James, at the 8th comment above asks me supplementary questions. I’ll try and answer these as best I can.

    1. Nigel: “The ‘Australian’ method allows less information to be expressed than does the AV ‘UK’ method.” James: “How? Same method. Last eliminated, 2nd preferences are distributed. Repeat.
    ——————–

    I’ll use the same example as I did on Samizdata: 4 candidates and a close-run thing (so approximately equal a priori probabilities).

    With FPTP, there are 5 possible ways of voting: one for each of the candidates and a blank/spoilt ballot paper.

    With the AV method used for Australian national elections to their lower house of parliament (the ‘Australian’ method), each voter must rank every one of the 4 candidates; papers with less voted rankings are (usually) viewed as invalid (that is named there as ‘informal’ – and let’s not get into the law on this just now). So the ‘Australian’ method has 24 possible legal ways of voting: 4 first choices times 3 second choices times 2 third choices, and no choice for the last choice.

    The AV method proposed for the UK allows one (in this example) to rank zero, one, two, three or four candidates. There are 24 ways of ranking 4 candidates (as with the ‘Australian’ method), also 24 ways of ranking 3 candidates, 12 ways of ranking 2 candidates, 4 ways of ranking 1 candidate and 1 way of ranking zero candidates. This gives a total number of ways of voting of 24+24+12+4+1, which is 65 ways of voting.

    Note that, if the proposed AV system for the UK were the ‘Australian’ method, I would be actively against it.

    2. James: “Where does this RON come into it?”
    ——————–

    The addition of the Re-open nominations (RON) method is only meaningful if one has not ranked all the candidates (ie, in this example, ranked only 0, 1, 2 or 3 candidates).

    The Re-open Nominations (RON) method allows the voter to make an official abstention after having ranked as many candidates as (s)he chooses. There is more than one way of implementing RON, but the simplest is as follows. There is a separate box for RON which can be ticked or not ticked by the voter. On each completed ballot paper, any RON vote come into effect when all the ranked candidates have had their vote reallocated; the RON vote is never reallocated but remains for all subsequent rounds of voting. Without ticking the RON box, the ballot paper would be ignored after all ranked candidates have had their votes reallocated. Thus, with RON, the final round will be the one remaining candidate versus the RON votes: if RON has more votes, a new election is required; otherwise the last remaining candidate is elected.

    [Aside: if the RON option is not used, all completed ballot papers that do not make it to the last round are ignored. Thus the 50% level is eventually reached all cases, by one of the candidates (though there can be a tie - extremely rare and usually resolved by a recount or by the Returning Officer having a casting vote).]

    With RON, on the ballot paper where there are 0, 1, 2 or 3 ranked candidates, the number of options is doubled. So we have 24 ways of simply ranking for 4 candidates, 2*24 ways (two times for RON/notRON) of voting ranks for 3 candidates, 2*12 ways of voting ranks for 2 candidates (also two times for RON/notRON), 2*4 ways of voting for 1 candidate (also two times for RON/notRON) and 2 ways of voting for zero candidates (that is just RON/nonRON). This adds up to 24+2*24+2*12+2*4+2, which is 106 different ways of voting.

    The use of entropy, measured in bits/ballot paper, is a convenient and meaningful way of comparing the different voting methods, that is more obvious (both in information theory and common understanding) that the total number of ways of voting. The entropy measured in bits is, in this simple and approximate example, the logarithm to base 2 of the number of voting options. Thus each extra bit (ie binary digit) of entropy represents twice as many ways of voting.

    In my example of 4 candidates, assuming all ballot completions are equally likely (which is only a coarse approximation on average), the summary is as follows:

    FPTP (with blank vote option)………..5 ways to vote…….approx 2.32 bits of information
    AV (‘Australian’ method)……………24 ways to vote…….approx 4.59 bits of information
    AV (proposed UK method)…………….65 ways to vote…….approx 6.02 bits of information
    AV (with RON option)………………106 ways to vote…….approx 6.73 bits of information

    IMHO, more ways of voting, within the range likely for House of Commons elections, is a good thing. It gives the electorate more ways of expressing their view.

    3. James: link to liberalburblings.co.uk
    ——————–

    This video is a nice and simple explanation of the proposed AV method in the referendum. Though clearly partisan, IMHO it has no bias in its presentation of the description of the method of voting or of the method of counting.

    4. James: “One gets the donkey vote and reverse donkey, as people just put 1-2-3-4 etc. So a candidate named Adams benefits in this over a candidate named Zog.”
    ——————–

    The order of candidates on ballot papers is usually randomised after nominations are closed, so there is no favouritism of ‘A’ over ‘Z’, etc. Thus candidates cannot obtain this advantage by changing their names by deed poll for the election period.

    However, the ‘donkey’ vote is a problem in Australia, where people are forced to vote where they might have no opinion, or might be actively and equally against the election of any further candidates than those they have ranked. Likewise, voting is compulsory in Australian national elections: IMHO not a good idea.

    If the voting method in the UK is properly explained to people, they should be aware that such ‘donkey’ votes are unnecessary and also may well cause an effect that the voter does not want.

    5. James: “The system does require more time to count but what’s time compared to fairness?”
    ——————–

    Agreed that the cost/time issue should not be relevant in the case of FPTP versus AV.

    With scanning of ballot papers and semi-automatic optical character recognition (ie manual checking of a proportion of ballots and especially those the OCR ‘thinks’ are doubtful), the time and cost is reduced. In any case, I would recommend automatic counting and reallocation of votes; it is only getting the votes into the computer that is expensive.

    6. James: “Is there a quorum for this referendum? For example, if 20% of the voting population turn out, is it valid? Should it be?”
    ——————–

    My understanding is that there is no quorum for this referendum, and that setting a quorum was considered: introduced in the House of Lords and defeated in the House of Commons.

    Personally, I think that setting a quorum would be difficult, especially if it was missed by a small amount.

    It would be better, if there was a very low turnout, to allow Parliament to introduce a new Bill revoking the old Bill before the next election. I’d also be in favour of the Queen exercising her choice in the event of a very low turnout – requiring that we have another go in say 2 or 3 years, about a year before the next general election.

    It is worrying for me that people and the MSM (and bloggosphere too) are only taking a serious interest in this issue during the last few days before the vote.

    Best regards

    —————————————————————-
    For Nourishing Obscurity; Thu 05 May 2011 at 07:03

    I’ll come back later on James’ request to me for further clarification. However, I am a little disappointed in that he seems to be including me in his “therefore wildly speculating as to how it works”. I’ll have you all know that I was returning officer for the elections to the Royal College of Science Union, back in 1973. That used AV (STV with single seat constituencies) for, IIRC, 3 union posts and over 1,000 voters. All the Imperial College Union and constituent college unions used AV/STV and everyone understood how it worked: including especially me and my counting team.

    More up to date, Armando Iannucci, creator of “The Thick of It” and other good things, writes today in support of AV in the Telegraph and, more extensively, in the Independent.

    Best regards
    Nigel Sedgwick

    I replied that I wouldn’t dare include him in those who wildly speculate. Anyway, I’m off up town now to cast the vote.

    I am a little disappointed in that he seems to be including me in his “therefore wildly speculating as to how it works”

United Kingdom Time

Subscribe

Email us at contact orphans of liberty [all one word] at gmail dot com

Authors

For more about these renegades, click on the name to go to a short profile:

AK Haart
Churchmouse
James Higham
JuliaM
The Quiet Man

Orphans logo


Feel free to take this for your sidebar.