Drowned out by the news of the Olympic Dam expansion being shelved yesterday was the release of the final report of South Australia’s state electoral redistribution. This is a fairly dry topic at the best of times, this one at first promised to be reasonably interesting, as state redistributions go. South Australia’s redistribution commissioners, who perform their work between every election, have uniquely been given direction to seek “electoral fairness” ever since a provision to that effect was inserted in the legislation after Labor’s lucky escape in 1989, when John Bannon won a third and final election from a base of 48.1% of the two-party vote.
Successive redistributions have sought to achieve this by drawing boundaries that would deliver victory at the subsequent election to the party with the greater share of the two-party vote, assuming a perfectly even swing. This eminently rational approach could not overcome the basic flaw of the endeavour, which is that election results can never be so neatly predicated on the basis of what happened last time. The 2010 election was a remarkable case in point, with 22 of the state’s 47 seats recording double-digit swings against Labor, but the two most marginal Labor seats actually swinging in their favour (the only ones to do so). Labor was thus able to suffer a net loss of just two seats in the face of a plunge in their two-party vote from 56.8% to 48.4%, emerging with a solid majority of 26 out of 47.
That left the redistribution commissioners with a formidable task in drawing boundaries which met the electoral fairness requirement as it had previously been conceived. From a psephological perspective, the contortions required to burden marginal seat Labor MPs with the requisite Liberal-voting areas, assuming there were any nearby, promised to be something to behold. Instead, the draft boundaries published in May showed the commissioners had simply thrown up their hands and dispensed with the Mackerras-pendulum derived notion of “fairness” which had previously been applied. Their rationale for doing so makes for interesting reading, as it essentially argues that the Liberals’ defeat was down to political failings a redistribution can’t be expected to account for:
As many of the seats held by Labor were marginal, little would have been required for an effective campaign to influence the final result … Had the Liberal Party achieved a uniform swing it would have formed government. As quoted (in the findings of the 1991 Electoral Districts Boundaries Commission), “The Commission has no control over, and can accept no responsibility for, the quality of the candidates, policies and campaigns.”
That being so, the commissioners turned in an extremely conservative set of changes, and despite the protestations of the Liberal Party there has been no fundamental change in the final determination. However, the Liberals have been thrown the following bones:
• Bright has been given extra territory from its Liberal northern neighbour Morphett, turning Labor member Chloe Fox’s 0.3% margin in the original redistribution to a deficit of 0.1%. The Liberal margin in Morphett, which also cedes territory to Elder (see below), is accordingly down from 11.1% to 9.9%.
• Elder is redrawn in relation to its Liberal neighbours Morphett and Waite, cutting Pat Conlon’s margin from 3.4% to 1.7%.
• Waite also cedes territory to Ashford, so as to cut Stephanie Key’s margin in the latter electorate from 4.4% to 1.5%. The Liberal margin in Waite is reduced from 13.0% to 11.1%.
• Grace Portolesi’s 1.9% margin in Hartley has been cut to 0.5% by adding extra territory from neighbouring Bragg, where Vickie Chapman’s Liberal margin of 21.0% goes to 20.0%.
The redistribution is otherwise as described by Antony Green when the draft boundaries were published, the most notable changes being a boost in Labor’s margin in Little Para from 6.7% to 10.9% with the addition of territory in Elizabeth, the Liberal margin in Morialta dropping from 4.2% to 2.9%, and Norwood being renamed Dunstan in honour of its esteemed former member.
I also believe that Frome (currently held by Independent Geoff Brock) has gone from notionally Brock to notionally Liberal.
William,
Top post.
The folly of re distributing seats on the basis of “what happened last time” is that perhaps the swing against the incumbant has maxed out and the next election the swing is back to the incumbant Party while the challenger Party has their rock solid margins eroded at the same time and puts them in some danger of losing.
The fairness concept can be a crock in practice.
My other quarrel with the SA regime is that a good local member in a marginal seat who works hard and builds up their margin gets punished for it, because the redistribution commissioners have to bash their margin back down to keep the seat winnable for the other side.
That’s my biggest problem with the system, William. To be honest, if they wanted it to be more reflective of popular vote, they should’ve gone for PR.
Carey, Frome has gone from Labor to notional Liberal. They evidently did a Labor-versus-Liberal two-party preferred candidate there after the election which gave Labor a margin of 0.2% over the Liberals. But now they’ve added 1627 more rural voters, it’s 1.5% in favour of the Liberals.
I stand corrected.
That’s the only way to do it, CM, if fair representation of the parties is what matters to you. If it’s not what matters to you – if you prefer “stable government”, for instance – it seems to me that you should follow the idea through to its logical conclusion and advocate a directly elected executive.
Certainly. As long as the directly elected executive is balanced with a separate legislature. Separation of powers is just as important to me.
[That’s the only way to do it, CM, if fair representation of the parties is what matters to you. If it’s not what matters to you – if you prefer “stable government”, for instance – it seems to me that you should follow the idea through to its logical conclusion and advocate a directly elected executive.]
Why can’t we have both!?
Everyone gets two votes, one for who they want to run the executive and another vote for which party they want in the legislature which is then done by proportional representation, but with a minimum of, say, 5% to qualify for 1 seat.
I’ve sometimes wondered about extending the preference system.
It won’t happen, for a number of reasons, but why shouldn’t everyone’s preferences be counted, instead of just those voting for the losing candidates?
If there were more than 3 candidates, is it theoretically possible under the current system, for the candidate with the highest number of first preferences to lose to another candidate who is also disliked by more voters than the leader on first preferences?
As for SA, Labor’s vote last time was very low and they were saved by individual seat tactics.
Same again?
I’ve sometimes wondered about extending the preference system.
It won’t happen, for a number of reasons, but why shouldn’t everyone’s preferences be counted, instead of just those voting for the losing candidates?
If there were more than 3 candidates, is it theoretically possible under the current system, for the candidate with the highest number of first preferences to lose to another candidate who is also disliked by more voters than the leader on first preferences?
As for SA, Labor’s vote last time was very low and they were saved by individual seat tactics.
Same again?
I doubt it. I think whoever wins the next election will do so with the support of the majority. If Labor get back in, it will be because they had the trust of the people (although, after this redistribution, the onus is still on the Libs to gain support).
Although one could make the argument that the Libs just need to take 3 seats off Labor and they could be in the position to negotiate a minority government with the 3 indies (assuming they don’t win the seats from any of them)
The “electoral fairness” thing seems like a fundamentally unsound approach because governments can always try to dish out pork to ensure the swing isn’t uniform. It’s often not about opposition failure when this happens, but rather government advantage.
[If there were more than 3 candidates, is it theoretically possible under the current system, for the candidate with the highest number of first preferences to lose to another candidate who is also disliked by more voters than the leader on first preferences?]
So long as the system requires you to number each candidate consecutively (as opposed to allowing, say, Langer voting), this is logically impossible. However, what you can have is a situation where majorities prefer C to A, C to B and C to D, but C still loses as an arbitrary result of the order in which candidates are excluded. In this situation, you could have A coming first on first preferences, B coming second, C coming third and D coming fourth. The distribution of D’s preferences need not necessarily put C ahead of A or B, because a lot of the D voters who prefer C to A might give their second preference to B, and/or a lot of the D voters who prefer C to B might give their second preference to A. So the distribution of D’s preferences still leaves C in last place and accordingly excluded, and all those A and B votes which have C in second or third place count for nothing.
If you’d like the make the headache that no doubt gives you even worse, you can explore the wonderful world of Condorcet or “pairwise” electoral systems, which seek to find the winner out of all possible two-candidate preferred combinations (which in the example I’ve just discussed would, of course, be C).
Actually, I think I might have spoken too soon in saying SR’s scenario was logically impossible. The preferences of the fourth-placed candidate (D) might put the first placed candidate (A) into third place and thus see A excluded, even though A might win in a head-to-head contest between the eventual winner out of B or C.
William: many is the time I have stood behind vote counters while scrutineering and watched the same candidate get “2” over and over again, regardless of who the “1” was, and pondered whether this might, at the end of the day, actually be the compromise candidate to make everyone happy.
Instead, a major party candidate (of whatever flavour) just fell over the line because they had more votes with the number”1″ on them at the first hurdle – even if they ALSO had a lot more with the numbers “5” or “6”.
It would be interesting to have a system whereby all the numbers are added up and then divided by the number of votes, thereby (in the basis of who has the lowest average number) finding out who really is the “most preferred”.
Scrap the lowest average number crap. It would need to be value-based, so out of 4 candidates a 1st preference is worth 4, 2nd worth 3, 3rd worth 2 and 4th worth 1.
The HIGHEST number would then show who was, overall, the most popular candidate (or least hated, whichever way you want to do think about it ;-))
Thanks, William, that’s pretty much the scenario I was thinking of, hence me saying “more than 3 candidates”.
Danny Lewis’s statement about a candidate getting a lot of 2nd preferences is interesting, as I’ve often thought that and it was another scenario to look at.
I’d shared his thoughts about a value-based system, too.
Fair (?), but definitely unworkable in this instant gratification society.
William
I would like to ask if this topic was stimulated by my posting of the Adelaide Now reportage on the matter, in the larger thread.
[ It would need to be value-based, so out of 4 candidates a 1st preference is worth 4, 2nd worth 3, 3rd worth 2 and 4th worth 1.]
I find that systems like this seem like a good idea in principle but are very difficult to protect from strategic rorting in practice.
Example: in a two-candidate race, A has a 52-48 margin over B. If those are the only candidates, A wins.
However, B gets C to stand. C’s politics are on the same side as B’s but more extreme, so that A’s voters all prefer B to C, while B’s voters all prefer C to A. Hence 52% vote A-B-C and 48% vote B-C-A.
Now on the value-based method A has 52%*3 + 48%*1 = 204.
B has 52*2 + 48*3 = 248.
B wins the election, although the voters prefer A to B, because the presence of another candidate effectively doubled the value of the votes in favour of B.
Now suppose that A had envisaged this scenario and instead told his voters to strategically put C second, and they did. Now A wins with 204, C has 200, B has 196.
Which is back to how it was, except suppose someone leaked this plan to B and B decided he could not win but he would stop A at all costs. Therefore B told a small number of his supporters to thwart A’s strategic voters by voting C-B-A. Now C wins.
I totally agree that there is too much emphasis on the #1 vote in our voting systems, given that for many voters voting is not even about electing the most liked candidate, but more about voting against the most disliked. I tried to come up with a Senate system once in which a vote is distributed from both ends, counting +1 from the top and -1 from the bottom, but a trial simulation resulted in the Australian Democrats winning about 40 seats, so I scrapped it. 🙂