SA Labor banning above-the-line preferences in upper house


The South Australian Labor government has released legislation to reform the voting system used for the South Australian upper house. Like the federal and NSW upper houses before it, the legislation aims to eliminate the flawed and opaque group voting ticket system, but it comes up with a strange model which would have some odd outcomes.

I won’t spend a lot of time analysing the proposal in detail, for that check out the work by Kevin Bonham and Henry Schlechta.

In this piece I want to explore the theoretical problems with the SA Labor government’s proposed bill, and some alternatives, as well as raising some frustrating hypocrisy left over from this year’s Senate reform debate.

For the Senate and the upper houses in New South Wales, Victoria, Western Australia and South Australia, the voting systems are all similar. You can vote for a group above the line, or vote for individual candidates below the line.

Until 1999, all five bodies used the group voting ticket system – if you voted above the line, you just marked a ‘1’ for the group of your choice and all of your preferences would follow a predetermined order.

New South Wales abolished group voting tickets following the 1999 state election, and instead let people vote for multiple groups above the line. Earlier this year, the Senate went the same way, while also encouraging voters to number more than one box above the line. This worked very well, resulting in most votes flowing as preferences.

So there’s a clear precedent about a model which can be used to replace the group-voting-ticket system, which has now worked effectively for both the Senate and the NSW Legislative Council, and could easily be implemented for the upper houses in the three other mainland states.

Instead, the SA Labor proposal would keep below-the-line voting largely as-is, and abolish any preferencing for above-the-line votes.  If you choose to vote above-the-line, your vote will only go to the group who receives your vote, and no preferences will flow.

This is an astounding decision from a party which spent the early part of this year scaremongering about how Senate reform would result in millions of preferences exhausting. While Senate reform opened up the possibility of exhausted votes, the SA Labor proposal effectively mandates exhausted votes, with only below-the-line votes having any chance of flowing to a candidate in another group.

You can have a good proportional representation (PR) voting system without preferences. Most PR systems in mainland Europe don’t use preferences, and produce reasonably fair outcomes. They also have the advantage of being very simple, compared to those systems used in Australia: you mark a single box and your vote will count.

Most European PR systems use a “highest average” system. Seats are allocated one-by-one to parties based on their average vote per seat, so you get to an outcome where each party has a similar average number of votes per seat. The two most common versions of this are Saint-Lague and D’Hondt.

These systems work best when the proportion of the vote per seat is relatively small. For example, the 120 seats in the New Zealand parliament are allocated proportionally across the country, so you win one seat for approximately 0.8% of the vote. If the system is structured like this you don’t need to worry much about tactical voting: pretty much every party has a seat which is in play, and your vote is likely to be helpful (this is putting aside questions of a minimum-vote threshold).

As the number of seats to get elected gets smaller, the number of potentially wasted votes grows, and the value of casting a tactical vote becomes greater. If you are electing, say, five MPs in an area, it’s quite easy to see an unstrategic splitting of the vote producing a disproportionate outcome. When that is the case, it’s better to use a system which uses preferences to allocate overflow votes and the votes of minor parties and candidates who can’t get elected.

The simple PR approach has the advantage of simplicity. Ballots can be small, you just mark one box, and vote-counting is quick. PR with preferences is more complex, but allows for more nuanced votes – you have insurance in case your first-placed candidate isn’t in with a chance.

In this context, the proposed SA voting system is bizarre. It won’t make for a simpler ballot paper (although it may well lead to less minor parties standing, reducing the size). People will still have two options of how to vote, and below-the-line voting will still make the ballot paper very large. It will also be confusing as voters will be able to express preferences below-the-line but not above.

If you assume most voters will continue to vote above-the-line, the system will effectively become a largest-remainder voting system. Seats will be allocated according to the same quota as is used in a preference system, but the final seats may be allocated based on small fractions of a quota, with preferences becoming a minor factor. While there is a risk of this happening under the new Senate voting system, it would be almost guaranteed under the SA Labor proposal.

SA Labor should really choose one approach or the other: either maintain a preference system by moving to a model similar to the Senate or the NSW Legislative Council, or instead do away with preferences altogether and move to a highest-averages system, where seats are allocated to parties.

If you were to count the primary votes of the 2014 election using the Saint-Lague method, you would have produced the same result of 4 Labor, 4 Liberal, 1 NXT, 1 Greens and 1 Family First, but with a much easier method of casting and counting votes.

Basically parties are lined up by their primary votes, and the party with the highest vote wins a seat, and their vote is divided by three. Repeat this process, and if a party wins a second seat their vote is divided by five, then seven, then nine. The result looks like this, with winning votes underlined:


(This system would work even better for the NSW Legislative Council, but unfortunately the current voting system is embedded in the NSW constitution).

This may seem like a big change, but it was actually proposed by the same government prior to the 2014 election. I would suggest that this system would be superior to the new model, which seems stuck somewhere in the middle.

We’ve seen something like this happen before. When the ACT Legislative Assembly was first created, the Senate opted to use the D’Hondt system to allocate seats to parties, but also added Australian elements, including preferences. It was incredibly complicated and took a long time to count, and it was significantly worse than either a pure party list system (as seen in mainland Europe) or a Hare-Clark system which works best when the jurisdiction is divided up into smaller districts. I’d hope that the SA government won’t repeat this mistake.

Liked it? Take a second to support the Tally Room on Patreon!
Become a patron at Patreon!


  1. The problems with this proposed ballot structure are similar to those in the Federal Government’s first draft of the Senate reform bill earlier this year, before it was amended to reflect the recommendations of the federal Joint Standing Committee on Electoral Matters. Basically, there is no logic in the different approaches taken to above the line (ATL) and below the line (BTL) voting.

    (1) The SA draft makes it obligatory for BTL voters to show a preference for every candidate, but impossible for ATL voters to do that.

    (2) If I vote ATL for a group, I will be deemed to have shown preferences for the group’s candidates in their ballot paper order, and no other preferences. If I had expressed the same preferences for the same candidates BTL, and no others, my vote would be informal. As I said in the context of the Senate discussion earlier this year, this is like having different tax rates applicable depending on whether you lodge your tax return electronically or with a paper form.

    (3) Given the massive federal investment in publicising the new Senate voting system, it is highly likely that some SA voters will try to extend preferences ATL by marking squares 2, 3 and so on. Those preferences won’t count. By rights, the South Australian Electoral Commission should make a comparable effort to clarify to voters that if they want to extend preferences across groups, or reorder candidates within groups as the Tasmanian voters did with the ALP’s Senate group, they will have to vote BTL. To do that properly would not only cost a lot, but will also likely leave some voters confused when the next Senate election comes along.

    (4) The system will not only eliminate preference harvesting associated with ticket voting, but, in practice, any negotiated exchanges of preferences. This will distinguish it from every previously implementation of STV at parliamentary elections in Australia, including even those ones involving Robson rotation (all of which have incorporated some form of optional preferential voting which made preference swaps between groups at least more or less feasible).

    (5) This makes it more likely that minor party voters will have to vote strategically, choosing not necessarily the minor party which they most prefer, but rather the one with which they are broadly sympathetic and which seems most likey to exceed the threshold of exclusion. This feature is one generally associated with plurality (first past the post) systems. At least with ticket voting first preferences can be accepted as reflecting the true intention of the voter: under this system, even first preferences may be questionable.

    (6) This will be a classic example of a hybrid, internally inconsistent system, in which the basic principle of single transferable vote PR is contradicted by the ballot structure which virtually all voters will use.

    (7) There are no reasons in practice which would prevent the Senate system from being adopted holus-bolus; the AEC has proven that optical character recognition and computerised counting work.

    It is hard to avoid the conclusion that the ALP government in SA won’t adopt the Senate system simply because it would place in a very poor light their federal colleagues’ resistance to the Senate reforms earlier this year.

  2. Other big news from South Australia is the the electoral commission has FINALLY followed the state constitution and created fair electoral boundaries. And it wasn’t even that hard!

    All the rubbish that was spoken that it would need weirdly shaped electorates with no community of interest…was just that – complete rubbish.

    A couple of simple changes did the trick. Instead of bottling up ultra-safe Liberal areas in Morphett, they have now been spread between Morphett and Colton creating two moderately liberal leaning electorates in place of one ultra-safe Liberal seat and one ALP leaning.
    Instead of pulling all the labor leaning territory in the inner south into Elder, there is now some mixing with the Liberal leaning territory that predominates in the area.

    Just goes to show what a piss-poor effort the 2012 commission was.

  3. On the other hand, they couldn’t have done a worse job in the north east if they tried. Seats like King and Newland are truly dreadful, absolutely no community of interest whatsoever. I have always maintained that there should be as few as possible electorates crossing the hills.

  4. I just received an email from the SA boundary commission. If I understand correctly, the ALP are appealing against the redistribution in court.

    I didn’t think you were allowed to do that.

Comments are closed.