Is the Robson Rotation a Turn too Far?

It is a truth universally acknowledged that there is an advantage to be drawn from winning, or being given, a favourable position on the ballot paper. Anyone who has been at the draw of candidates’ names will have heard the whoop of joy from the candidate whose name is drawn first.

The election of Members of Parliament should not be dependent on luck.

To eliminate this element of luck some form of rotation of the order of candidates on ballot papers is essential.

The ACT and Tasmania use the Robson rotation. This, along with the absence of above-the-line voting and group voting tickets has in recent elections (contrary to results elsewhere in the country) prevented unrepresentative micro parties gaining election and unscrupulous joke party candidates from gaming the system.

As the number of members being elected per electorate rises, the number of variations under the Robson rotation also rises dramatically. The election of six Senators using the Robson rotation may be reasonable but its use in the election of twelve Senators in a double dissolution is not.

A slavish adherence to the Robson rotation also allows commentators who wish to be pedantic to have free rein.

In the ACT, the 5 member divisions require 60 different versions of the ballot paper, the 7-member district requires 420 versions. To get the rotations in every column equal, the number of rotations is equal to the first number divisible by every number less than the number of vacancies. But if you have more than 7 vacancies, the number of rotations required to get Robson Rotation working sky rockets, 2,520 for 9 person tickets, 27,220 for 11 and 360,360 for 13 or 15 person tickets, over a million when you get to 17. So I presume the number of candidates that can be nominated on a single ticket will be limited to overcome this problem.

– Antony Green, Tally Room (27 May 2009)

Green’s solution is to limit the number of candidates, not amend the method of rotation. Let’s instead look at simpler versions of rotation.

Single member electorates

In single member electorates, such as for the House of Representatives, the issue is simple. Political parties preselect a single candidate and then campaign to have that candidate elected. Occasionally, and probably increasingly, thirteen or more candidates will contest a single electorate.

Electoral Reform Australia recommends a simple rotation that will allow all candidates to share the top position equally, using the following process:

  1. Randomly draw the names of the candidates to give an initial order. This is the first ballot paper.
  2. Take the name of the candidate last on this list and place that candidate’s name first and move every other candidate down one place. This is the second ballot paper.
  3. Repeat this process until every candidate has been placed in the first position. With thirteen candidates there will be thirteen ballot paper variations, not millions.
  4. Then, reverse the initial order of the candidates and repeat the above process. With thirteen candidates there will be twenty six ballot paper variations.

Such a procedure can work for any number of candidates and the rotation of candidates ensures that those voters who do donkey vote down the ballot paper will not favour any one candidate. The reversal of the order will ensure that any preferences from such voters will also not favour any one candidate.

Multi-member STV electorates

In Senate and local government elections, the situation is different. The political parties, either formally constituted State or National organisations or just a group of like-minded people in a country town, choose a number of candidates to stand for election. Under fixed order ballot papers these political parties can be assured that their candidates, if elected at all, will be elected in the order of their choosing: most preferred to least preferred.

Candidates would be grouped in party columns, with the number of candidates limited by the number of positions to be filled. However, the number of candidates in a group is self-limiting. In a Senate election, 85.8% of the vote is required to obtain six quotas and 71.5% of the vote to obtain five quotas. To ensure that their vote is not spread too thinly and to limit exhausted votes, the major parties would limit their number of candidates to four. Minor parties would run two or three candidates. Should any party run six, the weaker candidates would be excluded during the count and their preferences would be transferred to the stronger candidates.

The Robson rotation undermines the influence of political parties, especially their organisational wings. The first Robson rotation  resulted from the spat between the political and organisational wings of the Tasmanian Labor Party that Neil Robson exploited to have his proposal carried, and has resulted in a substantial transfer of power from the organisational wing to the political wing of political parties.

Within the PR movement, the Robson rotation has been widely endorsed and little if any study has been given to its implications. But has this transfer of power necessarily been beneficial? Has the pendulum swung too far? Why should the preselectors who know the candidates and know what they want their party to achieve, and in many cases help pay the campaign expenses, be denied any influence in determining who is elected?

The rotation of candidates on the ballot paper – any rotation – is essential to help ensure that the number of candidates elected from popular political parties or groups is maximised. Naturally, any political party, once it is assured of getting its favourite sons or daughters elected, would be very pleased to have even more members elected.

The following rotation helps redress the balance, but does not guarantee that political parties will elect their preferred candidates. This rotation can save the political party from its own stupidity; should they preselect candidate(s) who are clearly unsuitable, voters still have the choice of voting for the party but not the candidate(s).

Recommendation for STV elections

First, each political party supplies a list of candidates in the order that they choose. This is the first ballot paper.

As with current fixed order ballot papers, a sensible order would be from most favoured to least favoured. If the party is confident of electing multiple numbers, there is room for the party to arrange candidates in a different order. The party may, for example, place the only female candidate immediately after the current Mayor. A council team with two popular sitting councillors may run a group of four with the unknown candidates separating them, hoping that any surplus will keep their running mates in the count longer. It is the party’s choice.

Second, take the name of the candidate last on this list and place that candidate’s name first and move every other candidate down one place. This is the second ballot paper.

Third, repeat this process until every candidate has been placed in the first position. There will be as many variations as there are candidates.

By arranging the ballot paper this way, the party is able to promote the leader of the team, perhaps a current Cabinet Minister or a potential candidate for Mayor, and individual voters may choose to vote for this candidate. In many instances, this candidate will gain a quota and be elected. Unless the voter exercises their democratic right to vote differently, preferences will generally flow to the next candidate on the list. This candidate has not been chosen randomly, as with the Robson rotation, but deliberately by the Party: the likelihood of their election is enhanced.

The rusted on voters who don’t know or care who the candidates are but just want to vote for the party will start at the top and vote down the party ticket. Because every candidate in the group has an equal chance of having the top position these votes will be very even. It is these voters who make the rotation system work by spreading the party vote across all the candidates in the group and helping to prevent the last candidate from being excluded early in the count.

The following scenario demonstrates that the rotation of candidates – any  rotation – has  a profound influence on an election’s outcome.

Traditional Fixed Order Ballot Paper
(2 to be elected)

Candidates Vote
A1 1.49
A2 0
B1 0.51
B2 0

After Preferences

Candidates Vote Seats
A1 1.0 quota 1
A2 0.49 quotas 0
B1 0.51 quotas 1
B2 0 0

With traditional ballot papers, and especially with above-the-line voting, the above result gives each party one seat despite the fact that Party A has almost three times the support.

Rotation Ballot Paper (2 to be elected)

Candidates Vote
A1 0.74 quotas
A2 0.75 quotas
B1 0.51 quotas
B2 0

After Preferences

Candidates Vote Seats
A1 0.74 quotas + B1 votes 1
A2 0.75 quotas + B1 votes 1
B1 0 quotas 0
B2 0 quotas 0

Party A received exactly the same number of votes as in the fixed order ballot but this time gained both seats despite all the votes in Party B being concentrated on one candidate. Although this is the perfect split for Party A, all that is necessary is for the weaker of Party A’s candidates to have more votes than the combined vote for Party B.

Local Government

Rotating candidates within party groups works regardless of the number of candidates to be elected. In fact, the more candidates to be elected, the fairer the result. In NSW, many councils elect their Councillors from single wards of nine, twelve and occasionally fifteen. Almost invariably, the last few candidates are elected with the largest remainder(s). Parties with multiple quotas fail to elect their last candidate and candidates with poor support scrape into the council with votes short of a quota.

At the last NSW Local Government elections in Campbelltown (15 to be elected from one ward), a group with 4.4 quotas failed to elect five candidates while three groups, none of which reached a quota, each elected one candidate. The group with 4.4 quotas had 5.7, 5.1 and 4.9 times the vote of the other groups.

Any form of candidate rotation would have ensured, at a minimum, that this group would secure five seats at the expense of at least one of the poorly supported candidates. The nonsense that each group was required to run a minimum of eight candidates did not help. The simple rule is: only candidates who obtain a quota should expect to be elected.

Further discussion

One of the arguments against STV for the House of Representatives is that it is unreasonable to expect busy MPs to have to campaign on local issues within a multi-member electorate because they fear losing their seat, not to the opposition, but to members of their own party.

Why should the Minister for Foreign Affairs, who is busy representing Australia on the other side of the world, have to be back in Bullamakanka to attend the local B & S ball? With single member electorates, this Minister would probably hold a safe seat; he or she would have been unlikely to have accepted the ministerial position otherwise.

Stephen Todd, an electoral reform advocate in New Zealand, has an ingenious and brilliant solution to the problem of intra-party fighting. He proposes that New Zealand should have nine electorates, two in the South Island and seven in the North Island, each returning either thirteen or fourteen members.

To quote Stephen Todd:

I have divided each of the 9 electoral districts into 7 precincts of roughly equal populations. The parties would be allowed to put one candidate of its choice at the head of its list in each precinct and the rest of the names would be listed in random order. That would enable, say, National, to put its leadership at the head of its list in all, or most, precincts in each electoral district in order to attract voters, but the transfer of their surplus votes would be determined by the voters.

This compromise would also allow the parties to promote the candidacies of ‘prominent locals’ or ‘local heroes’. For example, National could put a prominent Māori candidate at the head of its list in the precinct where that candidate resides, to attract Māori voters. In other words, the party leader(s) would not necessarily have to head the list in all 7 precincts of their electoral districts. Again, for example, prime minister John Key might only head the list in 4 of the 7 precincts in the Auckland isthmus electoral district, with other prominent Nationals each heading the list in the other 3. With experience, the parties could get quite creative in the way they allocate the head position to their candidates. [From private correspondence with the author]

In Australia, such precincts could also serve as administrative units for the Electoral Commission, similar to the current single-member electorates servicing Senate elections.

Stephen Todd also points out the advantages that these precincts would give candidates:

[The quota for election would be slightly less than half the turnout in each precinct.] That should assure individual major-party candidates that they don’t need to campaign over the entire electoral district to ensure election; they only need to campaign in their own precinct, plus perhaps the one or two precincts either side / nearby. The candidates can divide up the electorates the way they do in Ireland, and the way Scottish councillors started doing at their 2012 local elections (the second by STV).

Such a proposal does not guarantee election and the voters still have the opportunity to vary their own personal choices. The Australian constitutional requirement that Members of Parliament be directly elected by the voters would still be met.

These precincts also allow Members to develop an affinity with local areas not dissimilar to single member electorates. As there are twice as many Members as precincts, most voters would still be represented, perhaps in their own precinct, but certainly in the broader electorate, by an MP that they actually voted for.

Minor parties can place the head of their team in the lead position in all the precincts. This action does indicate that they believe that they are unlikely to win more than one position but does give maximum prominence to that candidate. In electorates that include both traditional Liberal and National Party areas the Coalition could vary the lead position accordingly.

Electoral Reform Australia would recommend that the subsequent positions in each precinct be rotated in the manner described earlier in this article for STV elections.


Candidates in an STV ballot can be arranged in a number of ways. A fixed ballot order guarantees the order of election of candidates but severely reduces the chances of electing an additional candidate from the group. Groups that fail to reach a quota are the main beneficiaries of this method.

Applying strict Robson rotation rules spreads the vote widely and ensures that second and subsequent preferences are also spread widely. This maximises the number of candidates that will be elected from groups that obtain multiple preferences. Since the party controls much of the campaign’s money and publicity, the first candidate preselected can usually be assured of election, but the election of subsequent candidates, especially in large impersonal elections such as for the Senate, can be a matter of luck.

A third option is to allow the parties to have some influence on the order of the ballot papers. This is a compromise between the other two options. Party members can be satisfied that their inside knowledge, influence and involvement is respected. Supporters of the party can be satisfied that they are able to vary the result should they believe that the party has made an error in the choice of candidates; these same supporters can also be satisfied that the prospects of electing more members is enhanced.

The influence of the organisational wing should neither dominate nor be irrelevant in the election of candidates. It merely needs to be diluted.

Leave a Reply

Your email address will not be published. Required fields are marked *