Victoria is the only Australian jurisdiction that continues to elect its upper house using the discredited Group Voting Ticket (GVT) system.
GVTs in Victoria give parties almost total control over the distribution of preferences, which flows through to controlling who wins the balance of power in the Legislative Council.
GVTs have been abolished in every state and for the Senate because they can be manipulated to elect parties with only a tiny percentage of the vote, a result that distorts the intended proportionality of the chamber’s electoral system.
In the lower house voters control preferences. Parties and candidates can only try to influence voters in how they complete their preferences. It is the same for the reformed Senate electoral system where voters now control the flow of between-party preferences, not parties.
As I explain in this post, the rottenness of GVTs is revealed when you examine the proportion of ‘above-the-line’ (ATL) votes that are under party control at Victorian Legislative Council elections compared to the related data for non-GVT Senate elections.
In Victoria, 100% of every ATL vote for every party, whether big or small, flows according to the party ticket.
In contrast, the 2022 Senate election saw major parties lucky to influence even the second preference of 50% of ATL votes, and the rate dropped precipitously for smaller parties.
It is without doubt that the reformed Senate system delivers an outcome that reflects the preferences of voters, where in Victoria the use of GVTs means the result reflects the decisions made by the tiny cabal of officials who negotiate the preference deals.
Electoral System Decision Rules
All electoral systems have a set of rules that determine how votes are translated into seats.
For simple majority voting in single member seats, the rule is the candidate with the most vote wins. (This is often incorrectly called ‘first past the post’ as an analogy with horse racing, but horse racing has a fixed winning post which must be passed, where there is no fixed winning post in simple majority voting.)
Under full preferential voting used for Victorian and Federal lower house single member elections, the winning candidate is the one that achieves 50% of the vote after the distribution of preferences. There is a fixed winning post.
Under optional preferential voting used in NSW, the winning post becomes variable as the winner needs only 50% of the vote remaining in the count, that is total votes minus ballot papers with exhausted preferences.
Where proportional representation is used in Australia, we use a fixed quota method of allocating seats. The first seats are declared with quotas filled on first preferences. The allocation of final seats is determined by distributing preferences to fill quotas.
It is important for the discussion below to understand that Australia’s single member and multi-member systems use the same quota formula to decide winners. The formula is called a ‘Droop quota’, and it is the (total formal vote) divided by the (number of vacancies to be filled plus one)
In the Victorian Legislative Council, there are five vacancies per region so the divisor is (5+1)=6. Expressing the quota for election in percentage terms, the quota needed for election to the Legislative Council is (100% divided by 6) or 16.7%.
In the lower house, there is only one vacancy so the divisor is 2 and the quota 50%. (50% + 1 for the pedants).
So both single and multi-member preferential systems are the same in using a fixed quota, and where no candidate has a quota, the allocation of the next seat (in single member contests the only seat) is done by distributing preferences.
The only difference between the two systems is that only lower polling candidates are excluded and have preferences distributed in single member seats. In multi-member systems, candidates with more than a quota of votes can also be a source of preferences, what are called “surplus to quota preferences”.
A Simplified Decision Tree for Predicting Seats Under Preferential Voting
The following decision tree is a model for predicting whether the leading candidate can win a seat in a preferential voting race. Note that I am talking about a prediction here, not how the votes are counted. The decision tree can be used for predicting a seat in a single member contest, or for predicting a succession of seats in a multi-member contest. The decision tree is –
- (1) Does the leading candidate have a quota? – candidate is elected
- (2) Leading candidate does not have a quota – Their chance of election depends on –
- (2a) – How close is the leading candidate to the quota
- (2b) – What is the leading candidate’s lead over trailing candidates
- (2c) – What are the preference flow rates against the leading candidate.
If no candidate has a quota, then the most important consideration is (2c) preference flow rate. (2a) and (2b) matter because in most cases preferences do not approach 100% against the leading candidate. There will be enough leakage for (2a) and (2b) to be important.
But if preference flows are 100% against the leading candidate, then (2a) and (2b) are irrelevant. It doesn’t matter how close to the quota a leading candidate is, or how far they are ahead of trailing candidates, 100% preference flows will deliver victory to the trailing candidate.
Here’s a single member contest example.
- Candidate A – 50 votes
- Candidate B – 26 votes
- Candidate C – 25 votes
- Total 101 votes, votes needed for election 51
If candidate C’s preferences flow 100% to candidate B, then Candidate B wins. Candidate A was only one vote short of winning, and only needed one preference vote from Candidate C to win, but a 100% preference transfer elects Candidate B. How close A was to a quota, and the size of A’s lead, became irrelevant at 100% preference flows.
This extreme example explains why GVTs distort election outcomes. GVTs allow preferences flows to approach 100%, and once they do, it doesn’t matter how close to a quota the leading party is, or what their lead is, preference flows approaching 100% can allow the trailing candidate to win.
ATL Voting Rates at the 2018 Victorian Legislative Council election
The table below shows the percentage vote by party at the 2018 Victorian Legislative Council election, and the proportion of each party’s vote cast as an ATL vote and allocated the party’s GVT preferences. The ATL% figure is effectively the number of votes cast for a party that were ATL votes and therefore flowed 100% according to the party ticket.
Overall 91.2% of votes were ATL votes tied to a GVT. Even a small party such as the Australian Country Party with only 0.68% of the vote could deliver 91.7% of its votes to other parties as preferences.
That so many votes were tied to tickets explains how some parties elected members from ridiculously low percentages in 2018.
- In Eastern Metropolitan Region, the Transport Matters Party polled 0.62% of the vote but won a seat ahead of the Greens on 9.0%
- In South-Eastern Metropolitan Region, the Liberal Democrats on 0.84% won a seat defeating the second Liberal candidate who began with 12.3%
- In Southern Metropolitan Region, the Sustainable Australia Party with 1.32% defeated the Greens on 13.5%.
Compare this with the result of the 2022 Victorian Senate election. Overall 94.8% of votes were counted as ATL votes and 5.2% as BTL. But without GVTs, all of those ATL votes had their own preferences. The between-party preferences of ATL votes were determined by voters.
The table below shows how this makes a difference. For all ATL party votes, I have accumulated and calculated percentages for the strongest second preference flow to another party.
At the 2018 Legislative Council election, all ATL votes delivered 100% preference flows to a 2nd preference by the party's GVT. In fact, the flow was 100% to every preference in order of the lodged GVT. If I re-calculated the table below to show the highest rate of 1,2,3 preferences, or 1,2,3,4 preferences, all the percentages would be smaller. Voters do not complete preferences that look like party GVTs.
The Liberal/National total had 97.2% ATL votes, and of those 41.2% had a second preference for the UAP, in line with the party's how-to-vote. Had GVTs been in use, that flow would have been a party determined 100%.
The Labor total had 96.6% ATL votes, and of those 54.8% had a second preference for the Greens, in line with the party's how-to-vote. Had GVTs been in use, that flow would have been a party determined 100%.
The Greens total had 91.5% ATL votes, and of those 41.5% had a second preference for Labor, despite the party's how-to-vote recommending a second preference for the Socialist Alliance. As the Socialist Alliance were excluded early in the count, Green preferences would have skipped through to Labor eventually.
The Liberal Democrat total had 90.9% ATL votes, and of those 20.7% had second preference for the United Australia Party, in line with the party's how-to-vote. Had GVT's been in use, that flow would have been 100%.
A look at the final counts in the 2022 Victorian Senate election shows the real impact of GVTs. When only three candidates remained in the race, The UAP led with 0.64 quotas to Labor on 0.55 quotas and the third Liberal 0.50. When the final Liberal candidate was excluded, Liberal ATL votes split 55.3% to UAP, 24.5% to Labor and 20.2% exhausted and the UAP's Ralph Babet was elected.
As the UAP's Ralph Babet was already leading before this distribution, Liberal preferences did not decide the outcome. But had GVTs been in use, and with 97.2% of Liberal votes being ATL, that final preference distribution would have seen 97.2% of Liberal preferences flowing to Babet.
This is a clear example of why the Victorian Legislative Council's electoral system using GVTs is rotten. Instead of voters controlling preference transfers, the GVTs put that power into the hands of party bosses who negotiate preference deals. Half-a-dozen MLCs elected at the 2018 Legislative Council election would have been defeated had voters controlled preferences rather than the party GVTs.
Remember you can't preference above the line in Victoria
Since the last state election in 2018, Victorians have voted at two Senate elections where the ballot paper instructions were to mark at least six square for an ATL vote. 92.1% of Victorian Senate ballot papers, around 3.5 million votes, had ATL preferences marked.
ATL preferences are not counted in Victoria. Every voter who uses the Senate method of voting will have their second and further ATL preferences ignored. The ballot paper will instead be allocated the GVT preference ticket of their first preference party. This means a different set of preferences will be applied other than the ones indicated by the voter.
There is only one way for a voter to give their own between-party preferences.
Vote below-the-line to control between-party preferences
Victoria at least has an easy way of casting a BTL vote. Only the numbers 1 to 5 need to be marked for a formal BTL vote. And you can vote for any candidate in any party in any order you like. And your vote counts as long as you have at least the first five preferences.
At Senate election, you can control between-party preferences with an ATL vote and only need to cast a BTL vote if you want to re-arrange the candidates.
In Victoria you cannot control between-party preferences with an ATL vote, the only option is voting below the line.
The more people vote below the line, the more that complex preference deals are unpicked, and the harder it becomes for parties with tiny votes to pass higher polling candidates and get elected.