Preferences

How GVTs Have Distorted Victorian Legislative Council Elections

Summary: Change is coming to the Victorian Legislative Council’s electoral system for November’s state election. Victoria becomes the last jurisdiction to abolish group voting tickets, a system that allowed parties to control preference flows and distort election results. Voters will now control between-party preferences and the method of voting mimics the Senate system in use since 2016. The change favours parties that campaign for votes over parties that rely on preferences. Based on past results, the big losers will be minor parties whose election from low votes has been achieved by “preference harvesting” tactics, something that is impossible with group voting tickets abolished.

The Victorian Labor government’s bill to abolish group voting tickets (GVTs) for Legislative Council elections was introduced on Tuesday 28 July, passed the lower house the next day with no opposition, and will be debated in the the upper house on Tuesday 11 August.

The bill is supported by the government, opposition and Greens and will pass by week’s end. The new voting rules will be in place for November’s state election.

(Update 9:30pm 11 August: the bill passed the upper house without amendment. The next step is royal assent by the Governor at which point the amendments in the bill are incorporated into the Electoral Act.)

Inside this post I look back at the five Victorian elections that have used GVTs and estimate how the results might have looked without GVTs. In summary –

  • The new system will favour parties that attract significant first preference support and disadvantage parties that rely heavily on preferences. Based on my analysis of post-GVT Senate results, (see this post), weaker preference flows will make it very difficult for a party to be elected with less than 6% in a region.
  • Across the 2014, 2018 and 2022 elections, 24 leading candidates who began the preference distribution with more than 6% of the vote were defeated by lower polling candidates. Three of the defeated candidates polled above 12% or 0.7 quotas.
  • At the same elections there were 20 candidates elected from under 6%, including nine that polled below 3%.
  • Without GVTs, the 2014 election would have seen five fewer minor party MLCs elected, the seats instead going one Labor, two Coalition and two Greens.
  • Without GVTs, the 2018 election would have seen seven fewer minor party MLCs elected, the seats instead going one Labor, three Coalition and three Greens.
  • Without GVTs, the 2022 result would have seen six fewer minor party MLCs elected, the seats instead going one Labor, one Coalition and four Greens.
  • Over the last three elections, the number of non-Green minor party members would have fallen from 22 to 4. The number of Greens elected would have risen from 10 to 19, Labor from 47 to 50, and the Coalition from 41 to 47.

Table 1 below shows a four party breakdown of seats won at elections since 2006, along with my estimated results had there been no GVTs. My estimates assume the same parties contested election and the same party vote shares were recorded. I explain my method for estimating results inside the post.

Table 1 – Legislative Council Seats 2006-2022 – With and Without GVTs
Election Alternatives ALP L/NP GRN OTH
2006 Actual result 19 17 3 1
Estimate without GVTs 20 17 3 ..
2010 Actual result 16 21 3 ..
Estimate without GVTs 16 21 3 ..
2014 Actual result 14 16 5 5
Estimate without GVTs 15 18 7 ..
2018 Actual result 18 11 1 10
Estimate without GVTs 19 14 4 3
2022 Actual result 15 14 4 7
Estimate without GVTs 16 15 8 1

Source: Estimated results without GVTs calculated by Antony Green as set out in the post.

The changed results are due to weaker preference flows with the abolition of GVTs. Preference ‘deals’ implemented by GVTs were previously used to engineer results. Even the smallest parties could deliver preference flows above 90%. GVT deals could deliver victory to parties with little electoral support at the expense parties with up to 14 times the vote. These remarkable come from behind victories were created by GVT preference deals, not by the actions of voters.

Three of the more remarkable results occurred in 2018. In Eastern Metropolitan Region, the Transport Matters Party was elected from 0.62%, defeating a Greens candidate who began the count with 9.0%. In Southern-Eastern Metropolitan Region, the Liberal Democrats polled just 0.84% but defeated the second Liberal candidate who began the count with 12.3%. In Southern Metropolitan Region, the Sustainable Australia Party polled 1.32% but defeated the Greens on 13.5%. (Data in this paragraph has been corrected.)

Results such as these could only happen under an electoral system where voters are herded into voting for above-the-line for tickets with preferences controlled by parties.

A decade of experience with Senate elections since GVTs were abolished, and evidence from Legislative Council elections in other states, shows that preference flows are much weaker without GVTs. Larger parties that campaign actively can still influence preference decisions and create stronger preference flows, but the flows of smaller parties become almost random. This decreases the ability of parties to win from trailing positions.

When voters control preferences rather than parties, the system rewards parties that attract first preference votes over parties that rely on opaque and labyrinthine preference deals. The days of preference ‘whisperers’, preference ‘harvesting’ and ‘corralling’ preferences disappear when GVTs are abolished. Voters first preference party choices play a greater role in determining which parties elect members.

But I need to stress that the analysis in this post is about the parties that have contested the last five Victorian Legislative Council elections, not the parties contested the 2026 election.

It is clear from polls that Pauline Hanson’s One Nation will poll strongly, will elect MLCs, and probably outpoll the Greens. I make no comment on whether abolishing GVTs increases or decreases the number of MLCs that One Nation might elect.

What is certain is that abolishing GVTs will see the number of MLCs elected by party more accurately reflect vote share by party. The new system will better translate the will of the electorate into representation than any election conducted with GVTs.

Read More »How GVTs Have Distorted Victorian Legislative Council Elections

The Impact on Senate Election Results of Abolishing GVTs – the Lessons for Victoria

Over the next week I will publish several posts looking at the abolition of Group Voting Tickets (GVTs) in Victoria and the implications this has for the state’s Legislative Council election in November.

Work I’ve undertaken in the past, about Senate elections both before and after the abolition of GVTs, produces clear conclusions.

Abolishing GVTs will increase the importance of first preference votes and diminish the value of preference transfers. Abolishing GVTs advantages parties that campaign for votes and disadvantages those that rely on deals and transfers to be elected.

Abolishing GVTs also ends the ability of parties to conspire and use GVTs to block the election of high polling parties. It also stops parties using GVTs to elect low-polling parties that few voters have heard of into important upper house balance of power seats.

The key aim of abolishing GVTs is to end party control over between-party preferences and leave it to voters to make the decision. With 95% of votes tied up in party tickets, too many elections had been decided by opaque preferences deals that voters had little knowledge of. A new form of above-the-line (ATL) voting was created allowing voters to show preferences for parties rather than candidates. Voters still had the option to vote below-the-line (BTL) for individual candidates of any party.

In other states this has been implemented with full optional preferential voting. A voter must indicate a first preference for a formal vote, but further preferences were optional. At elections conducted under the new rules, only a third of ATL ballot papers have preferences beyond ‘1’ and exhausted preferences have come to play a part in deciding who wins the final seats. (See my previous post that covered experience with the change in NSW.)

By contrast, the Senate instructions after the abolition of GVTs has been to mark at least six preferences, though a savings provision allows any first preference vote to remain formal. Under these instructions, around 95% of ballot papers are completed above the line (ATL) with between-party preferences. Given the exhaustion rate of preferences for minor party votes, there hasn’t been much difference between the role of preferences in the two ballot marking systems.

Abolishing GVTs makes it nearly impossible to win from a low vote AND from behind on preferences. Abolishing GVTs does not prevent a party winning from a low vote, nor prevent a trailing candidate from winning on preferences, though it does make it harder. But is becomes extremely difficult to do both at the same time.

A party winning from a low vote is possible without GVTs if there is an even split of vote between remaining parties. The United Australia Party’s Ralph Babet won a Victorian Senate seat in 2022 from only 0.28 quotas (4.0%). This was because there were five parties between 0.20 and 0.28 quotas competing for one seat. Without GVTs, preference flows were relatively weak and Babet maintained his lead to win the final seat. Preferences were not responsible for his victory.

And coming from behind is possible. In NSW at the 2025 Federal election, strong preference flows from the Coalition and small right-wing parties allowed One Nation on 0.42 quotas (6.1%) to pass Labor’s third candidate on 0.63 quotas (9.1%).

Without GVTs, the classic 2013 Victorian Senate case of Ricky Muir (Australian Motoring Enthusiasts Party) on 0.5% defeating Helen Kroger (Liberal) on 11.6% would be impossible. Muir’s election was the equivalent of a lower house candidate on 1.7% defeating a candidate with 40%.

As One Nation showed in 2025, is is possible for a candidate with two-thirds of another candidate’s vote to come from behind and win on preference transfers. Voters still have their say on preferences, but without GVTs, results can’t be engineered. Muir’s case of winning from behind with only one-23rd of the winning candidate’s vote is absolutely impossible without GVTs.

And to date, every jurisdiction except except Victoria has understood this inherent problem with GVTs and abolished them. Next week it appears Victoria will be the last to join the club.

Based on One Nation’s two come from behind Senate victories in 2025, I think around 0.4 quotas is the minimum a party needs for victory, 6.5% translated into Victorian Legislative Council numbers. That’s unless you get a case like Ralph Babet’s where a seat remains to be filled and all remaining parties have a low vote.

Looking at how many parties have been elected from trailing positions, the difference produced by the abolition of GVTs for Senate elections is clear.

At the 2013 half-Senate election, the last to use GVTs, 9 of the 36 state Senate positions were won by trailing candidates. At three half-Senate elections since the abolition of GVTs, only 2 of 108 state Senate seats have been won by trailing candidates.

And both of those trailing wins were aided by Senate instructions to mark 6 preferences. Trailing wins may be even harder under instructions for ‘1’ with further preferences optional.
Read More »The Impact on Senate Election Results of Abolishing GVTs – the Lessons for Victoria

Reports that Victoria is set to Abolish Group Voting Tickets

(Update Wednesday 29 July: – Legislation has been introduced today and will be debated and passed by the Assembly this afternoon. The legislation abolishes group voting tickets and adopts a similar method of ATL voting to the Senate. Voters are required to mark 5 ATL preferences and there is a savings provision so that any vote with at least a valid first preference vote remains formal. There is no change to the Victorian method of BTL voting where voters are required to complete preferences for five candidates.)

Two weeks ago I published a post highlighting that if Victoria wanted to abolish group voting tickets (GVTs) for the 2026 Legislative Council election, legislation had to be passed at the next sitting of parliament in the last week of July.

As I outlined in the post (you can find it at this link), abolishing GVTs would bring Victoria in line with the Commonwealth, New South Wales, South Australian and Western Australian parliaments. It would also implement a recommendation made twice in the last four years by the Victorian parliament’s Electoral Matters Committee.

Both The Age and Herald-Sun reported over the weekend that GVTs will be abolished and voters allowed to indicate their own preferences with an above-the-line (ATL) vote. What the reports don’t make clear is whether Victoria will adopt fully optional preferential voting as used in NSW, SA and WA. The alternative would be to copy the Senate model where ballot paper instructions are to complete a minimum number of ATL preferences. The Senate model still allows any vote with a valid ATL first preference to remain formal in line with ballot paper instructions for the GVT system.

The number of members to be elected per region may lean the Victorian decision in favour of the Senate system.

NSW, SA and WA elect their Legislative Councils using a state-wide electorate. NSW elects 21 members, South Australia 11, and Western Australia 37. A single ‘1’ vote in these systems is enough to produce a very proportional result. Over six elections since the NSW system was reformed, only twice have preferences altered the party elected as the 21st member. Two SA elections elections have seen preferences play no part in the result, while the first WA election after reform saw preferences change only one of the 37 members elected.

The complication in Victoria is members are not elected state-wide. The Legislative Council consists of five members elected from each of eight regions. The quota for election is a high 16.7%. A high quota makes it much harder to match the number of members elected by party in each region to the proportion of votes by party.

As with the Senate, where six members are elected per state, there may be an argument to encourage more ATL preferences to ensure the members elected reflect first preference votes and the preference intent of voters for excluded parties.

The Senate rule to mark at least six preferences encourages voters to complete preferences. At the 2025 Victorian Senate election, 83.4% of voters marked six preferences. Only 1.9% of voters marked a single first preference, a vote that remains formal under Senate voting rules.

Contrast this with lower preferencing rates at interstate Legislative Council experience where only a third of ballot papers have preferences. If fully optional preferential voting were adopted, then clearly there would be fewer preferences playing a part in determining the final seats in each region. It is a matter of opinion whether this is a good or bad outcome.

Fully optional preferences would work without difficulty if Victoria were electing 40 MLCs from a state-wide electorate. But Victoria can’t switch to state-wide election without a referendum, so the five members by eight regions model will be used in 2026.

In this post I’ll first summarise why the GVT system is so wrong using past election results. I’ll also look at evidence from NSW elections about how the Senate’s minimum preference numbers rule has caused more voters to complete further preferences at Legislative Council elections.Read More »Reports that Victoria is set to Abolish Group Voting Tickets

The Financial Review’s Inaccurate Preference Analysis

One Nation’s polling surge has triggered an intense dissection of recent election results.

One Nation made the final preference count at the recent Farrer by-election. One Nation also finished in the final preference pairing in more than half of South Australian electorates at the state election in March. These results offer a mine of data where One Nation is a recipient of preferences rather than a provider.

Trying its luck dissecting the results on 7 June was the Australian Financial Review. It published an analysis piece on Farrer and South Australia headlined “How One Nation is winning the preferences war (with the Greens’ help)”.

As I will outline in this post, the article has confused published preference distribution data with more accurate but unavailable preference flow data. And the remark about Greens preferences isn’t backed by my analysis of the data.

The article was published three days before the Farrer flow data was released and the South Australian analysis has been attributed to preference flow data that is not available.

With the South Australian results, the article uses preference transfer data to draw some odd conclusions about Green preferences, and illustrates flows for other parties that aren’t justified given the quality of the data.

That the analytical problems, which I’ll explain in this post, have passed without much comment from anyone is due to the published chart included below. The chart tries to show SA preference flows from other parties to One Nation.

The chart’s meaning and data source are hard to unravel, but mirth has been the main reaction due to the inclusion of imaginary party names. It has caused the AFR to publish a correction and remove the chart from the on-line article.

LCP, an acronym for the Legalise Cannabis Party, is rendered as the Liberal Catholic Party. FGA, for Sarah Game’s Fair Go Australia, is transformed into the Family Guardian Association.

To the best of my knowledge, neither party name has ever been used in Australia.

It looks to me like a case of AI ‘hallucination’, AI asked to convert codes into party names and guessing it wrong. I doubt a human made those errors.

But enough cheap shots over party names. What I really want to explain is the article’s problems with data source and analysis.

With Farrer, I want to explain how the chart used to demonstrate the composition of One Nation’s winning vote uses the wrong data. It results in an incorrect emphasis on whose preferences were responsible for One Nation’s final margin.

With South Australia, the available data cannot be used to isolate the preference flows shown in the now deleted AFR chart. And the data that is available does not support the headlined conclusions about Greens preferences.

In both cases the problem comes down to limitations on how you can use preference transfer data from published distributions. The data must be treated with caution and can’t be automatically compared with more accurate preference flow data.Read More »The Financial Review’s Inaccurate Preference Analysis

Do Green Voters Follow How-to-Votes?

Headline Summary – Strong flows of Green preferences to Labor are often cited as being due to preference “deals”. When you examine the data, the flows are due to the instinctive inclination of Green voters to put Labor second. Even in seats where the Greens recommend preferences to other candidates before Labor, many Green voters still preference Labor. It’s how Labor narrowly defeated Independent challengers in Fremantle (WA) and Bean (ACT) at the 2025 election.

One of the legacies of Senate group voting tickets (GVTs) is that political observers continue to talk about preferences as “deals”. Show and tell preference meetings and parties registering preference tickets at the same time were all part of Senate deal making with GVTs. And those deals stuck as GVTs could deliver preference flows above 90% by taking control of a voter’s ballot paper.

The GVT system was rightly abolished in 2016, ending deals that guaranteed preference flows. The Senate is now like the House where parties can only recommend preference, not direct them. With the exception of one brief experiment in Western Australia (see my previous post on the subject), lower house preferences have always been about recommendation rather than control.

What might be termed a “deal” is no more than agreement on printing preference recommendations. This can occur without an agreement as parties often have mutual self-interest in recommendations. For obvious ideological reasons, parties and candidates tend to recommend preferences within left or right blocks rather then recommend preferences across the political divide.

And whatever negotiations do occur, they have no impact unless most voters receive a party’s how-to-vote, and that voters choose to follow the recommendation. Voters can have their own views on where the preferences of their vote should flow, whatever is on a how-to-vote.

Many voters never see the how-to-vote of their first choice party, especially with minor parties. Without seeing a how-to-vote, a voter’s preferences can only match a how-to-vote by guesswork. Follow rates are always higher with fewer candidates on a ballot paper and when a how-to-vote recommendation follows an understood ideological order.

There are discussions between party representatives about preferences. But even without “deals”, it still makes sense for Labor and the Greens to recommend preferences for each other before parties of the right. It still makes sense for the Coalition and small right-of-centre parties to recommend preferences for each other before the left.

With the Coalition always attacking Labor for working with “radical left-wing Greens”, it also makes sense for the Coalition to recommend preferences for Labor before the Greens. That said, the Coalition has at times been flexible on whether to preference Labor or the Greens.

So what about the Coalition and One Nation? That may be a current hot topic, but it’s not the purpose of this post. All I’ll say for now is that strategy, tactics and practice on Coalition-One Nation preferences have changed depending on political circumstances.

This post looks at data on whether voters follow Green preference recommendations. At the 2025 Federal election, 88% of Greens preferences flowed to Labor before the Coalition. For the past 15 years, the percentage flow of Green preferences to Labor has drifted higher as shown in the chart below.

Increased candidate numbers may have removed some “random” Green votes cast as an objection to the major parties. I believe Coalition attacks on the Greens may have led to a decline in conservative voters giving a first preference to the Greens. Not being Greens helps explain why ‘teal’ Independents have had success attracting former Liberal voters.

The last time the Greens issued open how-to-votes without a recommendation to Labor was in 2010. It made little difference to the flow of preferences to Labor as I wrote about in 2011.

What about if the Greens recommended preferences for the Coalition ahead of Labor? Unfortunately data is very thin on the ground, down to a single case, the 2020 Johnston by-election for a seat in the NT Parliament. Even that was a preference throw between Labor and the short-lived Territory Alliance rather than the Country Liberal Party and was due to NT Labor policy on fracking. Preferences flows were much weaker but still 56% to Labor.

But the Greens putting Labor ahead of the Coalition does not mean Labor is placed second. As I will discuss in this post, there were several examples at the 2025 election which show that strong preference flows from Green candidates is not just a matter of voters following a how-to-vote.

If you look at Coalition-Independent contests, Green preference flows match those seen in Labor-Coalition contests. But the path by which Green preferences reached Independents was not a simple case of following a how-to-vote. And the situation is even more revealing in Labor-Independent contests with Labor retaining two seats in 2025 due to Greens voters making up their own minds on preferences.
Read More »Do Green Voters Follow How-to-Votes?

FED2025 – Independent Two-Party Preferred Preference Flow Data

The 2025 House of Representatives election was the first where more than a third of primary votes were cast for ‘Others’.

In total, minor parties and Independents recorded 33.6% of the primary vote. This was more than 31.8% for the Coalition and only one percent short of Labor’s vote.

The rise in support for ‘Others’ has changed the nature of seat contests at House elections. Elections today are far removed from the days when almost all seats finished as two-party preferred contests between Labor and the Coalition.

The word ‘classic’, which I’ve always associated with vintage cars and ice cream marketing, has been adopted by the Australian Electoral Commission (AEC) to describe traditional Labor versus Coalition contests. Or rather the AEC calls all other combinations of final candidates after preferences as ‘Non-Classic’, which it helpfully provides a list of on its election results website. (2025 list here)

In this post I’m providing information on two-party preferred preference flows in these Non-Classic seats, data that has not previously been available. This includes preference flow splits between Labor and the Coalition from votes cast for Independents in Non-Classic contests.

Until 1983, preferences were only distributed in seats where no candidate reached 50% of the primary vote. The AEC began to conduct distributions in all seats for ‘information purposes’ in 1984. That election saw every seat finish as a two-party preferred race.

Every election since has seen at least one Non-Classic contest. And as the chart below shows, the numbers have ballooned at the last four elections.

The chart above is based on all Non-Classic contests. The last three elections have seen a particular surge in these contests involving Independents. In 2019 there were seven final counts that involved Independents, rising to 16 in 2022 and 20 in 2025.

The AEC continues to do full distributions of preferences to the final pairing of candidates in every seat. Since the 1990s it has also undertaken alternative two-party preferred counts in every seat. (Except in Bradfield in 2025 for reasons explained later.)

The AEC counts preferences in two ways. First is by conducting the legally required full distribution of preferences in every seat to produce a two-candidate preferred count. Second, the AEC tallies what are called ‘preference flows’ in every seat for both two-candidate and two-party preferred finishes. I’ll describe how preference flows are calculated inside the post.Read More »FED2025 – Independent Two-Party Preferred Preference Flow Data