The Turnbull government’s changes to the Senate’s electoral system were first used at the 2016 double dissolution election, but the 2019 half-Senate election was always going to be a truer test of the changes.
Several of the minor party Senators elected in 2016 owed their seats to the lower 7.7% quota used for double dissolution elections. Two-thirds of minor party Senators were elected to the final vacancies in each state and were allocated to short term seats post-election. Those Senators faced re-election in 2019 when their chances of re-election would be made tougher by the 14.3% quota used at half-Senate elections, as well as the new electoral system.
As this post will explain, the new system worked as designed at the 2019 election, rewarding parties that polled well on first preferences, and disadvantaging parties that relied of harvesting preferences to win election.
The Overall 2019 Senate Result
The 2016 double dissolution election broke the Senate’s staggered terms and saw all Senators face election. The new Senate’s first job after the election was to allocate all Senators to either long (6-year) or short (3-year) terms. The Senate voted to allocate seats based on order of election, resulting in two-thirds of minor party Senators being allocated short term seats.
Disqualifications, re-counts and party defections made minor changes to who faced re-election in 2019 without altering the overall story. Twelve minor party Senators faced re-election, including one Green Senator from each state. There were three minor-party members facing re-election in NSW and two each in four other states. In 54 6-Senator contests between 1990 and 2013, there had been only seven instances of more than one minor party Senator being elected per state. Everything pointed to a winnowing of minor party numbers at the 2019 election.
When the results were in, the Coalition polled 38.0% (+2.8) and elect 19 Senators, a gain of four seats. Labor polled 28.8% (-1.0) and elected 13 Senators (no change), the Greens 10.2% (+1.6) electing six Senators (no change). One Nation polled 5.4% (+1.1) and elected one Senator (down two on the 2016 election before defections). Other parties polled 12.7% (-3.5), electing one Senator (Jacqui Lambie) and losing three seats (Derryn Hinch, one Liberal Democrat, one Centre Alliance turned Independent).
The Coalition won three of the six vacancies in the five mainland states, two in Tasmania and one-each in the two Territories. The Coalition were the beneficiary of crossbench losses in NSW, Victoria, Queensland and WA. The Liberals also gained a seat from Labor in Tasmania, cancelled out by the National Party’s Steve Martin losing to Jacqui Lambie. (Martin had first entered the Senate by re-count on Lambie’s disqualification before defecting to the Nationals.)
Labor’s Senate vote share declined from already low levels in 2016 and 2013, electing two Senators in five states, one in Queensland, and one from each Territory. The party gained seats from the crossbench in NSW and South Australia, but lost seats to the Coalition in Queensland and Tasmania. Labor’s vote fell short of two-quotas in WA and Queensland, its vote so low in Queensland that it lost the party’s usual second seat to the Greens.
Green support fell short of a quota in every state, but easily outpolled Labor in the race for the traditional ‘third’ seat in each state, except for Queensland where the Greens beat Labor for what was the second ‘left’ seat.
The Coalition polled roughly the same vote as Labor and the Greens combined, and Labor plus the Greens won the same number of seats as the Coalition, 19.
One Nation increased its vote, finishing third ahead of the Greens in Queensland, fourth in the other mainland states, and fifth behind Jacquie Lambie and the Greens in Tasmania. One Nation’s Malcolm Roberts was elected in Queensland, taking back the seat that had fallen to Fraser Anning in the re-count following Roberts’ disqualification. (Anning was defeated.) One Nation lost its seat in Western Australia, and lost the seat won by One Nation defector Brian Burston in NSW.
The results illustrated why the Coalition and Greens had backed Senate reform in 2016. In all states the Coalition fell short of three quotas and the Greens did not reach a quota in any state. Had group ticket voting still been in use, both parties could have been run down by the preference harvesting techniques that dogged the 2013 election.
In Victoria, the Liberal Party vote was 4.2 percentage points lower than when the party lost its third seat to Australian Motoring Enthusiast Party candidate Ricky Muir in 2013. Without group voting tickets in 2019, parties could no longer engineer the sorts of preference flows that elected Muir in 2013.
The rest of this post will explain how the result was altered by the operation of the new electoral system. But first, I need to explain a way to measure the impact of preferences on the proportional nature of the Senate’s electoral system.
Preferences and How to Analyse the Senate’s Electoral System
There are many different forms of proportional representation around the world. Australia uses one of the rarest – Proportional Representation by Single Transferable Vote (PR-STV).
Where most of the world’s proportional electoral systems allocate seats based on share of vote alone, PR-STV modifies strict proportionality based on vote share through the use of preferences.
Most of the academic analysis of PR-STV is based on how the system operates in Ireland. The literature identifies two types of preferences to explain the functioning of the system, intra-party or within-party preferences, and inter-party or between-party preferences.
In Ireland, ballot papers list candidates in alphabetic name order in a single column, not grouped together by party as in Australia. Preferences are also optional beyond a first preference in Ireland. Parties tend to nominate only as many candidates as they think can be elected. Parties aim to split their vote evenly between candidates and campaign to maximise preferences between their candidates.
The Australian practice of grouping candidates by party, and requiring a minimum number of preferences, maximises the flow of within-party preferences. For that reason, within-party preferences rarely matter in analysing the Senate’s electoral system. The combined vote for a party’s candidates can usually be treated as a single party total.
The multi-party nature of Irish politics makes between-party preferences important for helping to elect potential coalition partners. Potential coalition partners have an interest in maximising the flow of preferences between the candidates of their parties if a party has a surplus of votes that can’t elect a member. It is between-party preferences that determines who fills the final seats in each constituency.
In the Senate, it is between-party preferences that are critical to determining the balance of power. In each state, four or five seats in each 6-seat contest will be determined by party first preference vote shares. It is the allocation of between-party preferences that determine which party wins the final one or two seats in each state and varies the final result away from proportionality based on first preference vote share.
The group voting ticket system that applied at Senate elections from 1984 to 2013 delivered control over between-party preferences to parties. The new Senate system introduced in 2016 reverted to the pre-1984 position where voters controlled preferences and parties could only try to influence preference flows by campaigning and distributing how-to-vote material. The 2016 changes further weakened between-party preferences by abandoning full preferential voting and allowing preferences to exhaust.
Measuring How Preferences Have Altered Senate Proportionality
Measuring the impact of between-party preferences on the proportionality of Senate outcomes means comparing actual Senate outcomes to non-preferential forms of proportional representation where seats are allocated based only on party vote share.
The simplest model for comparison is list proportional representation (List-PR) using the same Droop quota as the Senate, but using a highest-remainder formula rather than preferences to allocate the final seats.
Under List-PR, all seats are allocated to parties either by having filled a quota for election, or by having the highest remainder amongst the party partial quotas.
Under PR-STV, it is also possible for a party to win a seat from a trailing position on preferences. It is the number trailing party victories that can be used as a measure of the affect of preferences on the proportionality of the system compared to seat allocation based on first preferences alone.
As an observation, the stronger the flow of between-party preferences, the more a Senate result will diverge from a List-PR for an election with the same party vote shares. The weaker the flows, the more PR-STV outcomes will match List-PR outcomes.
Table 1 below compares Senate results since 1977 with possible outcomes under the List-PR model outlined above. The notes below explain the columns in the table.
- Filled quotas – the number of Senators elected through filled quotas based on the total of first preference votes by party
- Highest remainder – the number of Senators elected from the highest remainder or leading partial quota position on first preference votes
- Trailing wins – the number of Senators elected after having trailed the party with the highest remainder on first preferences
- Parties passed – the number of higher polling parties passed by trailing winners.
Table 1 – Senate Results Compared to List-PR with Highest Remainder Allocations
| Election | Seats | Filled Quotas |
Highest Remainder |
Trailing Wins |
Parties Passed |
|---|---|---|---|---|---|
| 1977 | 30 | 24 | 5 | 1 | 1 |
| 1980 | 30 | 25 | 4 | 1 | 1 |
| 1983 | 60 | 54 | 3 | 3 | 2 |
| 1984 | 42 | 33 | 7 | 2 | 2 |
| 1987 | 72 | 62 | 6 | 4 | 6 |
| 1990 | 36 | 28 | 6 | 2 | 2 |
| 1993 | 36 | 30 | 4 | 2 | 2 |
| 1996 | 36 | 28 | 7 | 1 | 1 |
| 1998 | 36 | 24 | 7 | 5 | 5 |
| 2001 | 36 | 25 | 6 | 5 | 6 |
| 2004 | 36 | 29 | 6 | 1 | 3 |
| 2007 | 36 | 27 | 8 | 1 | 2 |
| 2010 | 36 | 27 | 7 | 2 | 4 |
| 2013 | 26 | 21 | 6 | 9 | 33 |
| 2016 | 72 | 52 | 18 | 2 | 7 |
| 2019 | 36 | 22 | 14 | .. | .. |
Source: Calculations by author. Excludes Territory Senator races. 2013 calculations based on original 2013 WA Senate result, not 2014 Senate re-election.
Under group voting tickets at the 2013 election, a quarter of the Senators elected were from trailing positions, and the ratio of parties passed to trailing wins was much higher than at any previous election.
In Western Australia, Wayne Dropulich of the Australian Sports Party was elected despite the party polling just 0.23 percent or 0.016 quotas. The Sports Party finished 21st of 27 parties on first preferences, but received ticket preferences from 20 different parties to achieve a quota, 15 of those parties having polled a higher first preference vote. Preferences allowed Dropulich to leap frog parties and defeat a Labor candidate who began the count with the highest remainder of 0.86 quotas. (You can read about how Mr Dropulich achieved his remarkable victory, and the missing ballot paper disaster that voided the result, in this blog post.)
In Victoria, Ricky Muir of the Australian Motoring Enthusiasts Party began the count with 0.51 percent or 0.035 quotas, receiving preferences from 22 other parties including nine with more votes, and passed a Liberal candidate who began the count with 0.81 quotas. (There is a count by count explanation of Mr Muir’s leap-frogging path to victory in this blog post.)
Another way to look at Table 1 is to plot all trailing party wins as a percentage of the vote for the party that began the count with the highest remainder. That data is plotted below, dots colour coded by party. The graph shows clearly how, in the last few elections where group ticket voting was used, there was an increase in the number of parties elected having started the count with only a tiny fraction of the vote for the party with the highest remainder.
As the data in Table 1 shows, the 2013 half-Senate election was clearly the high-point of preference harvesting, when party control over preferences distorted proportionality based on vote share. As the zero entry for trailing wins at the 2019 half-Senate election shows, the new electoral system reverted the Senate’s electoral system to one where first preference vote share was more important.
Votes Into Seats – Comparing the 2013 and 2019 Half-Senate elections
The 2019 election was the first to be held under the new system with the higher 14.3% state quota for a half-Senate election. It is possible to compare the 2019 results with the result of the last half-senate election under group voting tickets in 2013. As the 2013 election was the high point (or maybe low point) for the manipulation of results by group voting ticket preference deals, the comparison of results is illuminating.
For simplicity I will refer to the Senate system in place from 1984 to 2013 as the old Senate system, and the system used for the 2016 and 2019 elections as the new Senate system.
Table 1 (above) looked at results since 1977. The tables below concentrate more closely on how the old Senate system operated at the high point of preference harvesting in 2013, and how it worked with the new Senate system in 2019.
The comparisons is provided in Table 2 below. Nine Senators were elected from trailing partial quotas in 2013, but no trailing parties were elected under the new system in 2019. With the new electoral system responsible for weaker party control over preferences and a greater number of exhausted preferences, the new Senate system behaved more like List-PR than the old Senate system using group voting tickets.
Table 2 – Comparing the 2013 and 2019 Results with simulated List-PR results
| 2013 Senators Elected | 2019 Senators Elected | |
|---|---|---|
| Filled Quotas | 21 | 22 |
| Highest Remainder | 6 | 14 |
| Trailing Wins | 9 | .. |
Source: Calculations by author. Excludes the four Territory Senators. 2013 calculations based on original 2013 WA Senate result, not 2014 Senate re-election.
More detail on the 14 Senators elected from leading partial quotas in 2019 is shown in Table 3. On the left are the initial partial quotas and parties for successful Senators, while on the right are the partial quotas for the highest polling unsuccessful candidates and parties.
Table 3 – 2019 election – success from partial quotas
| State | Successful Party and Partial Quota | Unsuccessful Party and Partial Quota | |||
|---|---|---|---|---|---|
| NSW | 0.70 LIB | 0.61 GRN | 0.35 ONP | ||
| VIC | 0.74 GRN | 0.51 LIB | 0.20 ONP | 0.20 DHJP | |
| QLD | 0.72 LNP | 0.72 ONP | 0.70 GRN | 0.58 ALP | |
| WA | 0.93 ALP | 0.86 LIB | 0.83 GRN | 0.41 ONP | |
| SA | 0.76 GRN | 0.65 LIB | 0.34 ONP | ||
| TAS | 0.88 GRN | 0.62 JLN | 0.24 ONP | ||
Source: AEC Results, calculations by author. Minor party codes are DHJP – Derryn Hinch’s Justice Party, JLN – Jacqui Lambie Network, ONP – Pauline Hanson’s One Nation.
Of the 15 parties with a partial quota above 0.5 at the start of the count, only Labor in Queensland failed to win a seat. That was a contest where four parties started with more than 0.5 partial quotas in a race for three seats. In the other five States, the partial quota for the next party in order, in each case One Nation, ranged from 0.20 quotas to 0.41 quotas, all well short of the partial quota of the successful sixth party.
It was a very different pattern in 2013 where every State except Queensland saw candidates elected from trailing partial quotas, shown in Table 4 by underlined text. In Victoria the ratio of the lowest elected party to the highest defeated candidate was 0.035 quotas to 0.81, and in Western Australia 0.016 quotas to 0.86.
Table 4 – 2013 election – success from partial quotas
| State | Successful Party and Partial Quota | Unsuccessful Party and Partial Quota | ||||
|---|---|---|---|---|---|---|
| NSW | 0.67 LDP | 0.39 L/NP | 0.55 GRN | |||
| VIC | 0.76 GRN | 0.04 AMEP | 0.81 LIB | |||
| QLD | 0.90 LNP | 0.69 PUP | 0.42 GRN | |||
| WA | 0.74 LIB | 0.66 GRN | 0.02 SPRT | 0.86 ALP | 0.35 PUP | 0.35 NAT |
| SA | 0.92 LIB | 0.50 GRN | 0.26 FFP | 0.74 XEN | 0.59 ALP | |
| TAS | 0.82 GRN | 0.46 PUP | 0.63 LIB | |||
Source: AEC Results, calculations by author. Bold text indicates parties successful from trailing partial quotas. WA result based on the original 2013 Senate result, not on the 2014 re-election. Minor party codes are LDP – Liberal Democrats, AMEP – Australian Motoring Enthusiasts Party, PUP – Palmer United Party, FFP – Family First, XEN – Nick Xenophon Team, SPRT – Australian Sports Party.
Table 5 below shows all successful and unsuccessful groups contesting the 2013 election under group voting tickets compared to the new system at the 2019 election, categorising all parties based on their initial partial quota.
Table 5 – Successful and Unsuccessful Groups based on Initial Partial Quotas – 2013 and 2019 Senate elections
| Number of Successful and Unsuccessful Groups | ||||
|---|---|---|---|---|
| 2013 Election (Old Senate) | 2019 Election (New Senate) | |||
| Initial Partial Quota | Successful | Unsuccessful | Successful | Unsuccessful |
| 0.9 to < 1.0 (14.3%) | 2 | .. | 1 | .. |
| 0.8 to < 0.9 (12.9%) | 2 | 2 | 3 | .. |
| 0.7 to < 0.8 (11.4%) | 2 | 1 | 4 | .. |
| 0.6 to < 0.7 (10.0%) | 3 | 1 | 5 | .. |
| 0.5 to < 0.6 (8.6%) | .. | 2 | 1 | 1 |
| 0.4 to < 0.5 (7.1%) | 2 | 1 | .. | 1 |
| 0.3 to < 0.4 (5.7%) | 1 | 2 | .. | 2 |
| 0.2 to < 0.3 (4.3%) | 1 | 8 | .. | 4 |
| 0.1 to < 0.2 (2.9%) | .. | 9 | .. | 27 |
| < 0.1 (1.5%) | 2 | 159 | .. | 100 |
Source: AEC Results, calculations by author. 2013 calculations based on original 2013 WA Senate result, not 2014 Senate re-election.
As noted earlier, 14 of the 15 parties that began the 2019 count with a partial quota above 0.50 were elected. This contrasts starkly with the 2013 result under group voting tickets. In 2013 there were six parties that were unsuccessful having started the count with more than 0.50 partial quotas. There were six parties successful after starting with a partial quota under 0.50 quotas, including two under 0.10 of a quota.
None of this analysis means that a party can’t win from a trailing position. At the 2016 Senate election, two parties were successful from trailing positions, Family First in South Australia and a second One Nation candidate in Queensland. The last three NSW Legislative Council elections have seen trailing parties win the final seat despite high rates of preference exhaustion.
However, the weaker flow of between-party preferences under the new Senate system means that the gap between leading and trailing parties must be narrower for a trailing party to win.
From the data in Table 5, the third Liberal candidate in Victoria at the 2013 Senate election started the count with 20 times the vote of the Motoring Enthusiasts Party, and in WA, the second Labor candidate had 40 times the vote of the Sports Party. A comparable result in a lower house single member seat would be a candidate on 43% being passed on preferences by a candidate that began the count on 2%.
This analysis confirms that the new Senate system weights the allocation of final seats in favour of parties with the highest remainders on first preferences. This is not the same as advantaging parties with the highest first preference vote. After the allocation of seats to filled quotas, the remaining partial quotas of parties polling above a quota must compete with the initial partial quotas of parties that polled less than a quota on first preferences. Who wins the final seats is then a battle of preferences, with exhausted preferences and weak preferences flows favouring the party with the largest initial partial quota.
Group voting tickets permitted parties to trade preferences in the race to fill final seats. The new Senate system hands the power over between-party preferences back to voters. At the first two elections under the new Senate system, voters made very different decisions on preferences compared to the complex group voting tickets previously lodged by parties.
The clearest between-party trend in preferences in the new system is that voters are more likely to direct preferences to parties they know over parties they don’t know. A party that polls poorly on first preferences tends to attract fewer preferences, while parties with a profile high enough to attract a significant first preference vote, also tend to attract more preferences from excluded parties.
The new system has already had an impact on nominations. There were 227 party or independent groups that nominated for the 2013 half-Senate election. That reduced to 206 groups at the 2016 double dissolution, and fell further to 163 groups for the 2019 half-Senate election.
Much of that decline has come about because parties are contesting fewer Senate races. In 2013 there were 27 parties that contested all six states, but only 11 in 2019. There were nine parties that contested a single state in 2013, and 15 in 2019.
Getting all parties to contest every state or region has been a feature of preference harvesting at Senate elections and Legislative Council elections in Victoria and Western Australia. Parties may have had a primary interest in only one race, but they would join the contest in other races to increase the pool of votes available for harvesting.
That contesting every race is a pointless exercise for micro-parties without group voting tickets is shown by the Senate nomination data, parties now choosing to contest only states where they consider they have a chance of gaining a significant first preference vote.
Under group ticket voting, minor and micro-parties could compete against each other and swap preferences, secure in the knowledge that group ticket votes could aggregate their vote, allowing them to leap-frog higher polling parties. The old system had no disadvantage for like minded parties that fragmented their vote. Group ticket voting provided no urge to merge for like minded parties.
The new Senate system punishes parties with small votes that compete in the same sector of the political spectrum. It seems that voters might give a first preference to a party they don’t know, but the last two elections show that voters then tend to preference the parties they know.
The question for future elections is, will the 2019 result produce a decline in the formation of so-called micro-parties. Will there be a consolidation of like-minded micro-parties into slightly larger minor parties?
If party consolidation produces a smaller ballot paper with a more recognisable array of parties, then voters will surely be able to make better informed choices.
Sources: This post is based on my chapter on Senate results for the forthcoming ANU Press book on the 2019 Federal Election. It also draws on submissions to the Victorian Parliament’s Electoral Matters Committee, to the Commonwealth Parliament’s Joint Standing Committee on Electoral Matters, and on a speech I gave recently to the WA Branch of the Australian Studies of Parliament group.
p.s. Apologies to Glenn Kefford. Yes, I’ve mixed micro and minor parties without defining the difference between the two. One day I’ll stop doing it.