I’ve written before about the peculiar way that single transferable vote works differently with districts electing an odd number of representatives, as opposed to an even number.
You particularly notice it in local councils with 2-member or 4-member wards, but it is also an issue at federal elections, with the Senate electing six members per state per election. In this post I’m going to explore how six-seat elections have worked, and how that has flowed through to who has control of the upper house.
The Senate was expanded from ten senators per state to twelve prior to the 1984 election, but we didn’t have a “normal” half-Senate election with six seats per state until 1990. The 1984 election saw the election of seven seats per state to bring the total up to the new number, while 1987 was a double dissolution.
If you take a simplified model of a single transferable vote election, you need about 42.9% to win three seats in a state (three quotas of 14.3%), but you need 57.1% to win four seats. That is a big gap. A party could gain a substantial swing and not change that pattern.
The domination of the two major parties was already significantly weakened by 1990, with the Democrats regularly winning a number of seats.
While the Democrats had started out as a centrist party founded by a former Liberal MP, by 1990 they tended to win one of the three “left” seats. From 1990 until their final wins in 2001, the Democrats won twenty seats. In seventeen of these cases, the other seats split as three Coalition and two Labor. Only two were 3 Labor and 2 Coalition. The final one involved a One Nation win.
For reasons that will become clear further on, this first table stops at the 2010 election. The 2013 election bucked the previous trends.
This table shows the number of cases of each type of result at the 48 state contests in the Senate from 1990 until 2010.
| Split | Cases |
| LNP3-ALP2-DEM1 | 17 |
| LNP3-ALP2-GRN1 | 12 |
| ALP3-LNP3 | 8 |
| ALP3-LNP2-IND1 | 2 |
| ALP2-LNP2-DEM1-ON1 | 1 |
| ALP2-LNP2-GRN1-DLP1 | 1 |
| Even 3-3 | 41 |
| ALP3-LNP2-DEM1 | 2 |
| ALP3-LNP2-GRN1 | 2 |
| Left 4-2 | 4 |
| LNP4-ALP2 | 1 |
| LNP3-ALP2-FF1 | 1 |
| Right 4-2 | 2 |
| ALP2-LNP2-GRN1-XEN1 | 1 |
| Left 3-2-1 | 1 |
The vast majority of cases (41/48, or 85%) involved three wins for the left and three wins for the right. This included 17 cases where the Democrats won one of those three left seats, and 12 where one seat went to the Greens. There were also 8 contests where every seat went to a major party, as well as two contests involving Brian Harradine (a right-wing independent) as well as single victories for One Nation and the Democratic Labor Party.
The left managed 4 seats on just four occasions. The Democrats won a seat alongside three Labor in 1990 and 1998, both in New South Wales. The 1990 election saw Fred Nile’s party withhold preferences from a particular moderate Liberal senator, which ended up seeing Nile’s preferences electing a third Labor member instead.
The left then managed 4-2 splits in Tasmania in 2007 and 2010.
The only two right 4-2 splits in this period both took place in 2004. The Liberal and National parties ran separate tickets in Queensland that year, with the Liberal Party winning three seats and the Nationals’ Barnaby Joyce winning a fourth seat for the Coalition. Family First candidate Steve Fielding also won a seat that would have been expected to go to the left in Victoria.
During the 1990-2010 period, the only race I couldn’t define was the 2007 South Australian result, where the left won three seats, the right two, and Nick Xenophon won the other seat. But the 2013 election elected a lot more senators that are hard to define in either group.
With most races producing an even 3-3 split, this makes those 4-2 splits crucial to determining the overall balance. That will become clearer after we look at this next chart, which shows a simplified version of the result for every 6-seat result from 1990 to 2022.
Since Senate terms last for six years, the overall make-up of the Senate is determined by the splits at the 12 state contests at the two most recent elections.
So after the 1993 election, there had been eleven even splits and one left 4-2 split. This produced a Senate where Labor, the Democrats and the WA Greens held a 39-37 majority over the right (the Coalition and Brian Harradine).
There were no more 4-2 splits in 1996, so the 1996-1999 Senate included half its members from Labor, the Greens and Democrats, or at least that was the case prior to some defections.
The 4-2 split to the left in NSW in 1998 meant that the 1999-2002 Senate had a 39-seat Labor-Greens-Democrats majority. With no 4-2 splits in 2001, the ALP, Greens and Democrats would have maintained a majority until 2005 if it weren't for defections.
It is easy to see why the right managed a clear majority in 2004. The left had lost their 1998 advantage, while the right had won two 4-2 splits.
It is also clear why the Rudd government didn't have a progressive Senate in 2008. The 2004 and 2007 elections had produced two right 4-2 splits, one left 4-2 splits, and a 3-2-1 split to the left in South Australia.
The removal of those 2004 results in 2010 meant that there was a clear progressive majority in the Senate, despite Labor doing less well at that election. Those 4-2 splits in Tasmania were crucial to the Greens holding a clean balance of power.
The 2013 election saw a breakdown of this simple categorisation, with numerous other senators winning seats. But you can see that the left clearly went backwards - the left only managed three seats in two states, compared to winning a 4-2 split in 2007 and even splits in every other state, so that's a loss of five seats.
The old system has mostly been restored in 2022, with the exception of Jacqui Lambie in Tasmania. If you classified her as 'right', then that would make it look neater, but I think that's an oversimplification.
If you were to just look at the state results, you would think the left had won half the seats in 2022. The right 4-2 split in Queensland in 2019 was cancelled out by a left 4-2 split in Western Australia in 2022. The left got a clear majority thanks to David Pocock doing something never achieved before: a 2-0 split in a territory. Indeed I've left the territories off this chart because, prior to Pocock's win, every other territory result split 1-1 between the major parties.
Now this is a simplification in a number of ways. Defining the Democrats as the "left" may be disputed by some. They were far more willing to negotiate with the Howard government than the Greens have been with the more recent conservative governments. But this system is still useful for understanding the way that results translated into different balance of power scenarios.
The reason I'm writing about this today is the story earlier this week about expanding the size of parliament.
A modest increase to 14 senators per state would totally change this dynamic. You would normally expect the left and right to win three seats each, with the seventh seat going to the side that is stronger, or to a candidate outside of these two sides.
A similar situation led to the expansion of the Tasmanian House of Assembly in 1959. Each electorate had been electing six members per election. The two previous elections had produced perfect deadlocks, with every electorate dividing 3-3. The new seven-member electorates produced a small but clear majority for a Liberal-independent alliance in 1959.
This change would not eliminate Senate deadlocks, which have two causes. Those causes are the six-seat contests, and the six-year terms. Seven-seat contests would be much more responsive to voting patterns, but senators elected at two (potentially quite different) elections would still be combined to produce the Senate.
We saw this in 2008, when a strongly progressive 2007 Senate cohort was combined with a conservative 2004 cohort to produce a deadlocked Senate. It's similar to what we're seeing in the new New South Wales Legislative Council. The NSW upper house is quite proportional and responsive to the election results, but this is undermined by the inclusion of the members elected four years earlier.
While we can't deal with the six-year terms without a referendum (and we already tried to do this in 1988), a change to seven-seat elections would be an improvement.
the problem with 7 seats it gives an unfair advantage to one side. the best solution is to expand to 8. that would inrease the senate to 100 and would need to inrease the house to around 200
I don’t see any unfair advantage? If one side wins more votes than the other, they should win more seats. What’s unfair about that?
Agree Ben, with the ever-increasing minor party vote the likelihood of 3-3-1 splits (3 each for left and right sides, with 1 centrist/non-aligned party elected) would be substantially greater and would be no different from the current 3-3 split arrangement.
Would be interested in a guestimate of what a 7 seat result would have looked like in the last few elections
My guess in the last election might be a few more ALP seats and maybe a few ON wins
Greens would comfortable in all states
because it would be seen as gerrymandering meaning they increased it solely to increase their own numbers.
its the same as lowering the voting age to 16. it would be done only to increase their own vote.
Whose own numbers would it increase? It’s not anything like gerrymandering, since all sides would have the potential to pick up the seventh seat in different states and at different elections.
Well lowering the voting age likely would benefit the left. Not to say it isn’t right, but yes there would be a benefit.
Who are you suggesting would benefit from 7-senator elections?
Hi Ben, you’re missing 2016 for “Seat splits in each state at each 6-seat Senate election”.
Actually, my mistake, what I meant was including the 2016 seat splits even though that was a double dissolution.
Yeah 2016 wasn’t a 6-seat election. Leaving it out is not a mistake.
In discussions to change numbers of members to break deadlocks in the Senate (or even the Tasmanian Legislative Assembly) the use of an odd number of Senators at a half election (or odd number of members in a Tasmanian electorate) it is overlooked that the intention was to make it hard to make laws. Hence the Senate President does not have a casting vote (only a deliberative vote). In the US, the VP has the casting vote in the event of a tie (but this is rarely used). And the creation of classes of Senators so that they are not going to be caught up in a “populist cause”.
From the US Senate Webpage quoting the Federalist Papers:
Writing in the Federalist Papers, James Madison explained the unique nature of the Senate and the cautious, deliberative role it would play in American government. In Number 62, James Madison eloquently stated…
Another advantage accruing from this ingredient in the constitution of the senate is, the additional impediment it must prove against improper acts of legislation. No law or resolution can now be passed without the concurrence, first, of a majority of the people, and then, of a majority of the states. It must be acknowledged that this complicated check on legislation may, in some instances, be injurious as well as beneficial; and that the peculiar defence which it involves in favour of the smaller states, would be more rational, if any interests common to them, and distinct from those of the other states, would otherwise be exposed to peculiar danger. But as the larger states will always be able, by their power over the supplies, to defeat unreasonable exertions of this prerogative of the lesser states; and as the facility and excess of law-making seem to be the diseases to which our governments are most liable, it is not impossible that this part of the constitution may be more convenient in practice, than it appears to many in contemplation. . . .
. . . The necessity of a senate is not less indicated by the propensity of all single and numerous assemblies, to yield to the impulse of sudden and violent passions, and to be seduced by factious leaders into intemperate and pernicious resolutions. . . . All that need be remarked is, that a body which is to correct this infirmity, ought itself to be free from it, and consequently ought to be less numerous. It ought moreover to possess great firmness, and consequently ought to hold its authority by a tenure of considerable duration. . . .
Back to Pollster – If parliament can’t pass laws it can’t diminish the rights of the population it purports to protect…..
Best Pollster
Thanks Ben. It would be interesting to retrofit results from recent half Senate elections to see (approximately) what they’d look like if 7 seats were up in each State at a half Senate elections. I’m in favour of the size of the Federal Parliament being increased. However with the reduction in the Senate quota, in order to minimise election of fringe micro parties, I’d like to see a minimum primary vote threshold requirement- ie parties or ungrouped individuals receiving less than say, 5% of primary votes, should have all their preferences distributed and could not receive any preferences from other parties to reach a quota.
Ben, perhaps this “modest increase” would also be quite politically expedient in the current context? (I’m referring to both the Labor Government’s legislative appetite and the current media context).
So for me, this 14 Senator per state model looks excellent, and possibly also the most realistic at the moment. That said, I still think a Citizen’s Assembly is needed to investigate this and issue their recommendations.
Also, it seems a bit unfortunate to force these results into a binary left-right split (for the most part). While I understand the preference for simplicity, this devalues the role of the crossbench, especially for those Senators who are more collaborative with whoever sits in government. With the ongoing trend away from the major parties and towards various alternatives, it’s becoming increasingly important to present politics as a three-sided contest at the minimum.
Ben
In 1959, Labor formed the government in Tasmania. The result was 17 Labor, 16 Liberal and 2 independents.
max –
There is absolutely no need for an arbitrary threshold. The principle of proportional representation says another X votes (possibly via preferences) gets you another representative. That includes the first one.
(I’m open to iterative systems like the Meek method which can better deal with exhaustion.)
In any event even with 18 Senators per state in a double dissolution the quota’s still less than in NSW, and it doesn’t have an issue.
With Group Voting Tickets a fading memory, nobody’s getting elected without genuine voter support.
Increasing the senate count by one per state could agitate the LNP. This will make them lose influence in the Senate when compared to left-leaning parties.
Having 7 senate spots per state would most likely produce a 3 ALP, 3 LNP, 1 Green result in some or most states (not Queensland nor Tasmania). It would be easier for the ALP to get a third seat. It could also make a minor party like UAP, ONP, Lambie or Legalise Cannabis pick up the last Senate spot.
Regarding increasing the senate count in the territories, it will increase the Greens’ prospects in NT and ACT (when David Pocock isn’t running). Ironically, the ALP, LNP and Greens will just see themselves as shoo-ins and not compete for the senate seats like they did in the ACT in 2022.
If you use Andrew Conway’s site (https://vote.andrewconway.org/) then you can do recounts for varying numbers of vacancies. I’ve done so for 2016 onwards, assuming 14 Senators per State and 3 per Territory.
In 2016, if a DD goes ahead you get 45 right-wing Senators, 40 left-wing and 5 in the middle (NXT, Lambie, Hinch). Minor party pickups are FFP in Vic and Qld, LDP in Qld and PHON in SA.
Interestingly, looking at 2016 as a half-Senate election on its own shows 21 left, 18 right and 3 centre from the states (plus 4 left, 2 right from the territories).
NSW goes 3-3-1, reversing the DD lean, as O’Neill and Rhiannon stay ahead of Leyonhjelm and Burston.
WA goes 3-3-1 too (also reversing the DD lean) with Dodson doing very well on preferences to overtake PHON for the final seat (winning on ~0.73 quotas).
Meanwhile in Qld the left loses narrowly instead of heavily, and in Tasmania wins heavily instead of narrowly.
In SA it’s 3 Lib, 2 ALP, 2 NXT, so slightly worse for the right.
In 2019, the right wins NSW, Qld, WA and SA 4-3, while in Vic the left wins 4-3 and in Tas it’s 3-3-1. Territories 2-1 to the left. Total 2019 is 23 left, 24 right and 1 centre, while overall it’s 44 left, 42 right and 4 centre.
In 2022 the left wins NSW, Vic, WA and SA 4-3, while in Qld the right wins 4-3 and in Tas it’s 3-3-1. Territories 2-1 to the left. Total 2022 is 26 left, 21 right and 1 centre, while overall it’s 45 left, 43 right and 2 centre.
Now about those Territory results: it’s not always 1-1-1. It was 2 for Labor in 2016 in the ACT, and in 2019 in the NT. But you’d imagine with a more winnable seat there the Greens would campaign harder for it.
I was meant to say in my above post is that I’m basing the produced outcomes on 2022 senate results.
In the territories, the Senate outcome would be: 1-1-1 (ALP-LNP-GRN) or 2-1-0 if ALP really excelled and David Pocock weren’t running. This would encourage the Greens to campaign harder as the above post suggested as they have a shot at winning at the expense of a second Labor seat. However, in ACT based on 2022 results, it would be ALP 1 LIB 1 Pocock 1, as the left vote would be split between Labor, Greens and Pocock.
Comments are closed.