The 2023 conference of the Australian Political Science Association is taking place in Sydney this week, and I will be presenting two papers – one today and one tomorrow.
I won’t be printing copies of my papers and so far they haven’t really been circulated so I thought I would share both papers here, both for blog readers who might find them interested, and also for conference attendees who want to download the paper and have a read.
The first paper is on how party systems manifest in NSW local council elections, and the second is on how the declining major party vote has changed how the electoral system for the House of Representatives is expressed.
Here are the links to the papers:
Hi Ben,
Thanks for sharing these and the work you do. Regarding the paper on preferences and the change away from the standard 2CP…
The AEC says:
“The AEC selects two candidates based on a number of factors including historical voting patterns in previous elections. The names of the two candidates are kept confidential until the close of the voting. In instances where it is evident that one or both of the leading candidates differ from the prediction, the TCP count is restarted and preferences are distributed to the correct two leading candidates in the days following election night. This has historically been referred to as a seat going ‘maverick’.”
This is only going to get “worse”, ie. become more frequent. Fussing around the edges and trying to make “better” guesstimates at the likely top 2 can still be somewhat of a coin toss.
Instead of tinkering around the edges, why not a more rigorous and robust model.
Given the 2CP count is only indicative, why not just NOT conduct it on election night. There will still be enough of a sentiment from the first preference counts and there’s no reason that there HAS to be outcome on the night.
Conduct an “indicatively elected” (2CP) count following the completion of the HoR Check Count. Then you KNOW who the (likely) top two candidates are. And no, you don’t need to wait for the last of the postal votes to arrive. If a contest is that close that you need to do a full preference distribution to declare the winner, then you were still going to have to do that anyway.
Thanks!
AEC also did an indicative 3CP count in the Melbourne Ports postcount in 2016 but it was late in the process and relatively un-publicised. Results weren’t published but were available to scrutineers.
I think that alternatively, and again Macnamara is probably the best possible example of this federally, if a seat is clearly going to have an uncertain 3CP order which ultimately decides the winner more than the actual 2CP count will, then in those cases they could just conduct a 3CP instead of 2CP count on the night.
That ties into G’s suggestion, in which obviously you’re not going to have a 2CP result displayed like the other seats, but that’s probably better than displaying a 2CP result and having poor Antony Green (or his replacement) have to keep qualifying that it’s not that simple.
Eg. Having the results show a clear double-digit Labor win but still having the seat as undecided because it’s the invisible 3CP count that means Labor might not even be in the 2CP count, and having to constantly explain that.
From a counting perspective on the night, it wouldn’t really be any harder. I have worked at an election before and conducted the count (funnily enough, at a booth in Melbourne Ports in 2016!) so I know that after counting the primary votes, it wouldn’t be any more difficult to then distribute them into 3 piles for whether ALP, LIB or GRN is higher than it would to distribute them into 2 piles, and then call in the 3CP results instead.
In that particular seat, not so much in 2016 where even the 2PP count was close but certainly now where the Liberals are no longer competitive, that would actually give the commentators on TV and online a much better indication of the winner, because it’s more likely that as long as you know *who* the 2CP will be between, you know who the winner will be as it will most likely be a double-digit win (unless it’s ALP v GRN).
On my experience in 2016 as well, I worked at a St Kilda booth where we could also see the potential for needing to do a recount.
When we counted the primary votes, the Greens pile was roughly as large as both ALP + LIB combined, then we had to distribute the preferences to ALP & LIB piles.
That actually took even longer than it would have to conduct a 3CP count, because rather than only then distributing the small pile of ‘Other’ votes, we actually had to distribute the largest pile AND all the others which accounted for more than 50% of the ballots (ALP + LIB only totalled about 47% in that booth).
So rather than have the 2CP take much longer at each booth due to redistributing more than 50% of the ballots, still have a completely unclear result because we don’t know the 3CP order, and have the TV screens showing a potentially misleading 2CP result, we could have actually saved time by doing a much quicker 3CP count.
And while in 2016 that still wouldn’t have made the result clearer, there would at least have been more data available to analyse regarding 3CP preference flows to apply to the remaining votes (Postal etc). And in future years if there is more of a gap between the 3CP totals, maybe it will give us a clearer result too.
Hi Ben,
Thanks for sharing your papers. I absolutely agree that the AEC should use 3CP counts wherever necessary.
For election night, these changing voting patterns aren’t just a technical question for how to best count the votes, there’s also a civic question. In particular, it’s an opportunity for the media to further clarify to voters how each 3-sided contest is turning out.
Some very rough ideas:
– Perhaps the live “in doubt” and / or “likely” categories could be adjusted to differentiate between the typical very close 2CP contests verses those where the final 2CP outcome could likely be much more convincing if only we were more confident of the 3CP order and preference flows?
– Perhaps 2CP contests should show 2 small candidates’ portraits (or just simple graphic outlines of heads and shoulders) side-by-side while 3CP contests should show all 3 (or more) portraits, so that there’s a clear visual illustration of a wider array of possible outcomes? Even a very minor detail like this could help voters to appreciate what is happening as the count updates…
As a minor critique, does your discussion get a little sidetracked on first preference proportionality? The argument for a mixed-member proportional voting system is relevant no matter whether we’re in the 1950’s or the 2050’s. But despite our trend in voting patterns over time, our current system is still successfully selecting the overall most preferred candidate in each area.
Peter, I think you’re missing my point.
This is not primarily a problem of how we report results. Yes, you could make some 3CP counts more regular to speed up understanding of the results, but ultimately that part is sorted out in the end. I do think the idea that just not counting 2CP on the night (which resolves who has won in almost all seats and thus makes it possible to know the direction of government) is “more rigorous and robust” is silly. It’s throwing out the baby with the bath water.
But this is primarily a problem of representation. It demonstrates that fragility of our single-member system – both because the results are not proportional, but also because leaving the decision of whether Labor or the Greens win an overwhelmingly progressive seat up to the Liberal Party HQ is really quite silly. And you can’t judge our system based on whether the most preferred candidate in that one area is chosen – you need to broaden out the context to the whole country.
Fair enough, Ben. I can see that. Perhaps rigorous / robust was a touch strong. There were multiple other considerations and factors to my post that I didn’t go into.
Do you have an idea for an election night counting / results model that takes into account the changing nature of the 2CP environment?
I think for now being able to switch to a 3CP count on the Sunday ASAP if a seat has a close third-place contest is fine. If it becomes a bigger issue, we can make allowance for the AEC to designate certain seats to have a 3CP count on the night, in the same way that they pick the top two for the 2CP now. The commenters are correct that in these sorts of seats it’s probably easier to do a notional 3CP count than a notional 2CP count.
I don’t think it’s a good idea to be trying to implement notional 3CP counts in most other seats.
@ Ben,
Counting and reporting changes in limited 3CP instances would certainly be relatively straightforward.
On the more substantive issue, can you please clarify if you’re interested in simply observing this trend in voting and its impacts, or do think changes are required?
I don’t think there’s any doubt about my position – we should replace our current electoral system with a proportional system – most simply it would involve districts electing 3+ members each. This issue with 3CP and non-monotonicity is not the only problem caused by the single-member system but I expect it to become more of a problem in coming years.
@ Ben,
Ok, thanks for clarifying. While I think a proportional system is a relevant idea generally, with respect, I think it’s a jump to a flawed conclusion that the electoral system might need to change because people are voting differently now.
(I understand you have other problems with the current system, so your argument for a proportional system should be separately focused on those.)
In the context of voting trends and their impacts, wouldn’t it be better to discuss any ways that this issue could be addressed within the current system?
As you point out, our current voting system has remained relatively stable. This means its “fragility” is inherent. Our overall election outcomes have always been exposed to the vagaries of close calls in a small number of seats, except when there is clear majority result.
Is it such a drama that parties’ how-to-vote recommendations can be influential? These recommendations are legitimate, so it’s also legitimate that they do influence the results in 3CP contests, especially compared to many 2CP contests where the whole preferencing debate can feel like a charade since most votes don’t go to preferences.
So under current electoral rules, it therefore seems to be the democratic right of any party HQ to make their own partisan assessment of what results would be the most favourable to achieving their political agenda in the next parliament.
So what, if anything, should be done?
Rule changes?
I don’t know if any JSCEM has discussed banning how-to-vote cards, but maybe that would help produce a more demonstrably representative response? (I’m on the fence here – happy to hear anyone’s comments for or against this option.) Or are there any other practical improvements that could be made to the current system?
Civic changes?
Education appears to need to be improved to help voters understand what impacts these voting trends have, for example:
– How preferences work,
– Following how-to-vote cards is optional,
– Campaign policies rely on winning support (and compromise, if necessary) in the next parliament to be successfully delivered, etc…
Political changes?
This is probably the most important factor, but it’s up to each party to decide what to do to win people’s support. No matter what the electoral system, once elected, MPs will always have to find a way to navigate the parliament that they are given, or go back to the voters. The current system can handle any outcome that the voters deliver, including a more European-style multipartisan assembly, if it ever comes to that.
So by all means if you want to recommend a more proportional system for its own reasons, go ahead. But I don’t think these voting trends in themselves justify any systemic changes.
The problem is that it’s a silly system that just elects a single person to represent an area, and all of the bizarre results flow on from that. And yes the results are getting less proportional as the party system has been changing.
Education is not the problem. Even with perfect education, the system potentially opens up the possibility of strategic voting dilemmas (something not really relevant when the top two candidates is clear).
Putting aside the merits of moving to a proportional system (which I think would be great), strictly talking about improving how we conduct counts for the current system that’s in place, my suggestion would be:
– For only the very rare seats where it is excepted that the 3CP order will determine the winner more than the 2CP count (Macnamara being an obvious one here), that the AEC simply have the option of putting a 3CP count in the envelope instead of picking a 2CP matchup
I don’t think that will have any impact on HTV cards, or how the party who can’t win (in this case the Liberal Party) approach contesting the seat, for two reasons:
1. The matchup will be confidential, as it already is, and only known on the night;
2. Party strategists, regardless of what gets notionally counted on the night, would already have a realistic view of what their chances are and what the likely 3CP/2CP outcomes might be. Eg. In this case, regardless of whether the AEC opts to do an ALP v LIB count or only conduct a 3CP count in Macnamara, the Liberals would already be well aware they can’t win the seat, but their placement in the 3CP could impact the outcome between Labor & Greens, and would already strategise around that.
I don’t think this solution would be used any more broadly than that, only in the handful of seats where the “likely” 2CP matchup really is completely unknown because it’s so close it could be anybody. For example, I wouldn’t apply this to the teal seats because they are far more likely to be IND vs LIB contests. So it’s not something that I suggest for all non-traditional seats.
There are a lot of benefits, as I mentioned not only would it give people a clearer picture on the night as well as more data for psephologists to use to predict/forecast results based on remaining votes to count, but it would actually even be easier & quicker for the booths to count on the night. For example a booth with 1000 votes would more likely be distributing about 80 votes into 3 piles rather than 400 votes into 2 piles.
So the results could come in quicker, and the booths can pack up and go home earlier.
@ Ben
“The problem is that it’s a silly system” is a different argument to the voting trends issue. It doesn’t address the flawed reasoning that ‘people are voting differently and therefore the whole election system needs to be changed’.
Note that the title of the article, “Not your parents’ electoral system”, is contradicted by the very next sentence, “The rules for the electoral system for the Australian House of Representatives have mostly stayed the same for a long time…” If the electoral system hasn’t changed, then how can it be blamed for the changes that voters have made?
I understand you like proportional voting, and that proportional voting is a useful tool in the democratic toolbox. But be careful to avoid taking the position of Maslow’s hammer, where every problem looks like a nail to a man with a hammer!
Hi Ben
Thanks for your work, and in this case especially for your paper on the vagaries of the current system in the Reps (and so on the resulting party of government).
Your analysis in this post adds to your and others’ arguments for changes to the electoral system to increase the extent of the Reps actually reflecting the proportionality of the national vote. For me – and I guess for all democrats – improving the proportionality of government-forming electoral outcomes should be a, or probably the, first order criterion for designing the electoral system. Especially at the national level, not just the local level where it’s a consideration for the AEC when drawing electorate boundaries.
Redesign based on the proportionality principle should definitely include, but also go beyond, the importance of the ‘who comes third’ issue which was so clearly illustrated by Macnamara last year, and the capricious outcomes in other three-way contests. It should also address the inequity in the Nationals capture of regional single member electorates, leading to their inordinately high number of seats compared to their national vote, and the corresponding lack of seats that the Greens, for example, typically win.
As a first approximation, your proposals for multi-member electorates looks like a good starting point for developing more proportional electoral models.
For readers interested in the history of related debates, I tip my hat to Dean Jaensch who picked up on the weakening of the loyalty- and interest- based 2 party Australian system in his 1983 book ‘The Australian party system’. He anticipated the rise of catch-all parties, but assumed that in Australia they would only compete for the swinging voters between ALP and Coalition, rather than for emerging new constituencies (eg the Greens) outside established center left and center right.
In USA Kircheimer introduced the catch-all term around the early 1960s, and meant something more like populist parties rather than just the soft centre that Jaensch picked up on. Kircheimer’s catch-all notion was fuzzy and remains contested, but it includes, for example, the direct personal connection between leader and electors that diminishes the importance of party ideology and organisation. His examples of this were the fascist parties after WW1. We might think of Trump, or perhaps the other side of that coin in Morrison’s visceral personal disconnection from electors in the Teal electorates.
“The electoral system” and “the rules for the electoral system” are not the same thing. The changes in how people vote is changing the way the system works. That’s the whole point of the paper!
And yes, if more people vote for parties other than the major parties and thus make the results less representative of how people vote, that does increase pressure to change the system. And in this specific case, the fact that having more close three-way contests opens up potential for strategic voting and makes the outcomes more arbitrary does mean the system is less fit for purpose.
All of these problems can be pinned on the decision to only elect one MP per seat.
Belgium, the Weimar Republic, Italy, and Israel are also good arguments *against* PR and I think that you and I would have a common interest in keeping the Greens – who want to bring back socialism, an ideology that is dead – and One Nation – who attract, well Fraser Anning – as far away from power and influence as possible.
@ Ben
“‘The electoral system’ and ‘the rules for the electoral system’ are not the same thing.”
It appears we have a semantic difference. For clarity’s sake, I’m referring to “the system” being equivalent to “the rules”, which are relatively static. How we use this system is much more dynamic.
Semantics aside…
You rightly identify several problems with the current system:
– Volatility in close 2CP contests
– Volatility in close 3CP contests at both 3rd and 2nd places
– Volatility in close results overall based on any close single seat results
… But these problems are inherent in this system, so they’re always present whether or not people are voting differently nowadays.
For this reason, your paper on voting trends would have more credibility if it discussed options for reforms in a more holistic manner, and then considered:
– Which options deliver smaller or larger improvements to the specific issue of voting trends?
– Which options are easier or harder to achieve in the context of our existing system?
Otherwise, in the absence of any alternatives, how can your suggestion that substantially changing the electoral system be adequately justified as the right or best approach? And what happens to your argument in case electoral system change isn’t achievable?
If the trends in voting are really as big a problem as you claim they are, then wouldn’t you want to avoid putting all your eggs into one basket?
My paper was not focused on options for reform. It was focused on identifying the trend and what it might change about how our electoral system works.
@ Ben,
Sure, it’s your paper. This is just feedback made with only constructive intent. Thanks for being open and responsive to these comments. I’ll leave that voting trend discussion there.
On the proportionality question, given that you favour changing the system for various reasons, can I please ask your opinion if it’s the voting method that should to change, or do you consider if proportionality could be achieved in other ways?
In principle, while I agree that more proportionality is better… in practice, I don’t think a proportionally voted house is the way. Our house is set up to be representative, and could perform much better if we simply enlarged it, making divisions much more individually distinct and collectively diverse.
A People’s Assembly formed by proportional random selection could be a superior model to a proportional house. The assembly would complement our existing system by providing something completely new – genuine informed *deliberation* between regular people on key issues – with findings reported back to the public and the parliament.
The assembly could be trialled by our existing parliament, making it easier to create, test, and refine. A proportional house could be harder to implement if it might require a committed step change from the existing system.
So how about we reform proportionality by introducing sortition?
I don’t see what sortition has to do with proportional representation.
@ Ben,
I’m suggesting that sortition and proportional representation are both proportionality reforms, and that introducing sortition is a better option.
We could create the assembly by selecting people randomly; in relative proportion for location, gender, age, cultural background, education, etc. This would make the assembly more proportional than if the house had 3+ member divisions.
– It would be more democratic, by inviting citizens to participate directly in the democratic process.
– It would be more deliberative, by including everyone in the assembly in a discussion of ideas and in making informed decisions.
– It could be quite viable, as it can be tested out within our current system to see how it works before anyone needs to commit to this change.
The findings from the assembly would reassure all the people who don’t have the inclination or the time to investigate issues themselves, and they would reassure governments by giving them a face-saving way to address wicked problems and avoid ‘wedge’ positions – they can announce, “We’ll take this issue to the People’s Assembly and implement the recommendations that they make”.
Does that help clarify the link between sortition and proportional representation?
No I don’t think sortition has the slightest thing to do with PR. One is about changing how our representative processes work so they are more representative and more inclusive, the other is about creating entirely different institutions. I don’t have all day to list all my issues with using sortition, but if you had such an institution you would need it to be overseen by bodies actually elected by the people and that gets you back to the original problem.
The point of PR is that the parliament should be proportional to how people vote – creating a body that is demographically representative doesn’t mean it is politically representative, and entirely ignores the role of parties in shaping and channeling public debate.
@ Ben,
Your concerns can be readily addressed:
“…but if you had such an institution you would need it to be overseen by bodies actually elected by the people…”
– Yep, the house and the senate.
“…and that gets you back to the original problem.
The point of PR is that the parliament should be proportional to how people vote…”
– We can lessen this issue by expanding parliament, and we arguably need to do that either way. ASAP.
“…creating a body that is demographically representative doesn’t mean it is politically representative…”
– What? Yes it does. If it’s demographically representative, then within reasonable margins, it contains the political points of view in proportion with the extent to which those views are held.
“…and entirely ignores the role of parties in shaping and channeling public debate.”
– No, the government, opposition, and crossbench would all still be formed in the house and senate. They would all still make and review policy, and the government of the day would still run the country. Any big issues would continue to be publicly debated as usual, but would also be deliberated on within the People’s Assembly, and their findings would help to inform the public debate.
Ultimately, if proportionality is worth pursuing, then why not seek to make the most proportional and beneficial reform that we can?
Still, you’re welcome to disagree. Thanks again for engaging. It’s great that you’ve shared your paper and your thoughts.
Comments are closed.