Enoggera – Brisbane 2024

Quintessa Denniz, a lawyer, has been selected as the Greens candidate.

https://www.couriermail.com.au/news/queensland/qld-politics/brisbane-city-council-election-2024-greens-to-target-key-lnp-labor-wards/news-story/b23f421614d295647c939be050500896

It’s starting to sound like I was incorrect about Holland Park being the surprisingly competitive contest, the jungle drums are beating for Enoggera instead. The Greens seem to have campaigned very hard in this ward, bringing in big guns like Stephen Bates and Mehreen Faruqi to assist. And according to the Australian, the Greens internal polling suggests this is now a close contest between the Greens and the LNP.

Whether it will fall is uncertain. The Greens finished just short in Paddington last time despite a strong campaign. But they did beat everyone’s expectations in the 2022 federal election, and most of the ward lies within federal seats that fell to the Greens then. One to watch closely on election night.

Agree Wilson, Andrew wines in enoggera has a much smaller margin than krista Adams in Holland Park. His status as a kind of backbencher without a significant portfolio also counts against him.

I think the result will probably like the federal seat of Brisbane, a three way tie between Labor, greens and lnp. Steven bates won easily under cpv for the federal seat but I think the result will be much closer under opv, with the small possibility of the lnp still managing to squeeze out a narrow win.

I am surprised by the Greens’s optimism here. For the Greens to win this seat in optional preferential voting they’re going to need to see a big decay in the LNP and ALP swings, and they’re going to have to be of similar magnitude. If the LNP drops hard, like -7% swing in first preference but the ALP drops by -3%, even if Greens make up the 10% difference there (it’s not that simple, but let’s assume they do) they will still lose.

The LNP and ALP would have to their first preferences drop by at least -6%, and the Greens would need to scoop that all up (i.e. capture that 12%). If they manage to do that, which would be a sensational result, they MIGHT win a preferential tightrope race against the ALP.

The dynamics here are quite complicated. I don’t really think any side should be expressing optimism apart from the ALP, it is much easier for them to win this seat even if they drop on the primary vote.

Ah I just found the Australian article you mentioned @Wilson. One thing to note about the method the Greens use to determine their swings which was developed by Max Chandler-Mather is not designed for optional preferential voting. Their method is designed for compulsory preferential voting, and it seems to work well for that. Max Chandler-Mather is many things, but he is not a statistician, and you do need to adjust an electoral model with respect to the rules of the election.

They might be messaging with that in mind, and being conservative about where they are polling well, but I think it is more likely that they are over-confident in their ability to poll with this method since the 2022 “Greenslide” in Brisbane.

Perhaps. Certainly the 40+% reported primaries for Paddington and Walter Taylor should be enough. I don’t expect them to win all of the other three, but they only said they were close there rather than on track to win. Even getting close in Enoggera would be a big achievement in my opinion, coming from a previous primary vote of just 16%.

That is incorrect Cyrus. It has nothing to do with the voting system. Its a methodology that estimates the party’s primary vote using a formula based on the support levels of people that are doorknocked. I ain’t gonna spill the formula or the type of data used beyond that (ie how that data is defined and categorised), because its reasonably novel and I don’t want lab/lib getting a hold of it, but its proven ridiculously accurate, much moreso than any seat polling that major parties do, because disengaged people that are impossible to poll are directly included in the data and you can get extremely large samples due to their mass doorknocking.

One reason why they wouldnt be trumpeting their chances of winning central, coorparoo etc as loudly as paddo and walto is because their primary vote may be in the ball park for a win, but not high enough for an outright win regardless of preferences. IE, they dont have data on labor/liberal primaries and the result of the election would depend on those because the greens primary vote polling isnt high enough alone. With Paddo and Walto above 41%, those wards are won regardless of lab/lib primaries. With a Greens primary in the mid 30s though, that leaves some uncertainty about whether they’d win or not.

Yes, Waltor Taylor and Paddington should be solid wins for them. Any miss on either of those wards would be a big gut punch.

@Cyrus, I think it’s reasonable to assume that that the Greens internal polling accounts for the impact of optional preferential voting. Many Greens would probably think they have been “done over” by the OPV system, especially given they couldn’t win Paddington in 2020 under the OPV system and would’ve won under CPV – so the OPV system “biasing” the LNP is probably a front-of-mind concern.

Additionally, Jonathan Sriranganathan did an interview with the “No Fibs” publication this month where he stated that “the one thing that’s really working against us is optional preferential voting because we don’t just need to win a big swing; we need a huge swing in order to have enough of a buffer for when people forget to number every box.”

@GPPS I would not be so sure about that. If it’s MCM’s method they are using (which is what they say), it does not account for OPV. Ask your local candidates team if you’re in touch with them. It’s getting to the tail end so they may elaborate on it, if are in the know.

And yeah, OPV entrenches the LNP’s dominant position in the BCC with a smaller support base than they would need to achieve the same result under CPV. As the web master here has very thoroughly summarised in a recent article.

The idea that no one in the Greens has noticed that OPV and CPV are different and require their models to be adjusted seems ridiculous to me. It isn’t hard to estimate what sort of primary vote the Greens need to be fairly sure of a win, or to be within reach of a win but uncertain due to being more dependent on preference flows, as ‘Cyrus is wrong’ has just outlined.

I love that I have sparked such a lively debate! I have an OPV model that I am using for my comments here, so I will use it to outline why I am so skeptical that the Greens have an OPV model.

Let’s assume this scenario, -4% for the ALP, -4% for the LNP and +12% for the Greens (equally, thereabouts, the Greens swing in Waltor Taylor in 2020, their biggest swing that year). Just ignore that this is a bit hard to imagine… it’s hard to have the 1st preferences fully add up to 100% because of informals, but let’s just imagine it does, the 1st preferences changes like this:
ALP 34.4% –> 30.4% (-4%), LNP 45.1% –> 41.1% (-4%), GRN 16.1% –> 28.1% (+12%)

Looks pretty good right? If you model that scenario out over 100 simulations, the ALP win 96, and the LNP win 4. The median final share after distribution is 49.2% for LNP and 50.8% for the ALP. The Greens do not make the runoff even with this massive swing and uniform drop in both of the majors.

If the LNP drop harder, it doesn’t matter, the ALP are more likely to win in that situation. If the LNP is more resistant, same issue, except the LNP are likelier to walk away with the ward. If the ALP drop like a rock – also bad for the Greens! They need both to drop basically equally. The Greens both need to run a massive swing here, and run the ALP and LNP into the ground.

To express confidence here, the Greens must be sure that both major parties are completely in the toilet and they are going to secure a historic swing here. I am very doubtful that even with their doorknocking campaign they can know both of those facts.

Just to be clear, I don’t think I’m an expert. I’m an amateur at election modelling (although I do have adjacent experience), but if anyone has an OPV model that differs significantly on Enoggera, I’d love to see and understand why.

My methods aren’t perfect, but they also definitely aren’t novel, nothing I’m doing above should be that earth shattering for an election modelling expert. I would expect other OPV models to have similar outcomes.

Cyrus, why have you ignored my earlier message? There is no preference model in the Greens polling. Its about their primary vote only.

Their polling doesnt even include the votes for lib/lab.

@CIW that is precisely my point haha. I’m not arguing against that. The fact is that in Enoggera they will need to account for ALP & LNP preferences dropping uniformly to win. As you have stated, they do not poll for that, and if they don’t know that, they (imo) don’t have a reasonable way of knowing if Enoggera is going to end up in the Greens’ column.

@CIW couldn’t you find a more creative username than “Cyrus is wrong”?

Anyway, Labor or the Greens can win with 25% primaries even if the LNP has 40%. Look at the seat of Brisbane in 2022: the LNP had 37.71% of the primary vote, Labor had 27.25% and the Greens at 27.24%, yet the Greens had 53.73% of the TPP vote to the LNP’s 46.27% (Labor had 54.40% of the notional TPP to the LNP’s 46.60%).

Nether Portal, that would only be the case under full (compulsory) preferences. Under optional preferences (OPV), either Labor or Greens would need something like 30% minimum to win due to exhausted preferences.

@Yoh An that is correct. 25% is simply not enough in OPV.

@Yoh An @Cyrus good point, I forgot that BCC elections are OPV.

@Cyrus Yes, the Greens need a very large swing to win Enoggera, needing something like a 15-20% 3CP swing if the preference flows are similar to last election. The main reason winning it is at all plausible for them is that the area is much better for the Greens at state & federal levels and they have not put much in the way of resources here before at council elections, while Labor is not trying that hard. They have also achieved very large primary swings (10%+) at recent state/federal/council elections in Brisbane when they’ve put in the resources. It is also plausible that the Labor to Greens preference flow may increase due to changes in voter behaviour relating to last election being early in the pandemic, but I don’t have any data there, that’s just speculation.

If the Greens primary is in the low to mid 30s, they could win the ward depending on the exact breakdown of Labor/LNP primaries and preference flows. It is possible to make reasonable assumptions about likely ranges for those based on last election (and a few other data points for things like LNP preference flows since there wasn’t a Labor/Greens 2CP seat last time), and use those to model the seat.

Cyrus, who’s to say that the Greens’ swing here must be limited to 12%? Their 2022 federal primary vote share in the booths in this area were usually in the order of 25-31%, and that was before they proved they could win seats. While people do vote differently at different levels of government, it’s possible that 2022 convinced a lot of people that they genuinely have a chance in this ward, so it’s not impossible for their primary to be above 30%.

@Babaluma what would you say is a reasonable swing estimation?

My -4% ALP, -4% LNP and +12% GRN scenario from above is (imo) quite unrealistic. But I’m curious as to what yours would be.

I’ll put your suggestions into my OPV and see what it spits out 🙂

@Wilson my +12% was based on their best performance in 2020 in Waltor Taylor, where they achieved a 11.8% swing. That was the only reason I picked it. I’m not suggesting that is their ceiling, but I think picking their best recent like-for-like result is not doing them a disservice. In fact I’d say that’s pretty considerate, because I am essentially suggesting that they’ll break their record for a BCC swing in Enoggera.

Happy to try some other values in my OPV model, if you want to suggest some.

You can just calculate the approximate 2CP for all combinations of primary from Greens at 28 to 40 and LNP at 30 to 45 (these are not predictions for the likely range but just illustrate the broader trends) and plot it, and that will give you an idea of what sort of primary is necessary for the Greens to win and where the swing needs to come from.

I would also just use the 3CP from last election as a starting point rather than the primary votes, given two independents got 4.3% together, about half of which exhausted, and there are only 3 candidates this time.

Not sure if that’s an invitation, but I went out and did that as you suggested with my OPV model.

https://docs.google.com/spreadsheets/d/1Frjwd9fii-Zosv6Z_7X5tG4zmX-zfEJHwsNSWf3tuSQ/edit?usp=sharing

Basically, from my calcs, if the LNP drop hard (by -10% or thereabouts) – the Greens still need to get a 17+% swing. This seems unlikely (both the LNP cratering, and the Greens picking up a swing that hard).

If the ALP drop (by about -6%) – the Greens need to get 15+% swing. The ALP dropping wouldn’t surprise my that much as I think the majors are generally in trouble, but the 15% swing will be tough.

If both drop (by about 5-6% for the LNP and about 4-5% for the ALP) – the Greens still need about a 16+% swing. This seems like the mostly likely for me. But even then, this is exactly why I’m confused about the confidence the Greens have. This will be very, very, hard to do. Looking forward to seeing what happens.

Your model seems a bit off – the 3CP swings aren’t zero-sum (it’s much easier to just work from last election’s 3CP given the independents that aren’t running again) and the primary values don’t add up to 100.

Based on last election’s preference flows, the lowest 3CP (so in most wards, primary vote) the Greens can win with is approximately 31%, with LNP on 39% and Labor on 30%. Last election Enoggera’s 3CP was roughly LNP 47%, Labor 36%, Greens 17%. That means that it is possible for the Greens to win it with a 14% 3CP swing, if 8% is from the LNP and 6% is from Labor. That isn’t easy but it may be within reach here, and they may also be able to get a better preference flow from Labor than last election. Obviously a higher primary than 31% is better for the Greens and makes them more likely to win, but that’s the edge-most case where they win.

The LNP primary obviously has to drop to 33% for it to be an ALP/Greens contest, which seems unlikely.

On the primary values not adding up to 100, I understand your concern, but it’s actually not necessary to add up to 100 in this sort of model.

Since the electoral base is growing, you don’t need to map every past vote 2020 to a new vote in 2024. You will have a higher sample size of voters. I’ve estimated that in my model, but it’s going to be off by a little bit. Furthermore, informals are increasing overtime, as is voter apathy. Less voters turn out for elections these days than any other time since voting became compulsory. What you’re talking about is a likely voter model. These are prevalent in American and UK politics, where voting is not compulsory.

I am ignoring the likely voter aspect of this entirely. I am assuming that the voter base will grow and therefore the distribution of swings (which I am assuming will be zero-sum, as I am ignoring informal votes) can be a bit over 100%.

These design choices, favour the Greens, it doesn’t disadvantage them as you suggest. I’m basically saying that there are more voters in the system, that will vote for the first time and disproportionately for the Greens.

Yes, the Greens can win with 31% as you suggest, my model agrees, with the parameters you suggest and ones you miss. Lines 23, 30 and 41 and 50 show that.

If they can achieve this, I don’t really think anything is off the table as it would have to be done with a pretty sizable reversion from either just the LNP (in the single case, a giant reversion) or both major parties.

If they can achieve Enoggera, it would be pretty reasonable to assume Sriranganathan will be within a point of, if not winning, the mayoralty.

Well, that’s quite a strange way to model an election. Swings aren’t about mapping every vote in the previous election to a new one, they are just about the change in the result in terms of the vote shares, regardless of the reason. Your methodology seems to just increase the complexity and decrease the legibility for no real benefit. You can of course model what a swing is likely to be based upon things like change in voter population, but it doesn’t make any real sense to present the swings and primary votes in this way which are not what anyone else means when talking about ‘swing’.

I am definitely not talking about a likely voter model (which is irrelevant to Australian politics), I am just talking about the 3CP swings and 3CP vote share (which in this election is just the primaries, but not last election) that the Greens need to win assuming last election’s preference flows.

I also don’t see any reason to expect that winning Enoggera would mean the Greens would also be in reach of winning the mayoralty. The Greens typically perform much stronger in their target seats (that they have poured resources into) than elsewhere, and are extremely unlikely to replicate an already difficult 14+% 3CP swing across the entire council.

Perhaps I am explaining myself poorly, or being a bit misrepresented.

> Swings aren’t about mapping every vote in the previous election to a new one, they are just about the change in the result in terms of the vote shares, regardless of the reason

Agreed, which my model is doing. When I said “you don’t need to map every past vote 2020 to a new vote in 2024”, I meant, you don’t need to map every past vote in 2020 to a new vote in 2024. You’ve suggested that I am doing something which I did explicitly state I wasn’t doing.

When you are concerned about say, the swings summing to 103%, what I am trying to say is that the extra 3% is just a byproduct of the growth in the vote in the electorate. Parties don’t have to win every vote from a previous voter, they can win new votes from new voters. And since my swings are based on the 2020 sample (i.e. a swing is a percentage of the 2020 vote, so it is related to the 2020 sample inherently), you can be a bit rough around the edges since the vote will grow in 2024.

Also, my model agrees with your scenarios you outlined above, and which I identified. If it is such a strange way, would that not also imply your reasoning is strange in the same way?

Upon re-reading your question, it doesn’t look like you are describing a likely voter model. But I would caution that it actually isn’t that irrelevant to Australian politics. Our voter participation has been dropping pretty consistently. It is, in fact, growing in relevancy.

As to your last point, large swings rarely happen in localised areas. Large swings happen across electorates. Like, you know, Ryan, Brisbane and Griffith in 2022. If the Greens win Enoggera.. I think it’s actually quite harsh to suggest that Sriranganathan won’t be within striking distance of the mayoralty. Look at the shares from Enoggera in 2020 to the mayoralty in 2020, you don’t need a model to figure out they look pretty similar! Perhaps Sriranganathan has other issues that draw your aversion, but I really don’t think it’s fair to say he’d underperform Enoggera in a sizeable way, he is the top of the ticket for the Greens after all.

Also, I’ve put my model where my mouth is. You profess to having much more expertise @Babaluma, and I believe you. As I said above, I’m an amateur at this.

Please build one, I’d love to compare!

‘They just need a 15% swing forehead’

um okay. seems like a tall order but good luck

I expect there will be reasonably large swings in the five wards which the Greens have put serious resources into and now think they have a chance of winning, and much less substantial swings in the other wards that they have not put those sort of resources into. That’s why winning Enoggera would not be indicative of anything like winning the mayoralty – there’s no reason to believe they could get that sort of result across the entire council.

I don’t know if they will win Enoggera but if they think they’ve gotten their primary into the 30s then they’d be within reach of it. That does require a large swing but it’s not as impossible as it seems given the Greens history of securing 10+% swings when they run a strong campaign and they already have proven strength at the state and federal levels in the area. The federal Greens primary vote for just Enoggera ward is in the high 20s, so if they were able to win over most of their federal voters and then build upon that further, that’s how such a swing could be achievable.

The most likely result in Enoggera is probably just that the Greens get a very substantial swing but still fail to get ahead of Labor, leading to Labor winning it despite not having put in much effort at all.

Since you asked, here is a very basic plot mapping 3CP to winners: https://i.imgur.com/W0NYmX4.png. The Labor 3CP is implicit. I am assuming the same average last-election preference flows from Ben’s recent article for preferences between Labor and Greens. Since there are no Labor/Greens BCC contests to refer to, I am assuming that if one occurs then 2/3rds of LNP votes will exhaust and that 1/6 each will go to Labor and Greens – this is a rough approximation of the preference flows observed in NSW state Labor/Greens contests since NSW also uses OPV. I’m not any sort of expert though, just a hobbyist.

When people talk about a say, 5% swing, they do not mean in absolute terms, they’re not referring to a comparison to the raw vote count from last election. The comparison between elections is simply in relation to the overall vote share. This is why having swings that produce primary votes that do not sum to 100 is strange methodology – you are just obscuring what the swings actually are because you eventually have to normalise everything so it does sum to 100. Since you have explained what you’re doing I’m no longer confused about your results, merely your methodology.

On your last point, yep I totally get it now. That is exactly what I did for my swings, and I now understand the issue. Thanks for explaining it. I can correct my model to allocate the correct first preference share w.r.t the growing voter base.

That being said, it wouldn’t change anything in the underlying results. My model allocates every single vote in every ward, 100 times and takes the average.

So the vote numbers will be the same, as will the outcome. The only thing that’d change is those FP shares.

In every single booth*.

I’ve updated my model with feedback from @Babaluma, thanks for your help.

It now shows 3CP correctly, but it does model election outcomes with some randomness applied to each booth return, so it has percentages for some results.

Available at this link: https://docs.google.com/spreadsheets/d/1Frjwd9fii-Zosv6Z_7X5tG4zmX-zfEJHwsNSWf3tuSQ/edit#gid=1362580988

Looks a little different to your Barumba, I find that the Greens can squeak out a narrow win at 29% 3CP when the LNP are at 41% 3CP. But this is the “chaos zone” where the ALP can also win.

The more I look at this the

Since you have some sort of probabilistic factor in your model while I assumed just a static preference flow, it makes sense that yours would find that cases a bit beyond the edges of what’s winnable for the Greens in mine can be winnable for them.

@Babluma “Since there are no Labor/Greens BCC contests to refer to, I am assuming that if one occurs then 2/3rds of LNP votes will exhaust and that 1/6 each will go to Labor and Greens – this is a rough approximation of the preference flows observed in NSW state Labor/Greens contests since NSW also uses OPV.”

Except … because of the number of seats the LNP hold, finishing anything other than third will mean that LNP preferences won’t be distributed EXCEPT in seats where it’s an ALP/Green battle. The only one of those I can see happening is Deagon. I’m willing to bet that the LNP unpreferenced vote will be closer to 90 percent but in the vast majority of cases that won’t matter.

The big decider will be what proportion of votes exhaust from the ALP and Greens and whether the Greens can be a viable second party in enough seats to overtake Labor. There’s an awful lot of LNP “Just Vote 1” signs around and if they look familiar it because Peter Beattie used almost the same design in the state election when he introduced OPV.

The ALP is not getting The Gabba ward back without a serious redistribution. In the long run I think the ALP is vulnerable to The Greens in Moorooka and Morningside.

Yes, it’s not particularly likely that there will be Labor/Greens 2CP contests, but modelling them is necessary for my plot. I think the mostly likely ward to be a Labor/Greens 2CP contest is The Gabba (though Labor won’t win it regardless) and there very are outside chances of that happening in Morningside (though Labor are the strong favourites there regardless) and Enoggera (which is the only true three-way contest, though the Libs finishing third is definitely the least likely). I truly have no idea why you think the Greens will do well in Deagon though – I don’t expect them to come close to second there.

I think the Greens will probably make it to the 2CP in a few more wards than they did in 2020 though – Enogerra is an obvious possibility, Holland Park seems possible, Hamilton is likely (though they won’t come anywhere near winning), Morningside and the Gap are outside possibilities.

The NSW Liberals also advocate for just voting 1 and about 1/3 of NSW Lib voters still preference in Labor/Greens contests, but it could be lower here, we don’t have any real data yeah.

So… why were the Greens so confident about this one?

Anyone on the inside wanting to give the tea? I’ve already said my opinion above – I don’t think they actually have an election model and/or a good understanding of the 3CP relationship here.

The Greens were pretty damn good at converting Labor votes to Green votes but had issues converting LNP voters to Greens as Enogerra demonstrates.

Clearly the nature of the swing yesterday council-wise was very different to the swing in 2022 federal and 2020 BCC where I think you can say with a high degree of certainty that there was a fairly substantial Liberal to Green shift – at least much more than what occurred last night. Probably many of those previous Lib–>GRN switchers stayed with the Greens, but the Greens weren’t able to convince many more came on board.

Maybe those factors driving the 2020/2022 swing were less present, or maybe there is limit to how many LNP–>GRN voters those 2020/2022 factors can drive to the Greens, but I think what we can say for certain is that the Greens had much more trouble under the current council environment converting LNP voters to Green than they did in 2020 and 2022.

They got a big swing, just not the sort of swing against the LNP that they needed. It isn’t hard to understand why they would have thought this one was in reach though, but just publicly expressing optimism about it isn’t the same as thinking they had it in the bag.

Apart from Walter-Taylor (which had a gigantic LNP margin that Labor had never even tried to touch at any level of government) there weren’t any big LNP to Greens swings in 2020 BCC, they were all more gradual chipping away at both LNP and Labor. Labor also ran up the margins in most of their own wards and got some sizeable swings against the LNP in a few places, which balanced out the Greens getting substantial swings against them in some places then.

@Cyrus the results at time of writing are similar to much stronger and more likely Green target wards. Polling likely correctly showed Greens were a decent amount ahead of Labor, and the seat was close over-all, close enough to publicly call it winnable.

I guess my point is that this seat in particular was reliant on a drop in the LNP primary vote (as was every other Greens target seat, but this one in particular).

I think it betrays a lack of understanding of the dynamics of the seat, for the Greens to express confidence in it. They’ve lost it harder than Labor did in 2020, so it’s basically a pyrrhic swing.

Obviously running and losing has its advantages, and puts you in a better position for 2028. But I think it’s fair to say that the Max Chandler Method of doorknocking to determine votes and swings showed its limitations on Saturday.

Cyrus, it’s only a pyrrhic swing if you’re approaching it from an anti-LNP perspective (or a pro-Labor perspective). Someone approaching it from a pro-Greens perspective would probably think that making themselves the alternative party if voters are unhappy with the LNP next time around is a big positive, even if it didn’t quite come off this time around. Even Max Chandler-Mather had to have an election where he got closer to victory, before he actually won Griffith.

Of course. That is what I’m saying without saying when I mention “Of course running and losing has its advantages”.

Maybe pyrrhic swing is a bit cynical. It does come from an anti-LNP framing, but I think that’s the only way the Greens take these seats, so I don’t think it’s that unreasonable.

This ward wasn’t particularly reliant on a drop in the LNP primary – if anything, this was the ward they were targeting that was most reliant on a drop in the Labor primary as well as being able to take a few points off the LNP, given how far they were behind Labor. The failure to secure a swing against the LNP was a broader failure so I hardly think they were misunderstanding the dynamic of this ward in particular.

The messaging from the Greens wasn’t that they were confident of a win either, just that it was a potential win, within reach, etc.

@Babaluma They achieved a cataclysmic drop in the Labor PV, and lost it by a wide margin.

So I’m not sure why you suggest that the ward wasn’t particularly reliant on a LNP PV drop. I guess I understand the particularly, because they needed the same drop in Paddington, Waltor Taylor, etc. but they did need a bigger drop here (at least -3 to -4%). They didn’t come close to that