Victoria 2026: The Election Where All Sorts Of Things Can Happen!

Victoria Aggregate Labor 26.0 Coalition 27.3 Green 12.8 One Nation 23.7 others 10.3
(polls are probably underestimating "others" vote)
2PP estimate 52.0 to Coalition vs ALP, shadow-2PP ALP 50.1 vs One Nation

Currently Coalition would be likely to govern in minority with One Nation support, with One Nation seat tally around the mid-teens

One Nation and Labor seat numbers are highly dependent on Coalition preferencing decisions

Both major parties are still capable of majority government with a modest swing from current primaries

------------------------------------------------------------------------------------------------

It's been forever - largely because of the amount of work I had on in the first half of the year - since the last time I looked at Victorian state polling in the leadup to this year's election in any depth but finally a good time to do so has arrived.  What I see, at the moment, is a picture with a lot of possibilities.  Arguments can be made on current polling that the Allan Labor Government, while currently behind, is not far from surviving.  Arguments can also be made that it is not far from losing the election so badly that it may not even be the second party.  For now this is a really difficult election to model - much harder than South Australia - but perhaps by polling day things will get a lot clearer.  This one is a rather mathsy modelling article that has been rated 3/5 on the Wonk Factor scale.  

One thing I cannot stress enough is that this looks like being the first Australian election for a long time where how to vote cards really matter.  At elections in recent decades how to vote cards have been vastly overhyped, especially by media, given that only about a third of voters follow them, and those who do follow them are mostly major party voters whose preferences are rarely distributed in competitive seats.  Furthermore among Greens voters the 15% or so who follow them would probably nearly all preference Labor above the Coalition even if the card said otherwise, so it's not clear that Green how to votes have any real impact these days on the two-party picture at all.  

But on current polling it could be that one major party or the other will be eliminated in close to two thirds of seats.  And at that point whether or not the Coalition recommends preferences to One Nation ahead of Labor will have a huge impact on what would currently be dozens of ALP/One Nation contests. This week saw the release of a Trades Hall Redbridge MRP poll where off 27% of the vote and based on preferences nominated by respondents, One Nation wins only 12 seats.  I am certain that if One Nation's vote is really that high and the Coalition recommends preferences to them, they will be winning many, many more.   But I don't think One Nation's vote is that high now, let alone whether they can hold it to the election.

Recent Polling

The Trades Hall Redbridge MRP (or is it Redbridge-Accent?) with a sample size of around 6500 raised some mysteries since it is unclear to me if it is fully independent of the AFR poll conducted June 17-28 with a sample size of 5516.  The primary vote figures of both have One Nation on 27 and both majors on 26.  The Greens primary in the Trades Hall poll was reported by The Australian as 12, however that was alongside Others on 8, and Redbridge normally forces the sum to 100.  The AFR has reported the Greens primary as 13 in an article on the Trades Hall poll, which is also what was in the AFR poll.  No dates for the Trades Hall poll have been reported anywhere (many people decided it must be fresh polling when it was reported on July 14, but with commissioned polling this is frequently not the case).  Unless advised otherwise I'll be assuming these two samples overlap.  In any case, I'm not considering the Trades Hall poll in estimating voting intention, as it is a union poll and there is plenty of public polling to go on.

I see again no need to resummarise poll data summarised elsewhere but there have been five recent public polls - the monster AFR Redbridge, two Resolves, one DemosAU and a Freshwater.  The DemosAU is an outlier with the Coalition leading Labor on primaries by nine points while Freshwater has the lead at four and the other three have the two major parties tied.  My own rough aggregate of this polling is Labor 26.0 Coalition 27.3 Green 12.8 One Nation 23.7 others 10.3, which I estimate at a classical two-party preferred of 52.0 to Coalition.  Bludgertrack has Labor 26.8 Coalition 27.1 Green 12.5 ON 23.0 others 10.6 with a 2PP of only 50.1 to Coalition.  Estimating 2PPs is difficult at the moment because One Nation has affected the nature of the "others" pool compared to past elections (meaning that a true last-election flow is not a sensible concept) so it's possible Labor gets a somewhat better overall flow than I think.  

Even my 2PP may sound like that's not too bad for Labor if One Nation is all a weird dream and the election behaves normally.  On a uniform swing basis the 2022 election produced a very skewed pendulum where even that 7% swing takes out only eleven of the 56 seats Labor won in 2022, leaving them with 45/88, a majority.  But there is always a level of effectively random variation in seat swings on top of cases where one party just outcampaigns the other in the marginals, and on top of personal vote effects, and in this case it's important to look just above the waterline implied by that swing.  Labor have five seats on between 7% and 8% and a further nine between 8% and 9%, but only three between 4% and 5% and three between 6% and 7%.  So if it really is 52-48 to Coalition the Coalition is going to win the 2PP in some of the Labor seats that would not fall by uniform swing, and indeed on my usual seat model Labor only wins the 2PP in an expected 43.3 seats after considering personal vote effects, and that includes an expected 3.9 2PP wins in seats held by the Greens.  That's only just a loss on average, but it gets worse.

The Redbridge MRP has the Coalition winning a median of 39 seats, Labor 35, One Nation 12 and Greens 2.  For their 2PP (53-47 to Coalition) that number of Labor 2PP wins (37) is pretty similar to what my model expects (it has 39.7 2PP wins).  But Labor would not do even that well in a real election where the Coalition recommended preferences to One Nation.  

I have remodelled the primaries from the Redbridge MRP based on how I would expect preferences to flow on this assumption.  Before I do this, a reminder that of course I know MRPs are not accurate on all specific seats and are designed to gauge an overall picture (One Nation for example will not be getting 18% in Northcote!).  I also think that Redbridge(/Accent) are underestimating the Others vote in general, in places considerably (I mentioned some of this issue here).  But anyway I find the following differences:

* Thanks to weak respondent preference flows from the Coalition, Redbridge has Labor beating One Nation in Greenvale, Kalkallo, Mill Park, Cranbourne, Narre Warren North, Sydenham, Lara, Macedon, Frankston and Carrum.   I have all of these on Redbridge's numbers as wins to One Nation, though nearly all of them by 2% or less.

* Redbridge has Bundoora and Essendon as Liberal gains from Labor, I have them on Redbridge's numbers as Labor retains.

* Redbridge has Prahran with primaries of Labor 26 Liberal 26 Green 25 One Nation 16 (nah) and others 7 as a win to Labor.  We have seen this movie many many times and when One Nation preferences split three ways between Labor, Liberal and Green, the Greens will get more of them than Labor (this even recently happened off a highish One Nation vote in Heysen SA).  Even if they didn't the Others vote in Prahran usually has an Animal Justice component.  So Labor would be eliminated and the Greens would win the seat.

* Redbridge has Gippsland East, Lowan and Ovens Valley as Coalition wins; I have them as One Nation wins on their numbers.  (Again, I am not saying any of this actually happens, just that it is my reading of what would happen on these primaries).  

So instead of Redbridge's 39-35-12-2, under this assumption of the Coalition recommending preferences to One Nation I get 34-26-25-3.  At which point Labor are on the edge of falling into third place in the party seat total - though if One Nation were willing to even momentarily provide supply and confidence while Labor were not, such a case might require rethinking the usual "most seats" convention for who the Opposition is.  

I should note also that the Greens winning anything but Prahran on Redbridge's numbers would be contingent on the Liberals recommending preferences to the Greens ahead of Labor.  Ted Baillieu's decision not to do this in 2010 was so well received that it arguably got the Liberals just over the line for their victory that year, and the Liberals' reversal on that score in 2022 seemed like a shallow grovel for a hung parliament.  In the different political landscape around the Israel/Palestine and antisemitism issues since October 2023 a Liberal decision to again recommend to the Greens just to make Labor fight in a handful of seats would seem like a massive free kick for One Nation. So maybe Labor would get Richmond and Melbourne then (again, this is only off Redbridge's numbers, I am not saying that they are accurate!)

What if One Nation isn't so high?

My reworking of the Redbridge MRP is what I call the "hybrid vigour" scenario.  Labor isn't badly losing the two-party race against either the Coalition or One Nation but they get smashed in seat terms overall because of seats where they would beat the Coalition head to head but One Nation is in the final two and beating them instead.  (This applies on my numbers to seven of the seats where Redbridge has Labor winning - in Frankston, Carrum and Macedon my numbers have them narrowly losing the 2PP as well).  On Redbridge's respondent preferences One Nation can dislodge the Liberals in a seat that is a close 2PP battle but the flows to One Nation are so much weaker that Labor wins the seat much more easily.  But in South Australia this didn't happen.  We saw some seats where One Nation was actually competitive that the Liberals never get near winning - and that was against a very popular first-term Premier not a 12-year old government headed by a Premier with some of the worst personal ratings we have seen.

But also, Redbridge tends to get One Nation on the high side compared to other pollsters (this is a long-running thing) and my aggregate has them lower.  When I rescale Redbridge's sample to my own two-party estimates (I get Labor leading One Nation 50.1-49.9, compared to One Nation leading 51.2-48.8 if I use Redbridge's primaries) Labor now saves South Barwon and Eltham from the Liberals and Greenvale, Mill Park, Cranbourne, Narre Warren North and Lara from One Nation, while the Coalition now on my estimates by uniform swing off Redbridge's seat primaries beats One Nation in Berwick, Bendigo East, Gippsland East and Ovens Valley for a total of Coalition 36 seats Labor 33 One Nation 16 and Greens 3.  

Australian Election Forecasts (which was outstandingly accurate in South Australia) currently has a nowcast with midpoints of Coalition 37 Labor 30 One Nation 13 Greens 5 (the totals of their midpoints aren't forced to 88, so the majors would probably get one or two more) off primaries of Coalition 28.1 Labor 25.2 One Nation 22.1 Green 12.7 others 11.8 - a pretty similar outlook to my best remodelling of the Redbridge MRP above off primaries that are slightly better for the Coalition and worse for One Nation. Their forecast is that One Nation will come down a little bit more by election day.  I also want to mention a new and very thoroughly considered modelling site The Swing Is On which off primaries of Coalition 27 Labor 25.5 One Nation 25 Greens 12.8 others 9.7 has 37-33-13-5. 

Overall, it seems that if One Nation does get 27% and the Coalition recommends preferences to them, then they should get a pile of seats and not just the dozen in the Redbridge median estimate.  But what I am seeing here both in my own reworking of the Redbridge primaries and in the other modelling being produced is that when the One Nation vote drops from there to the 23-25 range those are falling off rapidly.  They're still doing much better than in South Australia but South Australia had Labor on a much bigger primary.  An objection I often see is that One Nation still have no candidates and how will they win seats anywhere where their candidates are nobodies.  But if this is a systematic issue then their vote will be somewhat lower.  Candidate effects only go so far and some little-known One Nation candidates were able to win in South Australia.

At some point I may also produce my own model of the party vote shares, using such things as past One Nation primaries (though the upper house primary in 2022 was so low with so many competing similar parties that it's not a great guide) and demographic indicators.  That would also allow for taking into account personal votes, as it seems to me that in South Australia the personal votes of some of the Liberal incumbents enabled them to hold seats that should otherwise have fallen, especially against One Nation.  One Nation could do particularly well in vacant seats since there is no "politics is stuffed but my local member is a good sort" kind of sentiment.

Can Labor win on both flanks?

I'm going to assume that if Labor and the Greens win 44 seats combined then they still govern; somebody would crack.  So from the numbers I have off the Redbridge MRP poll (my preference flows but its primaries) Labor needs to do enough to win 15 more.  And it turns out that with a uniform 2CP swing back of 2.75% they would do this.  That corresponds to a primary vote swing of about 3% to Labor if it comes fairly evenly off the Coalition and One Nation.  So, while things look bad for Labor on the primaries in current polling, it's possible they can retain government with a primary vote as low as 29% - but only if the split between the Coalition and One Nation stays.  If One Nation's primary falls and the Coalition's rises then this is no longer possible, because the Coalition's 2PP will rise in the classic Labor vs Coalition seats, and there will be more such seats as One Nation is increasingly knocked out in third.

In South Australia, One Nation's vote held up to its polling very well.  This may be different though; everybody knew who was winning that election and so protest voters were entirely free to vent.  In this case the Coalition may be able to drive the argument that this one really matters and that if you want to throw Labor out you shouldn't muck around with an unstable party.  

So let's also take this the other way, how close might the Coalition be to a majority?  I can find enough seats to get them there if I give them a 3.1% 2CP swing off my estimates off the Redbridge primaries in every seat.  It's a little trickier than that because a couple of them are Coalition-One Nation seats where a uniform Labor to Coalition swing isn't enough to beat One Nation, and above that 3.1% the winnable seats for the Coalition become scarce quickly.  But if the Coalition can get, say, 29.5% then that could be enough for them for a majority.  It's quite possible if the One Nation vote stays in the 20s that either major party can win outright if it can get its vote to 30.  Of course, this will leave precious little for the other one.  I am already seeing ads claiming that the Coalition cannot win alone and will require One Nation's support, but it is way, way, way too early to be sure that this is true.  

As it is possible at the moment for Labor to survive, it's also possible they get really, really smashed.  Redbridge had 25 seats as a lower bound on their numbers but if one imagines it is really 56-44 instead of 53-47, then Labor can be easily knocked below 20 seats.  I'm not really expecting that at this stage though; I think more likely if the Coalition win outright it will be because One Nation have not won a lot.

A final thing about the Redbridge model is that it seems to have Labor slightly underperforming in 2PP wins.  Perhaps this would happen as a result of uneven local swings.  However this was also widely expected in MRP style models for a while in the leadup to the 2025 federal election, and it ended up being not the case at all.  

I've previously mentioned the overall historic picture - 12 year old federally dragged governments tend to lose a lot of seat share, Premiers this unpopular who make it to elections have always lost, and governments that go to their third leader in the same term as would happen if Labor changed leaders have lost (usually heavily) except for a single case in the NT.  And with these historic priors I find it hard not to see the Coalition winning somehow.  But the split in the right and the soap-opera chaos of the current Opposition through this term raise questions as to whether the normal rules still apply.