Why Hunter 2025 Can't Be Used To Do Last-Election Labor vs One Nation Preference Flows

With the very rapid rise of One Nation in recent national primary voting intention polling we are starting to see some pollsters offer a national alternative Labor-vs-One-Nation two-party figure.  DemosAU did this in its national Jan 5-6 poll where it found One Nation tied with Labor 50-50 while Labor led the Coalition 52-48, this off primaries of Labor 29 Coalition 23 One Nation 23 (more on that later) Greens 12 others 13.  A newish outfit curiously polling on the same dates, Fox&Hedgehog (founded post the 2025 election by a former Peter Dutton staffer) reported 56-44 to Labor vs One Nation off fairly similar primaries of Labor 29 Coalition 25 One Nation 21 Greens 14 others 11, compared to 53-47 for Labor vs Coalition.  So DemosAU has One Nation two points more competitive than Labor on a head to head with Coalition basis while Fox&Hedgehog has them three points worse.  (I'll add that by my last election preferences 48% 2PP for Coalition is pretty generous on the published DemosAU breakdowns, I get 47.4 as the average for their primaries.)

DemosAU attempted to use last-election preferences by using the flow in the seat of Hunter 2025 (the only federal case ever of a Labor vs One Nation finish) to model Coalition to One Nation flows, by assuming the Greens to One Nation vs Labor flow would be the same as the Greens to Coalition vs Labor flow, and also by assigning flows from Others 50-50 between Labor vs One Nation "As the composition of Others is not known".  (That last bit did strike me as a little curious unless they were also doing the same thing for Labor vs Coalition, given that the 2PP flow from others in 2025 was about 54.7% to Labor).  Fox&Hedgehog simply used respondent preferences.  

I happened yesterday to see a widely viewed episode of the YouTube Sloan Zone channel from last week that absolutely blasted DemosAU for using flows from Hunter to model nationwide flows of Coalition preferences to One Nation vs Labor.  (I should note the Hunter flow used by DemosAU was 82.9% to ON, not 88% as stated at one point in the video.  Also, the poll wasn't commissioned at all and was self-initiated).  Seeing this video and a high level of interest in this poll generally in online comments since it came out reminded me that I'd been meaning to crunch some numbers and try to get to the bottom of how representative Hunter really is or isn't.  So Hunter 2025 is the only seat to have produced this matchup in federal history, does that mean all we're left with is extremely wild guesses and people seeing what they want to see?  No of course it doesn't.  We can examine Senate preference flow data!

A standard relationship in House of Reps preference flows is that all else being equal if a candidate has polled a higher than normal primary vote in a booth, they will also receive a better than average preference flow in that booth.  Not every seat displays this pattern but it is so reliable that I use it every federal election to catch and draw attention to data errors in the booth counts.  It also tends to work at seat level too.  

So I went through every seat in NSW for the 2025 federal election and recorded a statistic that would combine the Senate strength of One Nation vs Labor in a given seat, which was (ON primary)/(ON+ALP primary) as a percentage.  And as a potentially dependent variable I found the above the line flow of Coalition preferences to One Nation vs ALP, with exhaust disregarded, using David Barry's Senate Preference Explorer.  (The very low rate of Coalition below the lines in NSW meant I didn't bother with the extremely fiddly business of including them.)  

Hunter is a very strong seat for One Nation but it's also a strong seat for Labor, so based on the usual relationship it should have a highish Coalition flow to One Nation but not among the very highest.  And that is what I got.  The relative One Nation vs (Labor + One Nation) primary vote strength figure ranged from 4.0% in Grayndler to 35.0% in Parkes, with a NSW-wide figure of 14.6% and a Hunter figure of 25.7%.  The One Nation share of Coalition preferences vs Labor ranged from 56.4% in Chifley to 91.5% in Page, with a NSW-wide figure of 78.3% and a Hunter figure of 86.2%.  The latter is reasonably similar to the Hunter Reps flow from Coalition to One Nation of 82.9% so all else being equal it's likely that the statewide Reps Coalition to One Nation flow would be several points lower than that, and probably quite similar to the statewide One Nation to Coalition Reps flow of 74.2%.

Here's a graph showing how this relationship operates across different NSW seats.  Seats where the primary Reps Coalition candidate was a National are shown in green, with Hunter marked with an H.

I was a bit surprised by Richmond (the leftmost green dot) as I thought the dampening of the Labor primary by the huge Greens vote there might mean Labor did better on Coalition preferences than their primary vote implies, but it seems Richmond Coalition voters really don't like Labor!  Overall though there's not much evidence for any National Party specific effect (the rightmost blue dot is Liberal Sussan Ley's seat of Farrer) - it's more that Nationals tend to be the Coalition candidate in rural seats where One Nation is strong and Labor often weak, and the voters who vote Coalition in those seats tend to have attitudes in common with and preference One Nation.  Plenty of rural Nationals voters would in fact vote for a Liberal in their seat if there was one to vote for.

There is clearly a quite strong overall relationship between the relative Senate primary votes of the parties and their preference flows from the Coalition, and for that reason alone Hunter needs to be treated as unlikely to be typical in the Reps.  The vast majority of NSW seats sit quite close to the trendline here - I should mention the few that don't in the lower left of the graph; these are the western Sydney seats of Chifley, Greenway and Fowler - three seats where One Nation doesn't poll terribly on primaries but does attract very weak preference flows from the Coalition.  This might be down to lack of Coalition how to vote card handout effort but I think it could also be that Coalition supporters would be more ethnically diverse and more suspicious of One Nation in these seats.  Contrawise, the teal seats tend to have somewhat higher Coalition to One Nation flows than would be expected given their parlous ON primary votes (and parlous even in the NSW Senate contest, where there are no teals).  

One might also ask how the NSW Coalition to One Nation preference flow with exhaust removed compares to other states.  Here comes a surprise: NSW, not Queensland, is actually the highest!  NSW 78.3 Qld 77.3 Vic 73.9 WA 73.2 SA 72.5 NT 70.6 Tas 63.2 (and Tas probably even lower after adding BTLs).  National 75.6%.  I am cautious about drawing too much from this because each state has its own ballot paper draw for the Senate, but we should not be too surprised from that if the national Reps flow from Coalition to One Nation was down around the low rather than the mid 70s.

I should also add that I looked at the Greens Senate flow and Greens preferences actually flowed more strongly to Labor vs One Nation than they did to Labor vs Coalition in every state, by an average of around 3.5%.  (Victoria is particularly striking with Greens preferences splitting 87.54-2.47 to Labor vs One Nation with remainder to exhaust).  So it's quite possible the national Greens to Labor vs One Nation Reps flow would be over 90%.  There are some minor right parties (especially Trumpet of Patriots) which had much stronger Senate flows to One Nation vs Labor than to Coalition vs Labor, but a lot of the Reps "others" votes is actually independents who would probably display the reverse pattern, especially in teal seats.   

My best guess at a last election Reps flow is that One Nation would have got something like 72% of Coalition preferences, something like 9% of Greens and maybe if lucky the same 43% of Others that the Coalition got, and if that's right then by last-election preferences the DemosAU comes out about 54.2-45.8 to Labor and the Fox&Hedgehog poll at 55-45 (not too different to their respondent preferences estimate).  

A further note about Hunter is the Stuart Bonds factor.  Bonds is a local candidate who polled very strongly for One Nation in 2019 and 2025; they fell in a hole when he ran as an indie in 2022.  It's likely he attracts preferences from some Coalition voters who wouldn't normally preference One Nation.  On the other hand, polling based speculation that One Nation would come second in Hunter may have driven a level of strategic voting for One Nation by people who would normally vote National, and so some of the people who would have normally voted National with preferences to One Nation may have just voted straight for Bonds, whose Reps primary was over 2.5% above the party's Senate primary. My overall view is that these two factors roughly cancel out.  

But Everywhere Is Hunter Now?

I would say based on the above that there's very little doubt that the Hunter Reps flow to One Nation would be unrepresentative of national House of Reps flows.  It is probably even worse than respondent preferences (shudder).

The counter-argument that the Reps flow from Coalition to One Nation could be Hunter's 82.9% next election or even higher comes from the Reps voting intention numbers.  My figures above use a range of relative One Nation vote strengths from 4% to 35% but in the DemosAU poll they're at 44.2% of the combined Labor/One Nation vote!  If One Nation become waaaaay more popular, won't that mean their preference flow from the Coalition rises?  

I'm not really convinced that it will.  Firstly this sort of relationship works for predicting preference flows within a given election, not necessarily between elections.  Secondly if there is a massive swing to One Nation at the Coalition's expense then this probably means the Coalition is losing a lot of its right flank voters who were the most likely to preference One Nation anyway; those who remain might be more inner-city and more resistant to doing so.  Indeed there's an argument that the reason One Nation did not already surge more in the 2025 election was that Coalition supporters flirting with One Nation were held back by Peter Dutton being Coalition leader.  And thirdly, it's not clear the flow to One Nation keeps rising as the strength of the One Nation vote flies off the chart; it seems to max out on average somewhere around the high 80s, and there will always be some seats where the party still polls badly.  

But people are welcome to whatever opinion they like on how preferences will really flow to One Nation in a world where they're polling over 20%.  It's one thing to have those opinions and another to treat the Hunter flow as a credible 2025-election preference flow.  It's simply not.  

I mentioned at the top that I'd say more about the 23% for One Nation.  One thing that is notable about DemosAU so far is that for whatever reasons it tends to get somewhat lower major party primaries than other polls.  The combined 52% in this poll was the lowest ever for any released poll, beating a 57% which was also by DemosAU.  It may well be that the combined major party primary is now several points below the 2025 federal election and it may even be that we will be seeing combined major party primaries like 52% or lower more often.  But for the time being this pattern requires some caution about exactly how large One Nation support is.  We will know more on that score when the more major pollsters return from the summer break.

PS (added 16 Jan): I should note that a risk with respondent preferences for modelling Coalition to One Nation flows is how to vote cards.  This is a bigger issue than for respondent preferences in Labor vs Coalition contests.  Coalition how to vote cards have quite high follow rates (around 40-45%) and at the 2025 election the Coalition recommended preferences to One Nation above Labor almost everywhere, exceptions being Banks, Bennelong, Chisholm, Goldstein, Kooyong, Menzies, Reid (and the ACT where One Nation didn't run).  There's a potential based on that for respondent preferences to underestimate the flow from Coalition to One Nation by something like 10 points, which at current polling levels is worth about 2 points to the bottom line (slightly undercut by a reverse of the same effect with the Greens).  My suspicion is that the Coalition voters who do follow the card are however at the conservative end and would be more likely to preference One Nation anyway so the impact may be not as large as this.    

This Person And Why They Are Wrong: Episode 1, Wasted Vote Guy

 

The gloriously cooked tweet above reminded me of a series I'd been intending to start where now and then I would cover someone known in the online psephosphere who has a particular gimmick that I haven't previously addressed in detail.  The rules for inclusion in this series are:

1.  the person in question needs to be a published author on elections and not just a rando twitter pest  (though this first one is really scraping the barrel on the first bit) 

2.  they need to have some defining pet argument or recurring MO that makes covering what they do in one article worthwhile and effective.

3. they need to be someone who I've not already written multiple articles debunking, so no Dennis Shanahans will feature in this series.  

I should note here that the subject of this article has written Substack articles unsuccessfully criticising my comments about his nonsense on multiple occasions.  (This did come after I blocked him on Twitter in May 2022 for bogus triumphalism and misrepresenting my arguments - he not long after deleted his side of that exchange.) He may be small fry, but from time to time I do come across someone who has taken his eccentric claims seriously.  Often these are well-meaning people who do share genuine concerns about the under-representation of the Greens in the House of Reps and just don't realise that this particular version of those concerns is silly.

Our subject here has been known under various usernames including djrobstep, wheelreinvent and similar, but in offline discussions he often just gets called "that wasted vote guy" or words to that effect.  The extremely blunt style with which he's pursued his argument on twitter over the years (though not so much in the last six months or so) is instantly recognisable, even when somebody else is describing it.  It has emerged in recent years that wasted vote guy is Robert Lechte.  He was published in Jacobin on his pet subject and then the editors of Crikey chose to put him in a cage match (shamelessly dubbed "Friday Fight") with William Bowe.  He doesn't usually use his full name and I'm not going to either; I'm going to keep calling him wasted vote guy (WVG for short) for the rest of this episode. 

WVG is a fanatical proportional representation supporter, but one who serious adherents tend to think gives it a bad name.  Aside from his argument being an unsound misapplication of theoretical concepts (proof that taking electoral theory articles on Wikipedia seriously will certainly rot the brain), there's another core problem with his output.   While he will sometimes say that preferential voting is much better than first past the post, he often attacks preferential voting with language that seeks to scandalise.  It's as if he wants to have his cake and eat it too by on the one hand claiming that our system is really terrible and corrupt and on the other hand trying to not appear stupid enough to miss the massive daylight between it and first past the post.   

The primary claim WVG makes is that most of the votes cast in House of Reps elections and other single-seat elections are "wasted".  Where this departs from a common argument made by PR advocates about the number of votes that end up with the loser, is that he also includes votes that end up with the winner over and above the loser's 2PP tally, which he refers to as excess.

Different conceptions of "wastage"

PR advocates who refer to unrepresented votes for the loser as "wasted" are using a term that also has a specific meaning in criticising first past the post and PR systems with threshholds.  That meaning is important and different and in my view the term should be largely reserved for that context.  

In first past the post, any vote for a candidate who does not finish in the top two in the seat not only has no effect on who wins the seat but also has no effect on the margin.  In terms of how much the winner won by, this vote may as well not have existed and the voter may as well not have bothered voting.  This creates a strategic dilemma for the voter thinking of voting for a candidate who appears unlikely to finish in the top two.  Should they vote with their heart and risk wasting their vote on the third placed finisher and thereby perhaps helping the nastier of the leading two candidates to win?  Or should they sell out and vote strategically for the more palatable of the expected top two, perhaps at the risk that if their ideal candidate does better than expected they might help the mediocre candidate beat the good candidate, or even cause the bad candidate to beat both?   It's in fact impossible for even expert level voters armed with polling-based models to be able to predict vote shares accurately enough to be sure of making the best strategic choice. The fact that this strategic dilemma applies to quite a lot of voters but nowhere near all means that first past the post discriminates between voters and means that it is a violation of what should be considered basic rights to equal treatment.  There may be excuses for some countries that have it to keep it, but there is no excuse for any country that has ditched it to go back.

In contrast, when some PR advocates try to call the votes that end up with the loser (or exhausted in a system where that's possible) "wasted", all they mean is that those votes did not finish up with anyone who won a seat.  This differs from FPTP in that there is not a "wasted vote problem" that subjects a voter to said tactical dilemma; it's just the case that elections have winners and losers, and single-seat systems have a lot of votes that are for, or in Australia's case finish with, losing candidates.

A better term to avoid confusion is "unrepresented" - in a PR system a much higher proportion of votes do end up directly with someone who wins than in any single seat system.  In 2025, on average the 2CP winner of each Reps seat finished up with 59.5% of the preferences in that seat and the other 40.5% finished up with the loser.  This figure doesn't change a great deal between elections.  In contrast in a Senate count it's normal for about 84% of vote values to finish up with a winner.  (In fact the proportion of voters whose votes at least partially contribute to electing someone is higher than 84%, and the proportion of voters whose votes entirely do so is lower, but never mind that for now.)  PR systems that use a very low quota can have even lower "unrepresented" vote levels than the c. 16% in Senate, though many such systems use threshholds that tend to result in about 10% of voters voting for parties that don't win a seat (these votes are then "wasted" just like votes for uncompetitive candidates in FPTP).

I actually don't think this unrepresented vote argument is a valid argument against our Reps system anyway. In a party system it's effectively a junk statistic.  It's true that about 40% of votes habitually don't end up with a seat winner, but in the case of votes for losing major party candidates there will (except in WA 2021 style wipeouts) be plenty of members of the party elected elsewhere, and the party may even win.  Every preference that reaches a major party that loses the seat contest in at least the competitive seats is a vote where the voter played a role in making the winning party work for that seat, which may in turn have helped the party that lost that seat to win elsewhere.  For some voters, not electing a specific major party candidate in their seat will seriously affect their view of the success of their voting experience, but the vast majority will care far more about the overall result.   The more effective argument for PR - which is not to say there are no counter-arguments - is overall disproportionality in single-member districts.   The leading party (in a close election both leading parties) gets a lot more seats than its vote share while minor parties with dispersed but reasonable support win few if any.  

Your Vote For The Winner Was Wasted??

Where wasted vote guy departs from the usual pro-PR complaint about votes that don't end up with the winner is that his conception of "wasted votes" includes many votes for the winner as well!  Specifically he regards the number of votes the 2PP winner gets above what they would have needed to beat the loser by one vote as "wasted", so if one side wins a seat 60-40, then he calls the 40% for the loser and also 20% from the winner's 60% "wasted", for a total of 60% supposedly wasted.  In optional preferential voting he includes all the exhausted votes as "waste" as well.

Now the first problem here is that this is simply absurd.  Someone who voted for the seat winner in our system is not going to think that their vote was wasted just because they could have voted for the other side and the result would have been the same.  They could (mileage varies) feel their vote for the winner was pointless if they live in an extremely safe seat where the result was never at any stage in doubt, but that's a smaller subset of the supposed wasted excess.  Victory for someone who really cares about their seat result is a collective experience and a win is a win is a win, especially if it's anywhere near a close one.  

There are some systems where a voter for the winner might have a genuine regret - if corrupt boundary drawers had deliberately put that voter in a seat their side was going to win anyway to stop them having more impact somewhere else.  The concept of wastage is useful in assessing deliberate gerrymandering of that sort (see below) but this does not happen in our system.  When one of the majors from time to time does slightly better than the other in terms of the point on the 2PP pendulum at which the two would break even on 2PP, this is because its vote is more efficiently distributed - through strategic, policy and campaign choices as well as personal vote effects, it just happens to be outdoing its opponent in bang for buck.  It is not because of any form of rigging.

In trying to avoid the absurdity of identifying individual "wasted votes" for winners, WVG has previously claimed that winning votes are "fungible", so that you know there are a certain number of wasted winning votes but nobody in particular's winning votes were identifiably wasted.  However WVG has also recently, perhaps out of rustiness as he's not been very active lately, claimed that "[..] 60% of voters receive literally zero representation (conceptually their votes are thrown in the bin due to being losing/excess winning single member votes)".  Well sure, if your concepualisation is daft and serves no purpose other than to prop up your own hostility to our system.  

A far more sensible way to look at it - and much more in keeping with WVG's "water" analogy - would be the way surpluses are distributed in the Senate.  When a candidate polls over a quota on primaries in the Senate every vote for the winner contributes an equal part of its value to electing that candidate, and the remainder of every vote is surplus to that candidate's requirements and flows on.  Likewise in the Reps, every voter whose vote contributes to the winner's election has equally helped the winner to win, and no winning voter is unrepresented.  As with a surplus over quota, it only makes sense here to talk about a surplus of total value over what is required.  It does not make any sense at all to say that there were any individual votes for the winner that did not contribute to the overall result.

A further issue I have raised with WVG's "wasted" mathematics is that if one assumes that all voters who voted formally were always going to do so, the excess in a compulsory preferencing election needs to be halved, because in that case if the voter didn't preference the winner they would preference their opponent, a net change of -2 to the margin.  

Herpetpsephology Fail!

As if dragging in the language of "wasted votes" from anti-FPTP theory is not bad enough, WVG also tries to pretend that the maintenance of our current single member system is a form of "gerrymandering".  The electoral map amphibian that is the gerrymander is widely misidentified in Australia (often being used to refer to malapportionment, for instance) so he is hardly unusual in this.  Gerrymandering is actually the manipulation of electoral boundaries to deliberately achieve certain results.  It is classically used as a pejorative for cases where, given a population relatively evenly split between two parties, the mapmaker might create five seats that fairly narrowly but comfortably enough favour one party and two that massively favour the others.  (It should be noted here that in some cases this kind of process can be used as a force for good - if an area too heavily favours one party it is sometimes good to draw a district that favours the minority so that they get some representation rather than none.)

Wasted vote guy has run a completely tinfoil (and IMO not completely intellectually honest) argument in which parties that vote to maintain the current Reps system are supposedly "engaged in the political manipulation of electoral district boundaries with the intent to create undue advantage".  In fact they're not manipulating the boundaries, they're simply preserving a system that has that many boundaries, and that same number of boundaries could be put anywhere within practicable reason and WVG would still complain and allege it was all a wasted vote plot.  

A further point here is that WVG in 2023 complained about the Nationals and teals doing so well in seat terms relative to their vote, but failed to mention that this is not only down to concentration of support but is also down to both these forces not running everywhere and, in the Nationals' case, being shielded from a potentially effective competitor (the Liberals) by the latter's voluntary withdrawal from nearly all the seats they contest.  

PR Obsessives For First Past The Post!

Another characteristic of WVG is the way he quite often flirts with anti-preferencer talking points.  The tweet depicted on this thread is an example (suggesting Albanese didn't really win, making a big fuss about Labor's modest primary vote and attempting to stir up contempt for the Prime Minister) but the Jacobin article contained others.  Despite him being outraged by me pointing out that he was pushing FPTP tropes, he did in fact in his Jacobin article claim that the Coalition would have won in 2022 under first past the post (they only led on primaries in 73 seats, which might or might not have been enough to cobble a minority government together, and would most likely not have even won all of those with strategic voting at play).  

He also said that "when used in combination with single-member electorates, it nevertheless creates a mechanism that reinforces the two-party duopoly on power by funneling minor-party votes back to major ones." (This is not true, it is the single-member electorates by themselves that are the issue, preferences tend to counteract it).  

Finally the "so-called illegitimacy" comment deserves a response, because I've unfortunately seen a fair few left-wing posters retweet claims that Maduro was a legitimate President into my feed - no it does not follow that just because he was abducted by Trump he was therefore a good guy.  The Venezuelan Presidential election "won" by Maduro was a massive, blatant and comically inept case of vote figure fraud, as confirmed by rounding fraud evidence, collected vote total returns, polling, exit polling and the lack of satisfactory official figures.  Nobody who knows anything about elections would use "so-called" in this context.  

That's the end of episode 1 (beyond any updates I may add based on the inevitable reaction).  There are at least another two I have thought of that may appear in this series in the future!

Helen Burnet Quits The Greens!

Witnesses to political history

Tasmanian politics has seldom seemed sober for long since the day in 2023 when two Liberal backbenchers announced they were quitting the government over the Macquarie Point stadium and other things.  2025 was especially deranged but things did seem to have largely settled down once it became obvious that the Rockliff Government would continue in office as nobody could be bothered removing it.  Would 2026 be a sane and normal year in Tasmanian politics?  Nope, we were only on day two before the familiar cries of "go home #politas you're drunk" again rang out among politics tragics as the scene reeled from another shock announcement.  In this case, it's that Clark MHA Helen Burnet, a continuously elected Green at council or state level for a state record of over 20 years had fronted the media in the North Hobart wombat sculpture park to declare that she had quit the party.  There are now six independents in the parliament, the most since the 1909 adoption of statewide Hare-Clark.  

At local council level, it's a common career path for candidates to be elected as Greens then become independents (usually as the end of their first term approaches) but Burnet is the first of 18 state-level Tasmanian Greens MPs to leave the party while in state parliament.  Around the country such defections have not been all that rare and I count six others at state level and two in the Senate (one of these, Dorinda Cox, to Labor).  About half of those defections were triggered by personal controversies.  This also makes this the fourth term of state parliament in a row to witness a defection of some kind.  

2025 Site Review

At the end of each year I post a review of the activities on this site in that year and 2025 was a big one.  For the first time since I started this site in 2012 there were Tasmanian and federal elections in the same year.  Not only that but they were back to back with one being caused before the dust from the other had fully settled.  

The following graph tells the story of the year in terms of user numbers per week.  This spiked at over 20,000 during the federal election and there was another big lift for the Tasmanian state election.  There were also smaller lifts from the Victorian by-elections that seem like about five years ago and the WA state election.  From about September on though there wasn't much going on.

Site activity as measured in total events was up 98% on 2024, which was itself probably the second-busiest year in the site's history, making 2025 easily the biggest year to date, overtaking 2022.  (Comparing 2022 and 2025 exactly is difficult because of Google's disgraceful handling of the transition from Universal Analytics to Analytics 4).  

Australia's Worst Oppositions: Phase 2 Not-A-Poll Results

 Secular seasons greetings and goodwill to all readers.  As noted almost every year it's an almost annual tradition on this site to release something every Christmas Day.  Click the Xmas tag for previous random examples.  Why do I do this?  Partly it's a present for those who like Christmas, which in an often lethargic and non-religious fashion includes your host, but it's also a present for those who don't want to deal with this particular Christmas or even generally cannot stand Christmas and just wish it was a normal day when normal things happened.  And what could be more normal than the results of a Not-A-Poll in this website's sidebar?  Therefore, the campaign against compulsory Christmasing brings you again ... whatever this is.  

This one is a very token present to you all and I can report that there is something far more thoroughly wonky and distinctive re the 2025 federal election and other recent elections in the pipeline, but it's not ready yet.  In truth, I've spent the whole year trying to recover my spare time and the volume of 2025 election detail I'd like to be posting on here from that moment when just as I was getting other things back on track after the federal election came the initially bold and exciting news that Dean Winter had placed a no-confidence motion against the Rockliff Liberal Government on the notice-paper.  

2025 Polling Year In Review

2PP Average For Year 53.9 to ALP (+3.0)

At the end of each year I release a review of the year in federal polling.  See the 2024 article here (ah, those days when the Coalition still seemed competitive!) and/or click on the "annual poll review" tab for all articles back to 2014.  As usual if any late polls arrive I will edit this article to update the relevant statistics.  

2025 was overall not a good year for the Australian polling industry.  The year started well with good results for the two main pollsters in the field for the WA state election but this was followed by a rather serious federal election miss, where pollsters had the right winner but were almost as far off on the 2PP and primary votes as when they were wrong in 2019.  As they had the right winner anyway there was little media interest in the error; there has also been precious little visible introspection from the industry since and none of the attempts at review processes we saw following 2019.  The Tasmanian state election was then a mixed bag with a good result for local firm EMRS but a big shocker for YouGov.  On the plus side, we are seeing more diversity in Australian polling, but still the average transparency level is ordinary.

This year welcomed one new entrant, Spectre, and also saw a low-key and unsuccessful return from Ipsos and a great step up in activity from DemosAU.  

EMRS: Liberals Slip But Labor Again Doesn't Pick Up The Scraps

EMRS Lib 34 ALP 25 Green 17 IND 19 others 5

Seat estimate off this poll if election "held now" Lib 13 (-1) ALP 10 Green 6 (+1) IND 5 SFF 1

Highest Green primary since November 2017

Another EMRS poll is out though at this stage the online documentation seems a little sparser than the usual.  I've seen a tweet and various secondary reporting, but not yet the usual poll report [EDIT 25/11: it is up now]. The poll is also missing the usual preferred Premier figure as a result of a "coding error".  My understanding is the error was the inclusion of Greens leader Rosalie Woodruff in at least the online component of the poll.  (EMRS used to do three-way preferred Premier up til 2014 but it tended to be misleading because of splitting of left respondents between the Labor and Green leaders, making it hard to compare to similar polls interstate).  

After pretty good numbers in the previous EMRS and the recent DemosAU this one would come as a bit of a downer for the government.  But in a movie we've seen quite a few times before, when the government takes a hit in the polls it's mostly not the Labor Opposition that mops up the slops, instead it is somebody else.  In this case it's the Greens, and I was surprised to find that the last time EMRS had the Greens on 17 or above was in November 2017.   In those days EMRS tended to overestimate the Greens at election time, something that became less apparent following significant methods changes in 2017.  The poll also has a high, though probably inflated, independent vote (we saw at the state election that 19% independent in generic polling is worth about 15 at the ballot box).  

With the splintered state of Tasmania's non-government forces at the moment we're increasingly seeing a broad green left (Greens plus greenish independents, who are most of the "independent" vote) outpolling the official Labor opposition.  

This poll also includes net likeability scores, in which Premier Rockliff has taken a tumble back to net -1 from net +18 three months ago.  Josh Willie has also dropped, from net +4 to net -4.  I don't at this stage have more detailed breakdowns.  

As a seat projection, in an election "held now" if this poll was accurate, the Greens would gain a seat from the Liberals in Braddon.  I wouldn't expect any change elsewhere as the Liberals were not far from winning in most of the remaining seats.  In Bass, the swing from the Liberals to the Greens would significantly change the exclusion order, pushing the second Green above George Razay at the point where they were excluded, but I suspect that Razay would overhaul them on preferences from fellow Independents and SF+F then go on to win anyway.  (If one allows for the Independent vote being overstated then Labor has slightly more votes than at the state election, but the Greens have significantly more, which also helps Razay if he is over them.)

What has happened here?  The changes from the last poll aren't massive, only just outside the in-theory "margin of error" and maybe the high Green vote is a blip.  But this poll does tie in with something seen in the state election campaign, that when the Macquarie Point stadium proposal dominates the news cycle, the government's polling tanked.  Whenever the government was talking about anything else, yes even TasInsure, things got better.  There might also be some sense of anticlimax that the Budget isn't any different on the debt front than its maligned predecessor, and the government did also encounter what should have been a scandal last week, but I am doubtful these were major factors.  

There's also still the possibility that EMRS's new half-online half-phone methods could have some impact on their results.  The EMRS online panel is opt-in and probably has a lot of politically engaged voters.  However the dual method was also used in their internal Liberal polling for the state election, which was accurate.  So I'm going with stadium fatigue as my primary hypothesis for this one!

Brad Battin's Booting Is A Poll History Outlier

(Note for Tasmanian audiences: I will have the usual article about the latest EMRS poll up sometime tonight or tomorrow)

The last fortnight has seen the leaders of three of Australia's current crop of feeble oppositions displaced while a fourth, Sussan Ley, is precarious (especially after another shocker Newspoll).  First to go was ACT Liberal leader Leanne Castley, who after an irregular (by Liberal Party standards) attempt to kick backbenchers Elizabeth Lee and Peter Cain out of the party room for crossing the floor over parliamentary sitting hours, resigned suddenly on November 10.  Victorian leader Brad Battin didn't recontest after a spill motion was passed 19-13 on Nov 18, and NSW Liberal leader Mark Speakman quit on Nov 21 facing the same fate.  In the ACT the leadership has passed to an old hand in third-term MLA Mark Parton, but both Victoria and NSW have gone for first-term female MPs, Jess Wilson and Kellie Sloane.  

The removal of Speakman requires no explanation following the NSW Liberals' run of poor polling and a terrible result in the Kiama by-election.  But the case of Battin is very unusual.  Since assuming the leadership in the final days of 2024, Brad Battin had led Jacinta Allan as better Premier in every single poll that asked such a question - eight in all by four different pollsters, with an average margin of 8.75%.  There have been occasional cases of Opposition Leaders being rolled who had sometimes led as better Premier (including Battin's predecessor John Pesutto who had led by a pyrrhic single point in his final Resolve sample) but an Opposition Leader who was leading solidly being removed is unprecedented, at least in the Newspoll era.

Normally, two things are more or less the death rattle for incumbent Premiers - personal netsats in negative double digits and trailing as better Premier when not facing a previous Premier as Opposition Leader.   And Jacinta Allan's personal ratings are not just run of the mill bad; in four polls taken while opposite Battin she averaged net -31.8.  Sure, Labor could replace her with Ben Carroll, but that would make them a three-leader government, and those have a poor past record with four losses from five attempts at state and territory level in the last fifty years, three of those heavy.  So there's an argument that Battin has been removed while he still had a very good chance of winning.

That is not to say that the Liberals were polling in an election-winning position at the time.  Prior to Battin's removal, the Victorian Coalition in the last six months had led 51-49 in two DemosAU polls and one Freshwater, and trailed 52-48 in two Redbridges and 51.5-48.5 in an earlier Redbridge, trailed 53-47 in the June Newspoll and trailed by my estimates 53-47 then 52.7-47.3 in two Resolves (Resolve doesn't reliably publish 2PP estimates).  All these 2PPs are rubbery because nobody really knows what the field will be and most importantly whether One Nation (currently polling through the roof federally) will run a serious statewide lower house campaign in the state election.  But even at the better end of the spectrum, 51-49 for the Coalition is probably not enough - though it's closer to enough than I realised before modelling it in detail.  

The reason for this lies in Victoria's tilted pendulum.  The 2022 election saw a shovelling of Labor seats from both the 2PP-marginal and 2PP-safe categories into the moderately safe range, meaning that by uniform swing the Coalition needs to poll into the 52s to take government (assuming that such Greens as win will not assist them).  Indeed a uniform swing of 6% would only yield eight seats, leaving Labor in majority government with about 48 seats (assuming the recaptures of Ringwood and South Barwon) to 37 for the Coalition and 3 for the Greens.  

Things are different on a probabilistic assessment because Labor holds a massive 18 seats on margins between 6% and 9% and on a swing around 6% the Coalition would, through variable swing, be bound to jag at least a few of those.  My usual model estimates 12-13 gains rather than the uniform swing gain of 8 seats, but that's only enough to put Labor into minority. For 16 gains and outright victory the Coalition probably needs more like 52-48 - though even if one or two seats short of that there always might be some independent or ALP discard who might help the Coalition into government. And one of these days Victoria, which alone of the states and territories has an even number of lower house seats, might get a deadlocked parliament (that would be interesting!)

Even so, the fact that the existing 51-49 readings came from Freshwater (the reputation of which took a serious knock at the 2025 federal election) and DemosAU (which is a relatively new pollster) probably meant that even these readings were treated with some scepticism.

Does it matter if an opposition is merely competitive and not actually winning against an unpopular Premier?  History suggests not really - unpopular Premiers find ways to get removed or lose, even if the 2PP polling is OK some time out.  Governments that appear to be polling OK also often nosedive in the final year if they are "federally dragged", meaning that the same party is in power federally (which is in general a large disadvantage at elections).  It can also be argued that the state Coalition parties have taken brand damage from the 2025 federal election, and that once the Albanese Government's federal honeymoon tapers off we could see a return to politics as normal - on which basis the Allan Government is very beatable.

Overall by historic standards there's a strong polling case that Battin still had a good chance of winning.  Why then was he removed?

The leadership change can be sourced to some obvious factors.  While Battin's personal polling was generally good (a minor exception being a net -5 satisfaction in the June Newspoll, though that was far from terrible) the Victorian Liberals had remained a mess on his watch.  A very poorly internally received reshuffle was a factor here.  There was also a perception that Battin's leadership style was too limited with too much of a focus on crime.  Law and order campaigning doesn't tend to win Australian elections by itself - even when crime is being talked about an unusual amount it tends to be only very salient for a few to several percent of voters.  It's also too easy for state governments to take action on compared to the economy and cost of living.  There was some perception hence that Battin wasn't up to winning an election.  Historically it was way short of a clearcut case.  The risk is that if the partyroom waited for clearer evidence to emerge, it could be too late for a new leader (especially a young new leader) to get a sufficient run in to the next election.

The leadership change might also be seen as running up the white flag on an outer suburbs dominated strategy.  In the leadup to the federal election we were bombarded with arguments that federal Labor was on the way out in the outer suburbs and if the Coalition focused there it could get large swings and win many seats.  In fact federal Labor wasn't on the way out anywhere, but despite the Coalition under Peter Dutton relentlessly focusing on such seats (egged on by internal polls that were wrong), they didn't get anywhere near winning most of them and would not have done so even had there not been a national swing against them.  Uniformly removing Labor's 2PP swing from the federal election would have gained the Coalition only Bullwinkel, Solomon and Bendigo and the Coalition would have still lost Dickson, Petrie, Leichhardt, Sturt, Bass and Braddon.

A few years back the theory was that Australia was realigning with higher educated inner suburbs moving to Labor and lower educated working class areas moving to the Coalition but with the current splintering of vote share on the right and following Labor's stellar performance in northern Tasmania in the federal election all that Piketty 101 stuff is feeling very 2022.  The Coalition won't win elections just in the outer suburbs, but leaders who are seen as too inner city might still play badly there, and this is a risk factor with Wilson.  Overall of the eighteen seats in the critical 6-9% range eight are outer suburban with four in the west (where Labor did badly in swing terms last time) and four in the east (where Labor held up well) but there was a lot of COVID issue overlay in the 2022 state and federal spatial vote patterns.  

The latest Newspoll

The first Newspoll of Jess Wilson's career was very quick out of the blocks and this makes it somewhat difficult to compare it with past first Newspolls for state Opposition Leaders.  Pre-2015 state Newspolls outside of campaign periods were aggregates taken from two or three months of polling (similar to Resolve now).  The most notable things in the new Newspoll are firstly Jacinta Allan's terrible -42 net satisfaction and secondly Wilson's 47-33 lead as Better Premier.  Allan's rating is nearly the worst for a Premier in Newspoll history (only John Cain and Anna Bligh at net -43 have been below it).  The only new Opposition Leaders who had not previously led their parties to jump to comparable leads in their first poll were Dean Brown in SA (19% lead) and Campbell Newman in Queensland (12%), and those were polls taken over longer periods.  Both went on to win elections.  

Frequently new Opposition Leaders are slow to get on the board as Better Premier because they are still building their profile.  On this basis a 14% lead after only a few days looks stellar, but because we don't often see such a quick poll after a change, we need to see more over coming weeks and months to be sure this isn't just some instant sugar hit for changing leader against a disliked government.  The other thing that was surprising here was Wilson's personal ratings at 33% satisfied 31% dissatisfied - the latter seems like a high figure for a new Opposition Leader who is also a first termer.  

History suggests it doesn't really matter if state Oppositions are a mess through their term so long as they can get their act together by election day.  The classic case was the WA Liberals in the 2005-8 term where after churning through four leadership changes in a term they had no-one left but to go back to Colin Barnett who had lost the previous election.  Barnett was however able to win (very narrowly).  

Heading into election year 2026 both sides in Victoria have reason to be nervous.  Labor has a Premier way underwater by historic standards and the government will be twelve years old and federally dragged.  The Coalition however faces federal brand damage, the splintering of the right-wing vote and the possibility that federal drag is not going to be such a big thing - as well as its own form for chaos. 

DemosAU: Liberals Increase Lead Ahead Of Budget

DemosAU: Lib 41 Labor 24 Green 15 IND 14 SFF 2 others 4
Liberals would be re-elected in minority, probably gaining one seat

A surprise DemosAU Tasmanian state poll has appeared.  The government will be grinning with a 17 point primary vote lead.  If there is a hostile reception to the coming interim Budget then the grin might not last too long.  That said, there have been some advance signs that the medicine won't be too harsh.  

This poll was self-initiated (not commissioned by anyone) and taken from Oct 16-27 with a sample size of 1021.  DemosAU scrubbed up pretty well in the state election, though not as well as EMRS.  They did significantly underestimate the Liberal primary and, like all pollsters, overestimate Independents, but their overall read of Labor's poor prospects in particular was on the money and their individual candidate breakdowns were very handy (for more detail on that see here). 

The Independent reading in this poll is noticeably lower than the c.19% readings seen before the election.  It's possible the use of both 2025 state and federal election voting as weightings will have toned down any impact of an overengaged sample on the previously over-polled independent vote, but I don't think that's the main reason why independent votes get overestimated in Tasmania anyway.  Rather I think some voters are looking for an independent but at state level never find the right one.  If that's so, this poll might be taken as pointing to some softening in independent support.  The August EMRS had found no such softening; the November EMRS will be interesting in this regard.