Talk of a potential boilover is increasingly focusing on the possibility of a radical change in preference flows, which loom ever larger in pollsters’ two-party preferred calculations as the major party vote inexorably declines. A piece in Inside Story from Peter Brent in February nicely summarised the conventional wisdom surrounding the matter: previous election flows are typically more reliable than respondent allocation, because some voters are guided by a how-to-vote card they don’t have when responding to an opinion poll. However, it is clear that the flow of preferences from One Nation to the Coalition have strengthened over time, and this is the basis of claims that pollsters’ two-party preferred results are selling the Coalition short.
The Financial Review reported yesterday that polling by JWS Research suggested it to have reached a game-changing 90% in a number of outer suburban seats that are normally safe for Labor, though this seems far too radical a change to be plausible and would in any case have been based off small samples. The was pursued in a column today talking up the Coalition’s chances in The Nightly by Andrew Carswell, a former press secretary to Scott Morrison, again with reference to polling from JWS Research.
Carswell makes the incorrect claim that the “vast majority of current polls … oddly determine preference flows based on 2022”: in fact, only RedBridge Group and DemosAU do so (and the former has included a respondent-allocated preferences result in its latest poll, as discussed in my other post). Roy Morgan, for one, uses respondent-allocated preferences for its headline figures, as well as providing a measure from the 2022 election. So, apparently, does Freshwater Strategy, which has been the strongest series for the Coalition for most of the term – though this has had a lot more to do with its reading of the primary vote than preference flows. Certainly it did so for its New South Wales election state polling, though beyond that I can find no hard information (UPDATE: Kevin Bonham notes in comments that a results display from the pollster on the Financial Review website specifies that non-major party votes are asked who they would favour if they had to choose).
The chart below records the overall share of independent and minor party preferences to Labor implied by the respondent-allocated two-party results of Morgan and Freshwater. That data in the former case is voluminous enough to allow for a trend calculation – the latter just joins the dots. The apparent decline in Freshwater Strategy’s flow to Labor this year is tempered by the fact that had in fact been crediting Labor throughout 2024 with even stronger preference flows than in 2022 – nonetheless, most of its two-party numbers this year would likely have come in a point higher for Labor if 2022 preference flows had been applied. Roy Morgan’s flows to Labor have consistently been even stronger to Labor than in 2022 (note that it has nonetheless produced stronger two-party preferred results for Labor on its previous election preferences measure because of its elevated result for independents, 63.8% of which has been allocated to Labor).
YouGov has lately pursued a middle course, based on some manner of blend of the 2022 flows and respondent-allocated preferences from the vast surveys the company has conducted for its MRP surveys. Its most recent methodology statement explains it has been giving Labor 80% of Greens preferences (85.7%), 59% from independents (63.8%, with help from teals), 33% from One Nation (35.7%, as noted) and 49% from the rest (45.3%). As of early February, Newspoll has bumped up the share of One Nation preferences to Labor by an unspecified amount, partly inspired by their 73.8% flow to the LNP at the Queensland election in October, compared with 66.9% in that state at the 2022 federal election. The observed difference since would neatly be explained if the share had increased from 64.3% to 70%.
Interestingly, none of the talk relates to the possibility that a record-breaking 85.7% flow of Greens preferences to Labor might not be repeated three years after the Morrison government made way for a Labor regime that has regularly disappointed progressive sentiment in relation to environmental approvals and the war in Gaza. If YouGov has it right though, this factor will do far more damage to Labor’s two-party share than anything Hanson or Palmer voters have in store.
I haven’t troubled to accommodate any of this in BludgerTrack, which follows 2022 election flows straight down the line: 85.7% to Labor from the Greens, 35.7% from One Nation and 50.0% from the remainder. Despite this, I will share in the general surprise if Labor indeed does that well on preferences this time. What follows is an attempt to determine the two-party vote under various proposed scenarios. The first step was to produce a crude but plausible-looking breakdown of the 12.7% “others” by splitting the difference between the most recent results from YouGov and DemosAU, both of whom provide results for Trumpet of Patriots and independents.
| Primary | 2022 | YouGov | Newspoll | Maximal | |
| ALP | 32.8% | 52.7% | 52.0% | 52.2% | 50.1% |
| L-NP | 34.3% | 47.3% | 48.0% | 47.8% | 49.9% |
| GRN | 12.4% | ||||
| ON | 7.8% | ||||
| UAP | 3.0% | ||||
| IND | 5.1% | ||||
| OTH | 4.7% |
The 2022 result is self-explanatory; YouGov uses the numbers from that pollster detailed above; Newspoll presumes I’m right about One Nation preferences splitting 70-30 and otherwise being unchanged from 2022; and “Maximal” indulges the claims being made from the JWS Research polling by inflating One Nation’s flow to the Coalition to 80% (from 65.2% in 2019 and 64.3% in 2022) and Trumpet of Patriots to 75% (from 65.1% for the United Australia Party in 2019 and 61.8% in 2022).
UPDATE: I should at least have mentioned Resolve Strategic, which got neglected because it rarely reports two-party results. It has during the campaign though, and the outfit’s director, Jim Reed, has emailed to say its respondent-allocated preferences have 78% of One Nation voters favouring the Coalition. However, Greens and teal independents have continued to favour Labor strongly enough that its overall preference flows have not been that different from 2022.
UPDATE 2: Kos Samaras of RedBridge Group relates: “We tested for any difference between historical preference flows and respondent-reported flows. There was none. ”
Measuring the possibilities of the crackling on an election roast!
I tend to think that the flow of preferences from One Nation to the Coalition will not be 80% either. Maybe increased from the historical average but not as much as going from ~65% to 80%. I think it might get up to 70-72%. Anyway I will have one data point from the booth I’m scrutineering on Saturday night. 🙂
What we are really dealing here is the limitations of opinion polling in a period in which the voting behaviour of the electorate is becoming increasingly diverse and complex.
As I have been saying all along, the important question in this election is not a 2PP one, because the Coalition has never had the remotest chance of winning a majority. The important question is whether Labor will be returned with a slim majority, or else there will be a hung parliament. A secondary question is, in a hung parliament scenario, whether Labor would be able to form a minority government without the toxic support of the Greens. (Note to Greens supporters on here: I’m not saying that the Greens are inherently toxic – although I am definitely not a fan of theirs in their post-Di Natale manifestation – but that it is toxic for Labor to depend on their support to form a government.)
The pollsters can make reasonable predictions about how the size of the first preference votes for Labor, the Coalition and the Greens, and also to some extent about other minor parties such as One Nation and Clive Palmer’s outfit (I can’t bring myself to type its ludicrous name). It gets trickier for them with the vote for independents, which I suspect is going to increase a bit this time. And I don’t think anyone has yet come up with a reliable methodology for predicting how preferences are going to go this time: including the increasingly important question of where major party preferences go in seats in which one of them finishes third.
For instance, when I look at Franklin, the seat in which I live, where the conventional wisdom is that most Liberal voters will follow their how to vote card and put Julie Collins before Peter George. But there is presumably a proportion of Liberal voters who always put Labor last (or second last, above the Greens). And another group who just do a donkey vote. Under both these scenarios, some Liberals will put George ahead of Labor. I’m not expecting George to get enough votes to make a fight of it but, if he were to do so, it’s really difficult to predict what proportion of Liberal preferences he might get.
Anyway, In terms of how to deal with these questions at an aggregated national level, I believe, as is usually the case, that William has got it 100% right. The only sensible thing to do with preferences is to stick with the time-honoured practice of allocating them as per the last election. The samples taken by pollsters are far too small to make respondent-allocated preferences at all reliable.
And William is also 100% right in feeling that allocating preferences as per the last election is almost certainly going to overstate the ALP 2PP vote. But there’s no systematic way of coming up with a lower 2PP estimate for Labor that is based on anything much more than a guess. So we’re just going to have to wait for election night to see what the preference flows here there and everywhere are going to be.
I think Kevin Bonham adds a full 0.5% to the 2PP figure for the Coalition to account for changing preference flows from 2022 to now.
0.5 up for LNP is also the current Newspoll adjustment and sounds like William pretty much agrees with this.
“ C@tmomma says:
Thursday, May 1, 2025 at 5:45 am
I tend to think that the flow of preferences from One Nation to the Coalition will not be 80% either. Maybe increased from the historical average but not as much as going from ~65% to 80%. I think it might get up to 70-72%. Anyway I will have one data point from the booth I’m scrutineering on Saturday night. ”
____
I think nationally the flow of ON preferences to the LNP will be over 70%: probably on account of a larger ON primary vote caused by an extra cohort of traditional LNP voters putting a 1 next to ON on their ballot paper as a ‘signal’ to Dutton to be just a little bit more racist, sexist and thuggish next time, before running home to mummy and placing a ‘2’ next to the LNP candidate (may be a 3 if they are also lured to put a 2 next to Trumpet on the way home).
The vexed issue is whether this 70%+ of ON preferences will be higher in seats like Werriwa, Hunter and Lalor AND whether that vote will include a cohort of disaffected labor voters. THIS is what Dutton and Littleproud have bet the house on when they did the ON preference deals and Dutton went into overdrive on Cultcha War punching down campaigning this past fortnight. Freshwater seems to be giving them ‘intelligence’ that this ‘cunning plan’ is actually an excellent tactic.
We just don’t know yet how effective it will be, but I suspect that the tactic will simply bake in (hello Nadia if you are out there) the dynamics in the rest of the election: Labor solidifies its position on all of its 2022 gains, as do the Teals and at best the LNP pick up 2-6 outer Urbs/regional centre seats – BUT still leaving them a million miles away from forming governmnet …. Not to mention doing irreparable damage to their ‘broad church’ brand. That’s the ‘best case’ outcome for the LNP. It’s as likely that they will not pick up more than 2-3 seats from the ALP nationally and these will be offset by loses across the board elsewhere.
This is a very interesting pseph issue, but I still see no reason to change from my ‘NY Punt’ that labor will end up with 75 seats (+/- 3 to 5 seats) and will form a stable government with very little extra scope for the Greens terrorists to do any more damage to Labor’s legislative agenda than it did over the past 3 years – if they chose to persist with this disgusting tactic.
As a curiosity, is there any way to access where bludger track was at historically? I’d like to know the precise figures for 1 month before the budget (which was also the week the campaign was officially launched). So I’m looking for late February/early march results.
There’s on for asking is that today’s updated bludgertrack has Labor up 2.6% on 2PP and 1.8% on primaries ‘since the budget’, but the graphs tend to indicate that the trend uptick in both started to occur about a month prior to the budget (which accords with my memory as well). … I’d just like to get some precise figures on that.
Andrew
You are right that I put my comments on the wrong thread. Further though my comment about this sites credibility being shot if the Polls are wrong has been stated in different ways by many different politcal commentators in different ways and is more a statement of fact. If the polls have this one wrong then what is the point of them going forward. I would also add at the very start of this campaign Anthony Green stated leaving no grey area that 2pp means very little in many of the seats that will decide this election, he was very strong in his wording I am being very careful as I don’t want to verbal him.
William, an excellent and informative article, thank you.
“ Steelydan says:
Thursday, May 1, 2025 at 10:06 am”
_____
fair enough, and I apologise for intemperate tone.
The info that Freshwater uses respondent prefs was apparently in this paywalled article; despite having ways around most paywalls this one is eluding me. If one hovers above the 2PP graphic there is a “If you had to choose [..]” question.
https://www.afr.com/politics/federal/freshwater-strategy-poll-20250128-p5l7p3
The chance of PHON voters being as disciplined in their preferences as Greens voters in 2022 is negligible. I would put an 80% flow to LNP as a conceivable upper limit, although I would think 75% would still be a pretty conservative upper limit. I will be very surprised if it is this high or higher.
Greens flows to ALP had a low of 78% in 2010 and haven’t been below 80% since. Yes, the Albanese government has been very disappointing (for Greens) but equally Dutton is a very well-known quantity. 75% flow to ALP is extremely conservative lower limit; it will almost certainly be upward of this and I will personally still be expecting upwards of 80% (although sure, maybe not 85%).
My One Nation adjustment is based on assuming a 7.5-point increase in the ON flow to Coalition, this comes out at either 0.5 points or 0.6 points after rounding and depending on my estimate of the One Nation vote at the time. At one stage I was calling it a recommended adjustment, now I’m treating it just as an alternative figure.
I agree with Martin B, except I wouldn’t be surprised if PHON prefs were 75% to Coalition, partly due to the reason Andrew Earlwood mentions above. Higher than 75% would be VERY interesting, though it would upset models a lot more if it coincided with Greens being <80% to Lab.
What puzzles me though, is the lack of discussion around the pref flows of Indys this time – probably because it's too hard and is absent of enough data, which almost needs to be at individual seat level.
But surely this is the biggest wildcard on pref flows? – given the relatively conservative economics many of them espouse, getting rid of the Coalition last time was one thing but keeping Labor in this time?
What is known about the HTV cards of the Indys running, both incumbents and challengers ? – including how they compare with 2022, and to what extent they were followed by their voters in 2022.
In terms of the Greens preference flows, the analysis at the last election showed that (from memory) around 2/3rds of Green senate voters preferenced Labor higher in NSW and Victoria than the Greens HTV directed them to.
I think the truth of the matter is that perhaps even a majority of people who vote Greens switch between Labor and the Greens based on how “progressive” Labor is presenting at the election.
The higher share of “softness” in Labor and Green votes captured by certain pollsters (eg redbridge, spectre) also support this.
So as much as Scommo may have been more on the nose in 2022 than 2019, what really was at play was a higher share of these Labor / Green voters choosing the Greens in response to Labor’s “big target” strategy in 2019 becoming a “small target” one in 2022.
The Teals also would be soaking up some of the Blue – Green vote leaving a large share of Red-Green in the Green’s PV.
TLDR: Paradoxically, more disappointment among progressives leads to the Green PV having a higher share of the Red-Green swing vote and so a higher share of Green preference flows to Labor
The ON adjustment gap just made it to 0.7 after rounding for the first time.
Good points from The Revisionist above.