DemosAU: More Friendly Fire Than Seat Swing?

DemosAU Lib 34.9 ALP 24.7 Green 15.6 Nat 2.7 SF+F 1.8 IND 20.3
Total of projected individual seat breakdowns for this poll Lib 13 ALP 10 Green 7 IND 4 Nat 1

(IND vote likely to be inflated because of format limitations)

(Green vote distribution appears unusual so real seat tally for this statewide vote share could be lower)

This article is part of my 2025 Tasmanian election coverage.  Link to main guide page including seat guides and effective voting advice.

---

One of the many remarkable elections in Tasmania was in 1986.  On the surface Robin Gray's first and only re-election sounds exremely dull; the Liberal Party won 19 seats just as it had in 1982, and Labor won 14, ditto.  Yet that election was a bloodbath with 14 incumbent MPs losing their seats, 11 in effect to their own party.  This sort of violence became less common in the 25-seat system, an especially tame case being 2006 when all 23 recontesting incumbents were returned.  (The only interesting thing about that was that nobody much thought it would occur).  Based on the recent Pulse Media DemosAU poll we could be heading for a milder case of this with at least six incumbents at serious risk of losing to their ticketmates but few clear signs of either major party gaining or losing more than the odd seat anywhere.

I have a copy of the full results of this poll [EDIT: which have now been posted publicly].  The full poll includes candidate breakdowns by electorate, which I have not seen since EMRS did it a few decades back.  While the writeup below is very detailed a reminder this is just one poll, and in the next few days I intend to produce my usual cross-poll aggregate.  

Overall the poll's results are even worse for Labor than the previous DemosAU and also the recent Liberal EMRS poll.  If this polling is correct it backs in the perception that voters don't like having their electoral sleep disrupted so soon after the federal election and are tending to blame Labor for doing so.  Sometimes Labor does a fair bit better than some of the polls suggest and this was certainly the case in 2024 when there were some very bleak polls late in the piece that proved to be not correct.  Nonetheless Labor needs to massively outperform its recent polling to have any path to governing without the Greens.  It may be that all the polls are underestimating the major party vote but at any rate it is not looking like a big surge back to the major parties after 2024, and it could even be a further swing against despite the absence of JLN.  

This particular poll is pointing to a state of things where the Liberals, unless they make major concessions to the crossbench, might continue in government but only for as long as a possibly chastened Labor wanted them to.  Even if the Liberals did a little better and got 15, they would still be likely to be dealing with a less friendly crossbench than the one they couldn't work with last time.  Labor in this poll aren't anywhere near more than 11, but even a few more would at best leave them depending on a bunch of indies who may as well be Greens.  Overall as with many recent polls the combined major party vote seems remarkably low, especially with no Lambie Network this year.  

I should unpack a few things about how the DemosAU poll works.  The poll used a combination of methods, with 76.3% IRV/robopoll, 20.2% internet panel and 3.5% random device engagement (a form of online polling that doesn't depend on membership of panels).  The robopoll component requires the respondent to be given a not too long list of options they can choose from at each step, which precludes offering the voter the list of all candidates at once.  So the poll was conducted at two stages.  The first stage gives the voter the list of the parties running and some of the key independents, who are named, and also an "Any other candidate" option.  If the voter picked a major party (or in some seats the Nationals or Greens) they were then given a follow-up to choose a candidate.  (Those undecided at this point were distributed equally across the group).  

The nine independents named were generally those with their own column, with some exceptions.  The only other candidates to be named in the first phase were the three Shooters, Fishers and Farmers lone candidates, and also Miriam Beswick (Nat) in Braddon.  (Andrew Roberts is also running as a National in Braddon but try telling that to his party who last acknowledged his existence on their official pages on July 27 2019 - there might be a reason for that).  In general I would expect the named independents to get a substantially higher response than if they were competing with named major party candidates in the same question.  If nothing else, the voter might hear the major party names and hear someone they liked who they did not realise was running in their electorate.  The crucial question is how large is this advantage.  

In my view the only way to avoid advantaging independents in current Tasmanian polling is to run an online survey that simulates the full ballot paper by listing all the candidates who are running together.  This could be done either by simulating the full ballot paper (for laptop/PC only as doing it on a mobile phone leads to exaggerated ballot order effects as shown with NSW iVote), or by offering the voter a list of parties and groups with the candidates running for each party or group shown in brackets after the party name.  I like the latter method and suggest pollsters use it in the future.

As well as the issues with the Independent vote, polling in Tasmania sometimes overestimates the Greens and sometimes doesn't.  I don't think this poll's overall figure for the Greens is implausible, but the breakdown between seats is rather polarised in their favour and the differences between some of the seats don't look quite right.  So it could be that they could get the vote suggested but not as many seats.  

Seat by seat

I give here the results for each seat in a compound form derived from the tables with my comments on the likely outcomes if the poll is broadly accurate.  I should note here that although the sample sizes are in the range 623-804 per electorate, individual electorate sampling is difficult and the individual candidate responses will often be a few dozen voters if that - some will be way off for sure.  But if one candidate's on 15% and another's on 2% that means something, and a lot of these breakdowns don't look wildly illogical.  

BASS 

Liberal 37.6 (Archer 19.6 Ferguson 7.5 Fairs 4.1 Wood 2.2 Gatenby 1.8 Sladden 1.5 Quaile 1.0)

Labor 26.0 (Finlay 9.4 Greene 5.1 Lyons 3.5 Moore 2.7 Anderson 2.1 Thomas 1.7 Gordon 1.5)

Greens 20.0

Nationals 3.5

Shooters Fishers and Farmers 0.5

IND Razay (named) 5.1

IND Pentland (named) 4.3

Any other 3.0

In this sample the Liberals have 3.01 quotas, Labor 2.08, the Greens an astonishing 1.60, Razay 0.41, Pentland 0.34. Nationals 0.28.  Obviously the Liberals get three, Labor two and the Greens one, the last one's rather messy ...

To place 20% for the Greens in context, they got 12.0% last time and have only got near 20% in their 2010 high water election where they got 21.0% there.  (The Greens were also high in the previous DemosAU wave with 18.8%).  It isn't easy to believe, and Tasmanian polls sometimes overestimate the Green vote.  But let's explore the case for this being anywhere near true: Bass hates the stadium, Labor has a Michelle O'Byrne sized hole in their ticket profile and there are no other prominent anti-stadium clearly left candidates to take votes from the Greens as there are in some other seats.  

Razay got 1.8% in 2024 having run in the 2022 federal election and been elected to Launceston council in 2022.  He has just run in another federal election as well and polled about the same as he did in 2022, so his 5.1% here gives some idea of how large I expect the advantage of being a named independent in this poll is. It's about what he got in the federal election where he was the only independent and I reckon he won't get that here though he may well do better than 2024.  If I'm right then polling 4.3% as a named independent is even worse for Pentland.  

The interesting threat to the Greens second seat here could actually be Labor.  With 2.08 quotas vs 1.6 Labor would lose, but it would not have to be much more of a gap before Labor had a path to three if they could keep their second and third candidates close to each other and ahead of the second Green - at, say, 2.3 vs 1.5 Labor would probably win.  This is especially so as the Greens suffered from high within-ticket leakage in 2024.  It's also not impossible here that the Nationals are ahead of Pentland and then there might be a fair flow from Pentland to Angela Armstrong (her former JLN ticketmate), so the Nationals might not be too far off competitive.

Estimate for this sample 3-2-2-0 (Liberal-Labor-Green-IND) but perhaps close to 3-3-1-0

BRADDON

Liberal 40.8 (Rockliff 24.6 Pearce 5.8 Ellis 5.4 Jaensch 2.4 Simpson 1.0 Parry 0.9 Wylie 0.6)

Labor 23.5 (Dow 10.4 Broad 3.4 Diprose 3.0 Hunt 2.3 Fuller 1.7 Luke 1.6 Woodhouse 1.1)

Greens 9.5

Nationals (Beswick named) 3.2

Shooters Fishers and Farmers 1.8

IND Garland (named) 13.7

IND Martin (named) 3.2

Any other 4.3

In Braddon the Liberals have 3.26 quotas, Labor 1.88, Garland 1.10, Greens 0.76 and the rest wouldn't be competitive.  This looks like a simple 3-2-1-1 with the disclaimer that the Liberals may not be far off four instead of the Greens if Roger Jaensch is close to Gavin Pearce and Felix Ellis after the distribution of the usual enormous Rockliff surplus.  Again, what they would need to do is have three candidates ahead of the lone Green, Even if he is well behind on primaries, as he is in the small Liberal sample in this poll, Jaensch has pulled this one off before, but it might be more difficult with Pearce in the mix because Pearce and Rockliff are both farmers.  It's a little surprising to be seeing polls where both Garland and the Greens might win in Braddon given there was not that much more than enough for a seat between them last time but I suspect a lot of the former JLN vote might be crossing to Garland, or to indies whose preferences will favour Garland if he needs them.  

Estimate for this sample 3-2-1-1 but perhaps close to 4-2-0-1

CLARK

Liberal 28.1 (Vermey 8.3 Behrakis 6.8 Ogilvie 6.2 Di Florio 3.4 Johnstone 1.5 Barnett 1.0 Wan 0.8)

Labor 25.6 (Willie 9.3 Haddad 7.0 McLaughlin 2.6 Martin 2.0 Kamara 1.8 Shirley 1.8 McLaren 1.2)

Greens 22.7 (Burnet 10.4 Bayley 7.1 other 5.2)

IND Johnston (named) 17.3

IND Archer (named) 4.3

Any other 2.1

In Clark the Liberals have 2.25 quotas, Labor 2.05 Greens 1.82 Johnston 1.38 Elise Archer 0.34.  Note that Johnston got 7.7% in 2024 so the high vote here could well be an overestimate from being named, but some potential vote for her was probably buried with independents not running this time (especially Sue Hickey).   There is interest here with the candidate breakdowns, which again suggest Marcus Vermey is competitive but don't show anything for Luke Martin (and also the Greens race could get spicy if their vote is lower than suggested)

Estimate for this sample 2-2-2-1

FRANKLIN

Liberal 31.7 (Abetz 10.8 Petrusma 7.6 Garvin 4.6 Street 4.4 Miller 2.0 Howlett 1.3 Young 1.0)

Labor 22.4 (Winter 10.1 Munday 4.2 Brown 4.0 Deane 2.0 Hannan 0.8 Di Virgilio 0.6 Meyers 0.6)

Greens 10.7

IND George (named) 18.2

IND O'Byrne (named) 13.7

Any other 3.4

Firstly as with the previous DemosAU poll I don't believe the Greens are getting only what they got in the seat at the federal election where their candidate was low profile and withdrew from active campaigning.  This is probably a sample pool issue.  

In this sample the Liberals are on 2.54 quotas, Labor 1.79, Greens 0.86 George 1.46 and O'Byrne 1.10.  Some of the major party candidate scores here look a bit wonky (Dean Young only 1% for instance) but the poll is consistent with Nic Street potentially struggling to keep up with Abetz and Petrusma.  An uneven split in the Liberal ticket would increase the risk of them dropping back to two; with an even split they can beat other contenders from slightly behind.  The poll also suggests that if Labor only wins two there will be a contest between Jessica Munday and Meg Brown for the second.  

I don't think George will get so close to his federal election vote and I suspect this is another case where the naming of an IND inflates their vote - but even with half of what he is polling here he would probably win.  With O'Byrne if the poll is over by, say, 5% he might come under some risk, but even that's not clear.   

Estimate for this sample 2-2-1-2 

LYONS

Liberal 35.5 (Howlett 13.5 Barnett 8.5 Shelton 4.9 Cameron 3.9 Hallett 3.0 Lyne 1.1 Groves 0.7)

Labor 26.2 (Butler 11.1 Mitchell 5.3 Farrell 2.8 Goss 2.5 Batt 1.6 Campbell 1.5 O'Donnell 1.4)

Greens 16.3

Nationals 6.1 (Tucker 3.8 Jenner 1.1 other 0.9)

Shooters Fishers and Farmers 4.1

IND Offord (named) 6.2

Any other 5.6

Note that there is a slight difference between these figures and the ones previously published - there's a minor inconsistency between two sections of the report but I believe the above are correct.  (Edit: yes, the previous headline breakdowns for Lyons had an error).

In this Lyons sample the Liberals have 2.84 Labor 2.10 Greens 1.30 Nats 0.49 Offord 0.50 (supposedly) others 0.45.  This is another example of the issue with naming some INDs and not others since Offord polled 0.3% with her own column in 2024, and would be somewhat more prominent now from having run in the federal election but is not likely to get anything like 6.2%, especially not if the Greens are getting quota.  What I think would actually happen with this sample is that John Tucker would get a pretty strong flow of preferences from the Shooters, Fishers and Farmers and win the final seat.  Neither major party spreads their vote evenly enough to stop him.  Here we see more potential within-party carnage with Andrew Jenner likely to lose to Tucker if the Nationals get a seat at all, and Casey Farrell at risk against Brian Mitchell.

Estimate for this sample 3-2-1-0-1 (Tucker)

Other comments

The Preferred Premier score is 40.7=31.1 to Rockliff.  Goven the skew to incumbents such questions contain this isn't a bad result for Dean Winter but it's another reading that suggests that Labor haven't put a serious dent in Rockliff by bringing down his government.  If the Liberals were heading for a thrashing I would expect to see even a relatively new Opposition Leader closer than this.  

The poll has a surprisingly large gender gap.  The Liberal vote is 8.2 points higher with male voters and the Labor vote is 8.6 points higher with female voters.  The poll has age/gender breakdowns which suggest this difference is especially strong among young voters.  

The poll also shows that the non-majors vote is highest at both ends of the age spectrum, with voters more likely to support independents with age.  The Liberal vote falls with rising education and the Greens and independent votes rise (save that there is a high Greens vote among "Completed year 12", presumably because this catches students and workers who haven't yet collected other qualifications.)

I am not sure how many other polls we may get - this looks like a less richly polled election than the previous -  though I do know YouGov have been in the field for what I presume to be a public poll that will be an interesting sequel to their previous.  To this stage the previous YouGov is the only poll to have Labor ahead since the election was called, but its average data age is nearly a month now.  

A reminder that I'll be live blogging from the tally room for Pulse Media on election night and the coverage will not be paywalled!  A link will be posted on this site on the night.