Queensland 2024: The Polls Aren't Getting Much Better For Labor

Recent polling LNP leads approx 55.5-44.5

Possible seat result of this 2PP LNP 56 ALP 29 Green 4 KAP 3 IND 1 

I last wrote about the Queensland polling leadup back five months ago - was it really that long? - in The Tide Is Going Out For Queensland Labor.  At that time, there had been a few polls out showing Labor trailing about 45-55 two-party preferred, which as I explained in the article is historically not surprising in the slightest.  Five months on and less than one to go til the election, they're still there.  

However it's not as if nothing at all has happened in the meantime.  Since my last article (which mentioned the 44-56 April YouGov and the 46-54 March Newspoll), things may have got worse and then got better for the Government.  There's no need for me to repeat all the details of polls that are recorded and linked to on Wikipedia but there was a string of shockers for the government through to early September.  On 2PP they had only 44.5% (est) in Resolve February to May, 43% (converted estimate) in Redbridge February+May (two waves, not a continuous sample), 43% in YouGov July 8-15, 45.5% in Redbridge May+August, 43% in Wolf + Smith 6-29 Aug (Wolf + Smith is a sort of Resolve spinoff), and 42% (est and possibly generous) in Resolve July through September.  

While the Resolve type polls in this mix have the Labor primary lower than others because of their handling of the independent vote, none of these six polls had the Labor primary with a 3 in front of it, and Redbridge's first sample had the LNP as high as 47.  The average major party primary gap across these polls was 17.5 points.

The latest Newspoll (12-18 Sep) and Freshwater (26-29 Sep) in contrast, aren't that bad.  They're also very nearly identical.  Newspoll has 45-55 off primaries of 30-42 to the majors, Greens 12, One Nation 8 and others 8.  Freshwater has 44-56 off 30-43, 12, 8 and 7.  So both polls have a 3 in front of the Labor primary, albeit barely, and the major party gap is down to 12 or 13.  It isn't clear what preferencing method Freshwater is using but an implied 51.9% of all preferences seems stingy if it is last-election preferences (Labor got 55.3% with a lower Green vote in 2020), so it may be respondent, or Labor might have done badly on rounding.  

Leadership polls continue to bear the hallmarks of a decisive handover.  David Crisafulli has beaten Steven Miles as preferred Premier nine times out of nine, by 7 and 8 points in Newspoll and Freshwater, despite these polls usually favouring incumbents.  He also beat Annastacia Palaszczuk six times in the second half of 2023.  Miles' net satisfaction/approval ratings are by no means terrible, usually in the negative teens, while Crisafulli scores in the net positive teens - since April they're averaging about net -14 and net +15 respectively.

No public seat polling and not much by way of reliable regional breakdown polling has been seen.  There was some attention for a Redbridge finding of a 14% snap back to Labor in inner city seats, attributable to 50c bus fares and the like, but the subsample size would have been tiny and the effect too large for me to take at anything like face value.  That said, the 50c bus fare policy does seem to have proved more popular than cynics were accepting, with the LNP forced to (to some degree at least) match it.

Some internal polling rumours have been seen, including one about an expensive internal poll of Shannon Fentiman's very safe (or is it) seat of Waterford (ALP 16%).  This is supposed to show a 13% 2PP swing to the LNP.  The fate of Fentiman is of interest as a potential leader to pick up the pieces from some distance from the current regime, though she may not have earned too many team player points by dumping on Palaszczuk's anointment of Miles as successor.

If anyone predicts a Labor victory in this election and is correct, I've seen odds as long as $30 on that from some bookies; it's been a while since a change of government anywhere in the country was seen as so locked in this far out.  (It's currently as of Oct 1 come in to about $11).  

Seat model

There have been some suggestions Labor might get belted to as low as 18-20 seats here, one of them in a puzzling AFR analysis of the Freshwater poll that referred to "The survey’s two-party-preferred swing of up to 12 points, which is unlikely to be replicated across the board" when the swing was actually only just over 9 points.  Perhaps Labor will lose that badly but it is not what the latest polls are implying on a state basis.  Here's my usual conditional seat probability 2PP model for a Labor 2PP of 44.5%.  I have just shown the Labor seats side since on a 2PP in the mid-50s, the LNP is unlikely to lose any seat to Labor that it won in 2020.  In the "proj" column the table shows the projected ALP margin (negative = loss) after considering current margin and personal vote effects, and the "prob" column shows what the model thinks is the chance Labor retains the seat if the 2PP is 44.5.  Note that I ignore by-elections where the incumbent retains, aside from making personal vote adjustments.  Where there is a change of ownership in a by-election (a "disrupted seat") I average the last-election and by-election margins for the new baseline.  (Ipswich West is a rare case of a disrupted seat where the by-election winner isn't recontesting, which might make things a little better for Labor than I have it here.)

In total, the model has Labor on for about 31 2PP wins in seats it currently holds off this 2PP, a net loss of 21 seats.  It could plausibly lose a few more to the Greens (see below).   The swing could always and almost certainly will break unevenly - for instance a bigger swing in the outer suburbs and regions is generally expected than in the inner cities - so some of the individual probabilities above will be nonsense.  But should that happen a number of wobbly Labor seats in the inner city will be saved, some regional or outer suburban seats will fall well above the swing line, but some of the LNP margin could well be wasted overkilling Labor's many non-Brisbane seats on 7% and below (especially the coastal retirement belt wins from 2020).  All else being equal it shouldn't make a lot of difference to the 2PP-to-seats conversion.  So for some of the scenarios doing the rounds to fly, I think LNP needs a bigger gun.  60-40, for instance, would be very likely to put Labor below 20 seats.   It is harder for the LNP to match the scale of the 2012 "Tarago election" where Labor were reduced to seven seats because compulsory preferencing means Labor will get a better 2PP off a given set of very bad primaries at this election than otherwise, even if the preference flow to them is weak.  Exhausting votes in optional preferential voting tend to magnify the primary vote leader's lead on 2PP.  

(A note re Murrumba - I have given Steven Miles a leader bonus, but I suspect his hold is shakier than the odds above imply, given outer suburban issues and the fact that he has been Premier for only a short time.)

ALP vs Green - And Associated Nonsense

Taking the Newspoll and Freshwater polls at their word, there is currently a 9.6% primary vote swing against Labor, 2.5% to the Greens and about 6.6 to the LNP.  If that is anything near uniform then the Greens easily knock Labor into third in Cooper and McConnel and therefore should win both (though the LNP might not be completely disinterested in either).  The next two targets a long way up the tree are Greenslopes and Miller, but Labor beat the Greens at the 3CP stage by 17.7% and 20.8% in those respectively last time, so the Greens would need to do three or four points better than the state Labor-to-Green swing in these seats to pick them up.  There is some history of the Greens outdoing their state swings in heavily targeted inner city seats and so these two in particular are worth keeping an eye on.  

The Greens' usual game of talking up their chances before the election whether they are competitive in the seats mentioned or not (they've been doing it for decades) has been met with disinformation from some ALP supporters.  Retiring Attorney-General Yvette D'Ath made the ludicrous claim that the Greens' thought they might pick up a few seats meant they had a backdoor preference deal with the LNP.   In fact, in none of the Greens' target pickups would they be relying on LNP preferences to win; in all of them the method would be pushing Labor into third and winning on Labor's preferences.  On uniform swing this would even happen in South Brisbane, which the Greens hold, rendering LNP preferences irrelevant to the Greens' prospects.  

D'Ath threw in the completely baseless claim that "I’m not quite sure why the Greens are celebrating the fact that a vote for them will put David Crisafulli into government,", which had overtones of the sort of a vote-for-X-is-a-vote-for-Y disinformation that Attorneys-General everywhere should be banning not promoting.  To the extent that there was any attempt at an argument behind it it was supposedly that Labor would need these particular seats to form the government Premier Miles has already by political standards conceded that they won't be forming anyway.  In any case no evidence was presented that the Greens would back the LNP in any hung parliament scenario even if it did somehow occur. 

Weeks later this nonsense resurfaced on Twitter in the form of several gossipping pro-Labor accounts who had "heard" (from unstated sources with zero evidence) that the Greens had done a preference deal with the LNP - as if it would be worth the embarrassment factor to the LNP of accepting said "deal" or as if any significant number of Greens voters would follow a card directing them to the LNP anyway.  

It is supposed to be the case that educated voters are realigning towards parties of the broad left.  It is a mystery to me, given that, that some ALP supporters continue to make these desperate and dishonest claims that seem calculated to repel any voter with a brain.

Others

There isn't a lot of useful intelligence about the likely fate of the now four Katters Australian Party MPs, after One Nation's sole MP Stephen Andrew was disendorsed and switched to KAP.  Four of the last five polls had KAP on 1% statewide with Wolf + Smith the sole holdout for them on 3%; uComms early in the year had 3.9% so I suspect the statewide polling for KAP just isn't all that useful. The disorder in Mirani looks like an opening for the LNP to get this one back in a "change" election.  They need very large swings to dislodge any of the other three, but Hinchinbrook will be worth keeping an eye on at least early on election night to see if there is any chance of that.  

There doesn't seem to be any buzz about independents at all with the exception of whether Sandy Bolton can retain Noosa, currently on a large margin.   [UPDATE: We should however pay attention to Margaret Strelow, IND for Rockhampton and long-time local Mayor, who nearly won in 2017 but was a victim of exclusion order with LNP prefs putting One Nation ahead of her and stopping her from beating Labor.]

Crisafulli Misleading Claims re OPV

False claims about the voting system are also coming (again) from Opposition Leader Crisafulli.  Asked about LNP preference recommendations between Labor and the Greens (the LNP are recommending preferences to Labor), Crisafulli said "Queenslanders shouldn't have to be forced to vote for anyone they don't want to.  It's undemocratic."

But compulsory preferential voting doesn't force anyone to vote for anyone they don't want to (it doesn't even force them to vote formally at all).  If a voter puts, say, Labor second last and the Greens last, they have hardly "voted for" those parties - they have voted against them, except that in the hypothetical case of those being the last two standing they have ranked one above the other.  And in 2020 that choice, as made by someone voting 1 LNP, became an active choice in just one out of 93 seats - this year it could be zero.  

I am personally sick of this snowflake argument for optional preferential.  Just get over it - if you dislike one party less than the other rank them accordingly, or if you don't care or want to make some kind of protest then you can randomise your preference.  The argument is self-indulgent and ignores the extent to which OPV can suppress real preferences that voters would have if asked, but that the voter will not exercise under OPV because of satisficing or misunderstanding the voting system.  I would be happy in principle if those who had an ideological objection to giving full preferences - and only those people - could declare their objection in advance and cast an exhausting section vote that would be counted - but such a system would carry privacy risks and the risk that other voters might think they could do it too. 

There is a real debate about the suitability of compulsory preferential as it currently exists, but that is about formality, not about voters who whinge about not being allowed to exhaust their ballot.

Crisafulli has also claimed that the reintroduction of compulsory preferences by Labor was "an electoral gerrymander that they have rigged", thereby comparing a choice of a different and widely used election system - albeit motivated by a belief it would bring an advantage - to gerrymandering, malapportionment and rigging.  (The term "gerrymander" is widely used in Queensland for the malapportionment Sir Joh Bjelke-Petersen thrived on, though he did not start it.)  And he has said " We know the Greens and Labor are one, and the Labor Party know that there is a real prospect of a larger contingent of Greens giving them preferences and giving them supply and enabling a fourth term in 14 years"  This makes no sense whatsoever since any elected Green, by definition, is not giving anyone else any preferences.  

Some other LNP operative has said "an LNP government would restore Queensland’s ‘just vote 1’ system and do away with Labor’s dealmaking system".  But OPV is only a "just vote 1" system if the voters and parties want it to be, and preference recommendation deals can occur under it just as they can under CPV.