Saturday, April 27, 2024

The Tide Is Going Out For Queensland Labor

...and when the tide goes out in Queensland, they say that it goes out a long way ...

Yesterday's YouGov poll finding the Miles government trailing 44-56 led to a minor outbreak of poll denialism on social media (I've so far seen versions of A4, C4, C6, C8 and C9), but Steven Miles himself was not denying the polling at all, commendably admitting that it looked "most likely" that his government would lose in October.  (Just whatever you do, Premier, don't actually concede before election day!) I haven't covered Queensland polling since I gave the Courier Mail a big roasting for some really bad poll reporting in December 2022 and a return to Queensland polls is overdue.  It happens this time that the poll is so bad for Labor that even the Courier Mail can't spin it as much worse than it is.

It's worth noting that Queensland Labor during its nine years in power has often polled indifferently.  In the 2015-7 term it trailed on 2PP in a third of the published polls, but never worse than 48-52.  In the 2017-20 term there was less polling and there had been a few shabby looking numbers (again no worse than 48-52) before the COVID-19 pandemic arrived in 2020 and lifted all governing boats.  The Palaszczuk government ended up slightly outperforming its final polling, but it was a very sparsely polled election.  Going into the 2024 contest that is now just six months away, it looks like we might see a higher volume.


In this term there was nothing to see polling-wise in the first year and a half.  Sparse polling through 2022 and early 2023 generally had the 2PP about even or slightly ahead (and the ahead came from Resolve Strategic, which was very favourable for Labor federally and in other states between the 2022 federal election and the end of 2023, though that trend has now weakened, at least federally).  

In early 2023 there was a YouGov poll with Labor behind 49-51 followed by a Resolve Strategic poll that suggested a lead around 52.5-47.5.  But the Resolve poll was an aggregate taken over three months while the YouGov poll was a week of data, meaning the latter had a lower average data age.  On that basis, the most recent eleven polls have now had the LNP ahead on two-party preferred, except for one uComms in February this year that still had a tie.  (See Wikipedia list of polling).  

I find uComms to be an erratic poll but one that does not have any consistent skew (it was, for instance, excellent in Dunkley but has been over 6% out in opposite directions at the last two Tasmanian elections).  But I suspect the uComms polls at the start of the year showing a close 2PP contest were broadly correct at the time taken.  There is a similarity with what happened in New Zealand, where there was a minor sugar hit for NZ Labour on the replacement of Jacinda Ardern with Chris Hipkins, but Labour was soon plunging to much worse numbers under Hipkins than under Ardern.  

Pendulum Favours Labor

While Labor has trailed in nearly every poll in the last 13 months, that doesn't mean it was necessarily in a losing position through that time.  The reason is that Labor has a large on-paper pendulum advantage in Queensland at the moment, as it does in Victoria and federally.  The 2020 election resulted in a pendulum in which, with uniform swings, Labor would retain a majority with a 2PP of 50.1%, but the LNP would need 52.5% for 47 seats, all else being equal.  Further favouring Labor on paper if the swing landed around the 3-6% range are personal vote sophomore effects for new MPs in marginals Bundaberg, Caloundra, Hervey Bay, Nicklin, Pumicestone and Mundingburra.  In Bundaberg and Nicklin the effect is boosted by defeating a recontesting LNP incumbent in 2020.  (As against this the retirements of Yvette D'Ath and Barry O'Rourke in Redcliffe and Rockhampton could hurt Labor if the swing is towards the upper end of that range.)

So while Labor has trailed in many polls in the last year, most of them have not pointed to outright losses.  A 48-52 2PP loss, for instance, would probably be an LNP win of some kind, but it could be a minority government, and it is even possible that Labor could scrape back in some fashion on such numbers.  The pendulum advantage should be treated with some caution because it's not unusual for Queensland to see regionally uneven swings, and the 2020 election had some weird aspects such as the COVID-induced swings to Labor in coastal retiree seats.

What we are starting to see now though is polls that are well into the LNP majority zone.  The first one was a Redbridge 55-45 to LNP last September, which was a bit of an outlier at the time.  It now has friends with a 54-46 Newspoll and the 56-44 YouGov as the two most recent polls.  

I should note that this YouGov is a different beast from the one active earlier in the cycle - it's the "post split" YouGov since the Pyxis split, and it has tried some different things, including the experimental use of Voice referendum vote as a weighting.  It's one poll and findings like the Greens on 15 always raise an eyebrow.  Outside Tasmania it is very rare for the Greens to record such swings at state elections. The party did get nearly 13% in Queensland federally and did get a substantial swing in Brisbane City Council, but still ...

What might this poll look like?

I'll be doing a probability-based model for Queensland later in the term but this is an early idea of things to look for if we get a big two-party swing combined with a high Green vote.

Labor won 52 seats at the last election to the LNP's 34, KAP won three, the Greens two, One Nation and independent Sandy Bolton one each.  The YouGov poll has primary votes of 44 for the LNP (+8.1%), 27 for Labor (-12.6%), Greens 15 (+5.5%), One Nation 10 (+2.9%), KAP 1 (-1.5%) and others 3%.  

On the two-party swing of 9.1% (NB my 2020 estimate is 53.1) about 21 classic seats would fall by uniform swing.  Ipswich West is above that uniform swing line from 2020, but has become what I call a "disrupted seat" by changing hands mid-term.  Disrupted seats tend to behave somewhere between their previous election and their by-election form, which for a 2PP swing this large suggests the LNP would probably retain it.  

The Greens would win Labor seats above the two-party swing line wherever they could knock Labor into third.  On a uniform swing this would give the Greens three extra seats, McConnel, Cooper and GreenslopesMiller would go close to falling too.  The LNP are hinting that they will not recommend preferences to the Greens in any seats, but on uniform swing they would not be excluded in any seats with all of these plus South Brisbane and Maiwar ending as Greens vs LNP contests.

Reading One Nation's single seat off the state picture didn't work at all last time but for what it's worth in Mirani, by uniform swing Labor would fall into third, in theory leaving One Nation and the LNP roughly equal on primaries.  However in 2020 One Nation outperformed the LNP on minor party preferences in Mirani.  In 2017 seven seats finished LNP vs ONP with an average flow at the Labor exclusion of 62-38 to LNP.  Based on this, Stephen Andrew would all else being equal be likely to lose the seat even with a primary vote swing to One Nation.  If KAP really polled only 1% they too would be likely to drop at least one seat, but that should be treated with caution as accurately representing KAP in a state poll is rather difficult!  We've seen that crossbench seats tend to be sticky against general trends so it will be interesting to see if the One Nation and KAP seats do hold anyway.  

All up a possible result off this poll would be LNP 58 seats, Labor 27, Greens 5, KAP 2, IND 1.  The more prosaic recent Newspoll (54-46 to LNP off primaries of LNP 42 ALP 30 Green 13 ON 8 others 7) comes out at about LNP 54 seats Labor 31 Green 4 KAP 3 IND 1.  There's not much difference here.  

At this stage this isn't looking like 2012 - partly because the LNP primary is not nearly so high and the Greens primary is higher, but also because Queensland now has compulsory preferencing which makes for less lopsided seat and 2PP results in wave elections.  (I am not sure how much longer Queensland will have compulsory preferencing for should the LNP win the election.)  Nonetheless whether this 56-44 YouGov is necessarily accurate right now, it's probably about my median expected result at the moment.  The election could still be close-ish at least in seat terms, or seriously lopsided.  

Leaderships

The leadership numbers in recent polling are also unpleasant for Labor.  Newspoll had Steven Miles at a net -11 rating while YouGov had him at net -22.  Newspoll had Miles trailing 37-43 as Better Premier, YouGov 27-40.  The history of state polling, unlike federal, is that once Premiers get into negative double figure netsats or fall behind at all as Better Premier, they will generally (not always) lose or be removed before the election.  David Crisafulli meanwhile is continuing to poll positively, with +14 netsats in both polls, although Opposition Leader ratings do not mean a great deal.  

This Should Be Expected

Irrespective of polling, it is historically expected that Labor should lose this election, which is not to say it will definitely happen, just that it is most likely.  The two biggest factors affecting state election seat share changes are the age of the state government and whether the party in power federally are the same; Queensland Labor has both of those against it.  According to regression off past such cases, Labor would be expected to lose about 17 seats at this election.  Federally dragged state governments of around the same age in recent decades (SA 1993 and WA 1993, Qld 2009, Vic and Tas 2010) have all gone backwards in seat share by between 9% and 28%.  The best result of these was Anna Bligh in Queensland 2009 losing only 8/89 seats, but that was while the Rudd federal government was still very popular and, perhaps more importantly, headed by a Queensland Prime Minister. 

Labor can only remind voters about Campbell Newman for so long - that was nine years ago and anger fades, and Newman's turbulent reign is actually an opportunity for the LNP to sell their guy by presenting him as different, which isn't hard.    This is not to say that a win for the LNP is inevitable - this party has displayed some talent when it comes to losing elections in the past - but that at this stage it appears very likely.  It's easy to make a flippant list of things that might save Labor, such as another pandemic, World War 3, the LNP listens to any of Newman's campaigning advice, or Peter Dutton promises to build nukes in <insert criticial seat list here>.  It's also easy to question whether even those would do it.

Resolve update added 23 May: Just adding a quick note that the recent Resolve sample (aggregate over three months) is very similar to YouGov's numbers (I get it at 55.5-44.5 in the LNP's favour by last-election preferences) and doesn't change the picture in this article at all.  

No comments:

Post a Comment

The comment system is unreliable. If you cannot submit comments you can email me a comment (via email link in profile) - email must be entitled: Comment for publication, followed by the name of the article you wish to comment on. Comments are accepted in full or not at all. Comments will be published under the name the email is sent from unless an alias is clearly requested and stated. If you submit a comment which is not accepted within a few days you can also email me and I will check if it has been received.