Queensland Polling Narrows Further In The Final Days

LAST-ELECTION PREFS AGGREGATE: 52.4-47.6 TO LNP 

SEAT PROJECTION OFF STATE POLLS IF THIS 2PP IS ACCURATE: LNP 48 ALP 37 GRN 4 KAP 3 IND 1

(AVERAGE OF FINAL POLLS BY RELEASED 2PP: 53.5-46.5 TO LNP, PROJECTION FOR THAT 2PP IS 51-34-4-3-1).  

As the final polls come out, we seem on for a closer Queensland election than earlier this year looked the case.  For much of the year the Miles Government has had classic hallmarks of a doomed state government - almost ten years in power, federally dragged, beset by crime complaints, and polling terribly.  Even four weeks ago there were signs of some recovery, but nothing that looked like life.  Now in the final week the LNP has recorded a couple of polls based off which it would be only slightly more likely than not to get a majority.  As the pendulum slightly favours Labor, it's even still plausible that if there is a modest polling error, Labor could scrape home.  Equally it's still plausible that the LNP could outdo the polls or get a good seat distribution and get a very solid majority.  But the very heavy drubbing that for so long looked so likely now seems a much more remote prospect.  If the late polls are spot on, Labor will almost certainly still lose, but they won't have trouble with saving the furniture.  Not that they needed the furniture the last time they were voted out.  

This has been accompanied by some remarkable changes in leadership ratings.  In the final Newspoll, Steven Miles has recorded a Better Premier lead, albeit of 3%, which is typically nowhere near enough because preferred leader polling skews to incumbents.  But such as it was, that was his first Better Premier lead ever, and the first for a Labor Premier since April 2023, snapping a run of 17 losses from various pollsters.  Crisafulli has gone from a personal rating of net +12 at the start of this campaign to net -3, his first negative rating of the term that I can find after at least 19 positives.  This sort of recovery by a state government that has started losing heavily in polling is very uncommon.

The three most recent polls (and I am not sure there will be any more) are:

* YouGov (10-16 Oct) 54.5-45.5 to LNP off primaries of ALP 31 LNP 41 Green 11 ON 11 KAP 1 others 5

* Resolve (14-19 Oct) 53-47 to LNP off primaries of ALP 32 LNP 40 Green 11 ON 9 KAP 2 IND 2 others 4

* Newspoll (18-24 Oct) 52.5-47.5 to LNP off primaries of ALP 33 LNP 42 Green 11 ON 8 others 6

(Resolve's 2PP is by respondent preferences.  The article also gave 52-48 by last-election preferences.  Where a pollster releases both and has not clearly stated one as the preferred measure, I treat whichever is published first in the article as the headline.)

Using my usual last-election aggregation methods but without any weighting for poll quality Queensland currently comes out at about 53.0-47.0.  (Australian Election Forecasts has the same.) In my seat model that's an LNP majority on average, but only just, with the median result being a 48-38 lead to the LNP in the 86 notionally two-party seats that the model includes, though this includes a couple of seats Labor has long been expected to drop to the Greens.  (The Newspoll comes out at 47-39).  AEF has the LNP one seat better than my model.  

There might be some upside for the LNP against my model's assumptions: for instance, they might recover at least one seat from KAP, and there are not that many seats they could realistically gain on 2PP but lose to a non-major candidate.  There might also be an improvement in preference flows for them from One Nation.  Regional differences don't seem to especially help the LNP on my reading though, because there are not so many regional seats in the 7-12% range, so a strong swing outside Brisbane plus a weak swing inside leaves the LNP struggling to take enough regionals to beat my model.  The worst scenario for the LNP is one in which the regional swing is not only strong but also uneven so that they miss out on some of the regional targets around 6%.  

I should also mention the Courier-Mail "exit poll" with an LNP lead around 58-42 across ten selected seats if only to point out that it wasn't a real poll, it was journalists untrained in polling techniques asking people at prepoll who they had voted for.  With the total demise of professional exit polling in Australia in the last few years there have been a lot more of these newspaper exit polls.  Usually, I find that aggregating enough of them and adjusting for the fact that they are prepoll makes for a reasonable broad idea of what will actually happen.  However, the problems with unscientific newspaper "exit polling" go beyond just the differences between prepoll and on the day votes.  These votes were taken very early in the prepoll period and it's probable that early prepolls break more to the right than late ones.  The time of day might also be significant.  

I have very little time today, not enough for a "why is it so" in terms of why the LNP's lead has narrowed, and want to get this article out this morning so I will just quickly note the crossbench seats of interest without going into much detail.  The Greens are defending Maiwar and South Brisbane.  A strong LNP vote could threaten them in Maiwar though only to the extent that the LNP either take votes directly from them or convert voters who previously voted ALP then Greens.  In South Brisbane Labor's improved polling raises the prospects that Labor stays in second rather than being pushed into third as has looked likely for much of the campaign.  If Labor does stay second they can pick up LNP how to vote card preferences (which favoured the Greens in 2020) posing an in-theory risk.  However if the Greens pick up the usual outsized first term personal vote for their MPs it's not likely to matter.  The remaining Greens targets are seats where they hope to push Labor into third and win on their preferences - by uniform swing this would still be likely in McConnel and Cooper but if Labor's vote is holding up unusually well in the inner city then this might not occur.  The other targets are Greenslopes and Miller but the state polling picture suggests that these are harder.

The three seats KAP retained by large margins in 2020 are hard to get a read on because some polls include KAP and some don't, and a statewide poll that returns a figure of 1%, 2% or 3% might not be that accurate when covering a party with intense support in a particular region - yet 1% or 2% could be all the difference to the party's fortunes.  KAP have picked up Mirani by defection from One Nation, but a problem for Stephen Andrew there is that his former party One Nation have recommended its voters put the LNP ahead of him, and plenty of One Nation voters who would have voted for him previously might do that of their own accord. On the plus side however, Labor have put him ahead of the LNP on their cards, so if Andrew makes the final two he should win.  If Andrew is to hold he will most likely need to either retain nearly all of the voters who voted for him as a One Nation MP or else have One Nation voters just ignore their party's card and generally prefer him to the LNP.  (Most of them do but even the 10-15% who usually follow it could harm his chances of making the top two.)

There hasn't been any suggestion I've seen of a threat to Sandy Bolton in Noosa, and one seat where an independent gain is possible is Rockhampton.  In 2017 Margaret Strelow beat Labor on two-candidate preferred but was knocked out in third place because LNP preferences put One Nation ahead of her.  One Nation aren't likely to amount to as much this time around and the Courier-Mail's tiny sample had no trouble finding Strelow voters even in prepoll (which is often bad for indies) so this is one to keep an eye on.

One Nation have a habit of spreading their vote too widely to turn it into many seats and in 2017 got 13.7% but won only Mirani.  Their statewide vote total in current polling suggests even a single seat could be difficult.  At this election Keppel (James Ashby) is the main one worth watching, together with how much of their vote they hold onto in Mirani after dumping their MP there.  

Tune in here for Queensland election night coverage from 6 pm Qld time Saturday and then postcount threads through the coming days.  What surprises will Queensland election night have in store for us this time?  

UPDATE (Friday Morning - edited Saturday): There is a late-breaking false claim by the Courier-Mail to have seen internal Labor polling suggesting that KAP could jump into second at Labor's expense in Mundingburra and Thuringowa and win the seats (see link in comments).  As always internal polling rumours should be treated with a lot of caution; even Robbie Katter seems to be struggling to believe it.  These are both seats that KAP has sometimes threatened to win in the past but nothing has been seen this election to suggest that KAP are surging!  As usual with such stories, no details of pollster, date or sample size, or even whether it is actually a poll and not an MRP model output.  (The same polling is supposed to have the LNP easily winning Mirani and Bolton holding Noosa, albeit narrowly from behind on primaries.)

I understand that this is not in fact Labor polling or even a poll at all, it is a union-commissioned DemosAU MRP model output (or similar), and the Courier-Mail's reporting was wrong and most likely careless.  The Australian has separately noted a claim that LNP "internal track polling" has a similar outcome in Mundingburra.  

(Friday 6 pm): The ABC is "reporting" - a feeble effort on the detail front at this stage - a uComms "robopoll" with a sample size of over 3000 taken last night and said to have a 51-49 to LNP 2PP, presumably by respondent preferences.  But there are at this stage no details of who commissioned it, how it was obtained, or what form of robopolling it was, let alone the primary votes.  uComms is an erratic poll and its previous Queensland poll in February while possibly capturing a brief honeymoon effect for Miles was followed by a far worse Newspoll one month later.  It is not however my experience that uComms consistently skews to Labor, more that it has a high average error.  UPDATE: The poll was not commissioned by anyone and the primaries - received by me before polls opened for polling day - are ALP 33.6 LNP 39.3 Green 12.9 ON 7.8 KAP 2.9 "Other minor party" 3.5.  Also of interest the poll has a large gender gap, which I estimate on 2PP as something like 57-43 to LNP for male voters and 55-45 to ALP for female.  uComms use respondent preferences; by last-election preferences this would be closer to 50-50.  The poll is an SMS poll of mobile numbers.    

Friday 7:30 Seat Betting Update: Seat betting is not necessarily predictive but fun to keep an eye on.  An incomplete market of seats has Labor at an implied chance below 50% in 21 of its own seats where the LNP is favourite,  2 where the Greens are favourite, and one where there is no party below $2; if all favourites won Labor would only win 29.  However there is an asymmetry in the number of close seats (which I define as two or more parties $3 or less, or favourite $2 or more) with 12 projected Labor losses being close while only four projected Labor holds are close.  On that basis 31 Labor holds would be a better read of what is expected.  On the pendulum Labor is favourite to lose everywhere below 11% except Cook and Macalister, and also to lose McConnel.  Given the state polling, something doesn't look right here!  The headline odds have Labor at an implied 13% chance to win, though the market probably expects less than that given longshot bias.  

Saturday: The Abortion Issue: I've added comments about this here because a live thread wasn't the place.    Although David Crisafulli ruled out any changes to abortion law in the term, Katter's Australian Party leader Robbie Katter announced an intention to move to repeal Labor's decriminalisation.  Crisafulli has been in a twist ever since because if abortion is a conscience vote, he can't control how his MPs will vote on it, and if they win by lots a repeal might then be possible whatever he says.  Helpful individuals like Freya Ostapovitch and Jacinta Price have continued to give the issue oxygen every time it looks like quietening down.  While it's not top of the issue list for most voters, 15% of female voters named it as one of their top two issues in Newspoll.  If the LNP really wanted to kill the issue off they could have declared that any private members bills on abortion in this term would not be brought on and that there would not be a conscience vote on that as it was an election commitment.  That would have avoided the situation of MPs having to vote against their "conscience", which usually really means their religion, directly on the issue of abortion itself.