This is a combined prospects/live post for this week's third consecutive by-election weekend, this one for the seat of Stafford in Queensland. Stafford is an old-fashioned vanilla two-party contest that has been overshadowed by Farrer but I think it's interesting enough to be worth covering in detail especially given some of the claims that have been made about it. A big question here is why are the LNP running for this seat. My initial impression was they might sit it out as governments often do, but in fact there is a strong strategic case on paper for giving it a go.
Stafford is an inner north Brisbane seat that was a Labor/Liberal swing seat in its first lifetime (1972-1992) but since its revival in 2001 it has been Labor-dominated, falling only in the 2012 wipeout with 2024 its only other return to technically marginal status. In its second life it has generally run about 9% above the Labor statewide 2PP with a low of 5.7% above in 2012. The seat has had an increasing Greens vote at recent elections with the party polling 18.06% in 2024 (their ninth best seat in the state). Conversely it is a dead zone for One Nation who polled 3.17% in 2024 (their eighth worst) and have decided not to risk a loss of momentum in this one. While One Nation are surging in Queensland state polling, it is not to the same extent as elsewhere, and they'd be lucky to get double digits if they ran in Stafford.
Stafford is vacant because of the death at only 44 of second-term incumbent Jimmy Sullivan, who was first elected as a Labor MP in 2020. Sullivan was evidently a troubled soul whose recent timeline included domestic violence allegations (abusive language), the loss of a child, post-traumatic stress, addiction and other health issues. He had been on a long period of leave from parliament from late 2024 and had been removed from the Labor caucus in May 2025 for allegedly failing to comply with a "safe return to work" plan. The exact cause of his death, described in some reports as a possible accident, is as yet not publicly known.
Stafford had a significant by-election in 2014 when its fall to Labor on a 19.1% 2PP swing presaged the sensational 2015 defeat of the one-term Campbell Newman LNP government. This has been used by Premier Crisafulli to goad Labor saying that if they can't get a double digit swing here then what are they doing. But the two are completely non-comparable. Stafford in 2014 was a government vacancy where swings are generally larger, and it was also a vacancy for a government that had scored a 2PP nine points higher than Crisafulli's at the previous election.
So what should we expect? The general pattern is that by-elections see swings against the government, but this is muted when they are opposition vacancy by-elections, which almost as often as not see small swings to government. This is partly because the opposition has lost the personal vote of its incumbent, and also because governments tend to cherrypick which such by-elections they bother contesting in the first place. There actually aren't that many recent Queensland cases (and a lot of them are unusual - pandemic, retiring ex-Premiers, local independents and so on) but nationally the average is about a 1% swing to the opposition in such cases, with a high standard deviation. There isn't generally any difference between by-elections caused by resignations and those caused by incumbent deaths.
The LNP has already made one by-election gain in this term, picking up Hinchinbrook with a huge swing after popular KAP MP Nick Dametto decided he would rather be the mayor of Townsville and didn't even clearly endorse his old party on the way out the door (though apparently he did vote for them). That was a special case but for what it's worth government gains in by-elections do most often occur when the state government is in its first term and the opposition is "federally dragged" by being the same party as governs in Canberra - both the case here.
The LNP government is also polling well (on sparse evidence) on average at the moment after a few iffy Resolve readings last Spring. It led 56-44 2PP in a Redbridge poll late last year and a DemosAU in February. Two small Resolve samples in Jan-Feb and March-April I estimate at 54.6 and 52.4 to LNP respectively.
Beyond the debates about issues like petrol costs and health services, the by-election background is unusually noisy with distractions at the moment including debate about an on-off-on affair between Ministers Tim Mander and Amanda Camm, Opposition Leader Steven Miles apologising after being found in contempt of Parliament for misleading the House and former Premier Palaszczuk's partner being charged with rape and deprivation of liberty, all of which he denies. And then of course there's the federal Budget.
Other points of interest
The Greens have decided to issue an open how to vote card at this by-election. I've been unable to find any clear public explanation (as distinct from apparent retro-justifications) for this decision. If it had been a head office call as an experiment or to send Labor a warning that might make some sort of sense but it seems to have been a local decision. I do wonder if it is inspired by the recent idea that the Greens need to be becoming more populist, like a One Nation of the left, instead of being so "establishment". It has sparked criticism including some from within.
Decades ago when the Greens used to issue open cards more commonly the impact on flow was low (5-10%). These days in this sort of educated seat the preference of Green voters for Labor is so strong that this may not make any real difference at all - though even a slight flow change could hurt Labor's 2PP by several tenths of a point because of the size of the Greens primary. In the federal seats of Deakin and Menzies in 2025 the Greens issued open HTVs on the main voting day only; the flow from the Greens to Labor nonetheless increased albeit by slightly less than the national increase - on that basis the impact of the decision on flows in the booths was at most a few percent, and could well have been nothing. In Clark 2022 (a special case by being in Tasmania where interest in how to vote cards is always low) an open HTV also had no impact on 2PP flows. My suspicion is that while about 15% of Greens voters will follow how to vote cards, a high proportion of them will preference Labor anyway if there is no card (as happens in Tasmanian Legislative Council elections where how to vote cards are banned). The 2024 count in Stafford saw 83.7% of Greens exclusion preferences (overwhelmingly 1 Green) flow to Labor. I will be surprised if that falls to even 75%.
The rest of the field consists of Family First, Legalise Cannabis, Animal Justice, a Libertarian and two candidates running as independents for currently unregistered parties. One of these, Damian Smart, is running for Gerard Rennick People First while Liam Parry is the candidate for the Queensland Socialists. The Socialist run is interesting firstly because it is the unofficial debut for this party in the sort of seat where they might poll significantly, and secondly because Parry is one of two people who have been charged under Queensland's laws restricting the use of certain pro-Palestinian slogans. Although the ban is supposed to only apply when the slogans are used in a manner that could be reasonably expected to cause someone to feel menaced, harassed or offended, police have been taking action against the slogans even where it is not remotely clear that that condition is met.
The Sunday Mail's "poll"
The Murdoch tabloids in Australia and particularly in Queensland have a long history of conducting "exit polls" mostly at prepoll stations and then interpreting them cluelessly and unprofessionally without expert analysis, as a result deceiving their readers about what the "polls" are actually saying in the interests of blatant hype. (Sometimes when interpreted correctly these "polls" are actually useful.) I pointed out what a debacle this was in the 2024 state election leadup but no amount of making these points appears to stop these papers from embarrassing themselves with this dishonest and sloppy poll reporting.
This has continued with the latest effort in which off the back of a doubtless uncontrolled sample of 300 voters at prepoll on May 7-8 we are told that the LNP is "on track to pull of a historic upset" and that Labor is "relying on unpredictable Greens preference flows to cling on to the seat by its fingernails."
The actual numbers in this sample have the LNP on 41.7%, Labor on 36.7% and the Greens on 12.7%. No breakdown is given for the remaining 8.9%.
The first thing to note here is that the remaining 8.9% are likely to skew left, perhaps heavily so depending on whether Parry is taking any substantial number of votes from the Greens. But even with a 55% flow from the others and a 75% flow from the Greens, Labor would still just win on these numbers, 51.1-48.9. The second problem is that comparing these numbers with a general election is not comparing like with like - they are prepolls. In 2024 within-electorate prepolls in Stafford were LNP 40.52% Labor 36.9% Greens 18.4% - very close to identical to the Mail's sample except for its lower Greens vote. (The prepolls were about 1.5% 2PP worse for Labor than the seat in general.) And furthermore the sample is not just prepoll, it is prepoll early in the prepoll period, which is likely to be more conservative than average. So what this sample if accurate is actually pointing to is a Labor retain, perhaps with a small swing against.
From the LNP's perspective when deciding to contest Stafford, they would probably detect some chance (even if only a small one) that they actually win the seat. A win for them in a seat that was ALP+7 to the state average at the last election would be a disaster for Labor, of the sort that causes leadership change. Even a very close result would be a very bad night for the ALP. But that aside there is a decent chance on paper that the LNP would get a 2PP swing of some kind (which would lead to significant bragging rights). A small swing against the LNP is easily dealt with by misleading nonsense about the average by-election swing being several points and even more misleading nonsense the swing being nothing like 2014. So it's not really easy for the LNP to clearly embarrass themselves in this one. And while a big swing (say 6% and up) against them would raise some questions, they would also learn things in the process.
Given the LNP's strong state polling but also that the LNP is coming off a high base election I would consider any 2PP swing to Labor above 3% to be a good result for Labor. Any swing to the LNP at all would be a good result for the LNP and a 0-3% swing to Labor is the inconclusive zone. And Saturday is not far away so soon enough we will see how they go.
