The Australian Electoral Commission has published its post-redistribution margins for New South Wales, Victoria and Western Australia, holding off for the time being on the Northern Territory as the redistribution there has not been finalised. This involves two sets of numbers: two-party preferred, which boil the issue down to Labor-versus-Coalition without regard to whether Greens or independents may have been in the mix, and two-candidate preferred, which tackles the sometimes knotty issue of estimating new margins between the parties and independents who actually made the final count at the last election.
The AEC’s report repeatedly observes that its numbers “may differ to the calculations of people external to the AEC”, by which they principally mean Antony Green but also me and Ben Raue at The Tally Room. Links to an extensive accounting of my own estimated margins were provided in an earlier post, which also offered a broad overview of the principles involved in making the calculations. I’m pleased to say my two-party margins are similar to the AEC’s: within 0.2% in 80 seats out of 100, and out by more than 0.5% only in the cases of Hume and Hasluck.
Now more than ever though, two-candidate preferred is a vexed question particularly where independents are involved, as they will not have been on the ballot paper in the parts of the electorate that have been added in the redistribution. Antony, Ben and I are all free to exercise common sense in treating the teals as a collective unit, which at least solves the problem in the cases of Warringah and Mackellar. Not only does the AEC feel it does not have the liberty to make such judgements, but Ben Raue also relates that its system is not designed to combine Labor-versus-Greens results from different electorates, which can readily be used to calculate fresh margins for Wills and Cooper — though not for Melbourne, which absorbs territory from Higgins and Macnamara, both of which had Labor-versus-Liberal two-candidate counts.
Ben Raue identifies the following electorates as ones in which mismatched two-candidate preferred counts must be combined from different parts of the electorate as redrawn by the redistributions (not counting those where the problem can be solved by falling back on two-party preferred, as can always be done where the seat is a “classic” Labor-versus-Coalition contest). These are Bradfield, Fowler, Grayndler, Mackellar, Sydney, Warringah and Wentworth in New South Wales, and Cooper, Goldstein, Kooyong, Melbourne, Nicholls, Wannon and Wills in Victoria. For reasons just explained, people external to the AEC are painlessly able to finesse the issue in Mackellar, Warringah, Bradfield, Cooper and Wills, which undoubtedly makes the non-AEC calculations more instructive in these cases. That leaves nine seats where varying degrees of creativity are required. In Labor-versus-Greens contests, this is a simple matter of estimating preference flows. But estimating support levels for independents in areas where they didn’t run last time is a very considerable challenge.
The differences in the various approaches taken are outlined at length in Antony Green’s post on the subject:
I base my estimates on a comparison of of House and Senate votes. Ben Raue uses an estimate based on the difference between two-party and two-candidate preferred results. William Bowe has not tried to adjust primary votes but rather allocates zero votes to the Independent and applies preference flows on accumulated primaries.
The chief virtue of my own method is the elegance involved in not requiring any data external to how people actually voted for the lower house of 2022, but it comes at the very substantial cost of crediting independents with very small vote shares in the newly added parts of their seats. However, the AEC’s approach is in this respect worse, as it apparently credits the independents with no votes in these areas at all (though I don’t see how that can be the case in Kooyong, where my own estimate for Monique Ryan is lower than the AEC’s). Antony Green’s and Ben Raue’s methods are of more practical value in addressing the task at hand, which is estimating how big the swing will need to be for the seat to change hands. Whether or not this is happening can be determined on election night by comparing the booths that have reported their results with the equivalent results from the previous election.
A few other bits and pieces from the last week or so:
• A second Victorian state by-election looms to go with the one to be held on February 8 in Prahran after Tim Pallas announced his resignation as Treasurer and member for Werribee, which he held in 2022 on a margin of 10.5%. The Age reports the Labor preselection front-runner is John Lister, a local teacher and Country Fire Authority volunteer.
• DemosAU has a poll on the ban on social media use for under-16s, which finds 64% supportive and only 24% opposed, but 53% expecting the law will be ineffective compared with only 34% for effective. The poll was conducted December 5 to 16 from a sample of 809.
• Keith Pitt, who has held the Bundaberg region seat of Hinkler for the Nationals since 2013, has announced he will retire at the election, taking the opportunity to call for the party to abandon net zero emissions targets and support coal-fired power. There has been no indication that I can see of who might succeed him in Hinkler.
• The Nationals have preselected Alison Penfold, senior adviser to party leader David Littleproud and former chief executive of the Australian Livestock Exporters Council, to succeed the retiring David Gillespie in the Mid North Coast New South Wales seat of Lyne. Penfold won preselection ahead of Melinda Pavey, former state member for the upper house and the corresponding lower house seat of Oxley, and Forster accountant Terry Murphy.
• Left-aligned Ashvini Ambihaipahar, St Vincent de Paul Society regional director, Georges River councillor and unsuccessful candidate for Oatley at last year’s state election, has been confirmed by Labor’s national executive as candidate for the southern Sydney seat of Barton, to be vacated at the election with the retirement of Linda Burney. One of those overlooked, former state upper house member Shaoquett Moselmane, resigned from a party position in protest, Elizabeth Pike of the St George Shire Standard reporting rumours he may run as an independent.
• James O’Doherty of the Daily Telegraph reports that Warren Mundine will seek Liberal preselection for the northern Sydney seat of Bradfield, to be vacated at the election with the retirement of David Fletcher and contested again by teal independent Nicolette Boele, who came within 4.2% after preferences in 2022. Mundine is a former Labor national president turned conservative who ran unsuccessfully for the Liberals in Gilmore in 2022, and was the public face of the campaign against the Indigenous Voice together with Northern Territory Senator Jacinta Price.
• Slightly old news now, but it had hitherto escaped my notice that Michelle Ananda-Rajah, Labor member for the abolished Melbourne seat of Higgins, will be making do with third position on the party’s Victorian Senate ticket.
J J Hall,
Thanks for the concern. My wife actually has a doctor’s appointment tomorrow. I’ll go with her and ask if I need antibiotics.
Cheers
I agree with Player One and mj,
A generalised comment about people spending money being used to deny a cost of living crisis fails to take into account simple demographics – such as who are the rent gouging property investors and who are unemployed renters living in poverty.
nadia88 at 8.33 pm
Thank you for the details and your estimate. Note that Dr Bonham says the TPP vote and the pendulum remain ultimately decisive:
https://kevinbonham.blogspot.com/2022/06/two-party-swing-decided-this-election.html
The issue I have with those polls is the small samples. Let’s take two of the most frequent one:
You Gov on 14 May with Labor on 35% primary (above Libs on 33%), then barely three weeks later on June 4 the same pollster has Labor on 24%, lower than your current floor, Libs at 41%.
The Payman debacle did not hit the media until a few weeks later. Was there such a big shift? Unlikely, but the Labor primary nationally diminished from winter 2023 until winter 2024.
Here is Bob Hogg on the 1990 campaign, with a bit of a focus on Victoria then:
“What you need to do, as when you’ve got a poll that has a 4 per cent shift in three days, is take it very sceptically and assume that at the very best what you are seeing is a correction on a poll that was wrong before. So you ended up getting it relatively correct, reasonably correct, that aspect. I could touch on others as well, but I’m not here on a witch hunt. The 4 per cent shift was sheer and utter nonsense.”
https://australianpolitics.com/1990/04/11/bob-hogg-1990-federal-election-analysis.html/
Note he says at best it is a correction, so at worst it is quite misleading.
What I conclude is that, for whatever specific reason linked to the small sample, the You Gov poll is unreliable. I suspect that is what the major parties have concluded.
The Dunkley by-election was on 2 March. The closest You Gov poll was:
Feb 26 until Mar 6, 2024 YouGov Online VIC 377 35 (Lab) 36 (Lib) 14 (G) 6 (Hanson)
Dunkley results: Lab 41% Lib 40% G 6% (lacklustre effort) Hanson absent; RW nutter 2.5%
Labor’s problem, evident in Dunkley, was not the primary vote, but fairly poor preference vote.
While that problem may persist, you say it is compounded by a significantly lower primary.
What you have noted correctly is that it matters greatly where the primary vote is spread. That was why Howard won from behind in 1998, because Labor’s vote was poorly spread, whereas Peacock had a stronger preference vote in 1990 yet lost, despite winning big only in Victoria.
If you think the last You Gov Vic Nov sample of 386 is adequate consider this sample of 386:
Aug 28 to Oct 12, 2023 Newspoll/Pyxis TAS 366 30 (Lab) 25 (Lib) 13 4 57 43
If true that would have indicated an 8% drop in the Lib primary since election. Unbelievable.
In general, and this applies especially to the US but also to state breakdowns here, the polling industry is woefully inefficient. They should poll less often with larger samples. Why not? Because the media have got much of their audience to believe in the horserace form of analysis.
Remember that the 1998 election showed the crucial role of candidate selection. Which 5 seats do you presume the Libs will win from Labor in Victoria, excluding Deakin?
Margins are:
Chisholm 3.3% (Lib candidate is former MP for Higgins, but Labor MP has sophomore surge)
Aston 3.6% on by-election, but 2.6% to Libs in 2022 (Lib candidate is Knox Mayor from 2004!)
McEwen 3.8% (held by Labor since 2007, won by 313 votes in 2013)
Bruce 5.3% (held by Labor since 1996)
Dunkley 6.8% (Libs have same stellar Irish bloke as candidate who lost the by-election)
Holt 7.1% (held by Labor since 1980)
Hawke 7.6% (new seat in 2022; Lib primary was 26.3%)
Corangamite 7.8% (margin increased by redistribution)
Isaacs 9.5% (held by Labor since 1996)
There you have 9 seats to choose 7 from, but the Libs won only 4 of those in 2013 landslide.
New thread.
‘Post discussion’ Nadia there was also one last Newspoll: Labor on 33%. That was missing from your list. I’m also not sure Resolve and Morgan somehow provide ‘post facto’ vindication to your bloviating for months that ‘29-31’ is baked in. Perhaps Labor is yet to suffer a further decline, but again, that doesn’t vindicate you saying – for months ‘29-31’ is baked in. In fact it underscores my point: you are not providing analysis: you are in fact campaigning.