The weekly Roy Morgan poll finds Labor’s two-party lead falling back slightly to 54.5-45.5, in from 55-45 last week. However, the pollster has switched from respondent-allocated preference to using the flows from 2019, the former of were producing results more favourable to Labor. The movements on the primary vote are actually in favour of Labor, who are up half a point to 35.5% with the Coalition down one to 34%. The Greens are steady on 13%, One Nation are up one to 4% and the United Australia Party is steady on 1%.
The state breakdowns, which cannot be directly compared to last week’s due to the change in the preference calculation, have Labor leading 51.5-48.5 in New South Wales (a swing to Labor of about 3.5%), 61-39 in Victoria (about 8%), 57.5-42.5 in Western Australia (about 13%) and 62.5-37.5 in South Australia (about 12%). The Coalition leads 53.5-46.5 in Queensland (a swing to Labor of about 5%) and 60-40 from the tiny sample in Tasmania. The poll was conducted Monday to Sunday from a sample of 1401.
Further chatter from around the traps:
• Karen Middleton of The Saturday Paper reports that Liberal polling shows Tim Wilson “headed for defeat” at the hands of independent Zoe Daniel in Goldstein with 37% of the primary vote, while Josh Frydenberg’s vote in Kooyong is “currently tracking at 42%”, putting him “about 2% lower than it needs to be” to hold out against independent Monique Ryan. In North Sydney, Trent Zimmerman could potentially lose to either independent Kylea Tink or Labor’s Catherine Renshaw; both parties’ polling suggests the Liberals are in “a losing position” in the Sydney seat of Reid and the Perth seats of Pearce and Swan; Boothby in Adelaide is “leaning strongly Labor’s way”; Hasluck in Perth and Bass in Tasmania are “tightening”, presumably to Labor’s advantage; and the Brisbane seats of Ryan and Brisbane are “at risk”, as is Casey on the fringes of Melbourne, which I haven’t heard mentioned before. Parramatta and Macquarie in Sydney “are currently looking like staying with Labor”. The government’s anti-China rhetoric is also said to have resulted in a “plunge” in Liberal support among the Chinese community, harming it in Chisholm and putting Bennelong in play for Labor. For all that, the Liberals “remain confident of winning Gilmore” and are “lineball” in Corangamite. They are also “hopeful of seizing McEwen”, although “Labor sources query this”.
• In contrast to the previous assessment, Greg Brown of The Australian reports Liberal sources are “increasingly confident” that Gladys Liu will retain Chisholm and “believe Labor is shifting resources towards Higgins, where incumbent MP Katie Allen’s primary vote has dropped to 42%”. However, Paul Sakkal of The Age reports Labor believes it will win Chisholm while also having a “serious chance” in Higgins, and will “probably” retain McEwen, Corangamite and Dunkley. Sakkal further reports that Anthony Albanese will appear with Daniel Andrews today, defying suggestions he is keeping his distance from the Premier, to announce a promised $2.2 billion in federal funding for the Suburban Rail Loop, a state government project opposed by the Morrison government. The initial stage of the project will cut a north-south path through the eastern suburbs that will run neatly through Chisholm.
• Contrary to Clive Palmer’s earlier position that the United Australia Party would direct preferences against all sitting members, The Guardian reports how-to-vote cards being distributed at pre-poll voting centres have Liberal incumbents ahead of Labor in Chisholm, Reid and Bass, and ahead of teal independents in Mackellar and Wentworth.
• Nine’s endeavour to rate audience response to Sunday night’s debate eventually settled on a tied result between Scott Morrison and Anthony Albanese, although this was based on an uncontrolled exercise open to anyone who get the website form to work. The overwhelming view was the combative nature of the debate did neither protagonist any favours. A third debate will be held tomorrow night on the Seven Network.
No leftieb sorry.
Cheers Upnorth, you are a true gentleman and a scholar.
My intel from NSW Libs is feds v worried about Longman internal polling, but these guys do hate scummo. Have heard it also said they want Feds out as they will run a ‘stop the wall to wall labor’ campaign for 2023. They are also hoping everyone would have broken their baseball bats on scummo.
https://www.tallyroom.com.au/47728
The questions around YouGov’s Multilevel regression with poststratification method that need to be answered are:
1. What demographic variables did you use to estimate individual seat 2PP figures from the national sample?
2. What was the Mean Absolute Error between the estimated actual 2PP figures in previous applications of this method in previous elections?
3. For previous elections where this method was used, what was the similarity in demographic variables between previous elections and this survey?
Many moons ago as a uni student with time on my hands, I spent a summer in the library trawling through census and election data trying to find correlations between demographic variables and 2PP figures across electorates.
I started out with high hopes thinking I could crack the voting code, but in the end, no matter how I sliced and diced the data, I found no strong correlations that could be used to predict voting based on demographics. There was just too much variability in 2PP for the demographics to be reliable. It wouldn’t be surprising to most people on here that the reason for this was that actual issues (rather than just demographics) play a greater role in determining voting intention.
Given that these actual issues and voter responses to them also vary widely across location makes me sceptical about the reliability of this technique.
In Curtin the result could hinge on who comes second, Labor or Chaney. Chaney’s preferences will split some back to Lib, Labor prefs won’t.
When there’s a big swing on, there will be seats which fall which no-one, including the successful candidates (for some of whom actually getting elected was definitely not in their career plans), saw coming. In Labor’s big 2002 Victorian win under Steve Bracks, I was involved in a seat where Labor fell a few hundred votes short of overturning a 15% margin, with a campaign budget which just about extended to printing the how-to-vote cards and a candidate who was an admin assistant at the Labor state office, put forward in the last week before close of nominations because no-one vaguely local could be found to stand.
The unionists trustingly and naievly going into the warm loving embrace of Boris and the Tories, what could possibly go wrong?
C@tmomma says:
Tuesday, May 10, 2022 at 10:16 pm
CarefullyRushed @ #1282 Tuesday, May 10th, 2022 – 10:10 pm
Apart from the sample size issue, UK’s lower house elections are first-past-the-post.
It seems like YouGov only asked who someone would vote for, not realising our system works on preferences…
I wouldn’t take this YouGov poll seriously.
Exactly as BH and I have commented. They asked us one question. Who are we going to vote for?
___________________________
I thought they used Newspolls as well from April to May as well to get their 19000, another question, what did they do with the undecideds, i like the idea of a model but it is a bit of over reach with definitive results. And we still dont know the seat totals !
Seat sample size of 125 and a MOE of 9% means that it is useless.
I’d be returning the troops to their barracks.
Someone call an ambulance. https://twitter.com/jonkudelka/status/1523901410265378816
Roger Miller @ #1301 Tuesday, May 10th, 2022 – 8:21 pm
Yes it is possible (though unlikely imo) that Labor finishes second in Curtin although I would imagine if it does it would be fairly narrow between Labor and Chaney. I think a fair number of Labor and Greens voters will either faithfully or tactically switch to Chaney. The Greens will probably poll ~10% in Curtin and these are likely to break to Chaney over Labor by some margin you’d think to push Chaney to the final 2-candidate preferred even in the event she finishes 3rd on primaries.
I’m ambitious for his future!
YouGov know what they are doing and their credibility is at stake.
However, from memory Clinton’s 2016 campaign was big on a form of hybrid demographic based polling that proved badly wrong.
Many moons ago, when I was a Uni student, I spent the summer getting pissed and listening to bands. 😀
Things are going more nuts.
The Canberra Times says Labor at risk for senate. It is possible Pocock and Zed will get the 2 spots which seems unbelievable.
>People might not realise this but at the last election in Warringah, if you take Steggall out of it the TPP between LNP/ALP is only 52/48 LNP
If Labor do well in the election, does anyone think it could be a surprise ALP gain?
In 2019, Labor in Warringah got 6.6% of a vote compared with Steggall on 43 and Abbott on 39. No, there is no chance of them getting in the final two (and a good chance that any loss of votes from the transphobe pushes Steggall above 50%).
michael @ #1288 Tuesday, May 10th, 2022 – 10:41 pm
Does it include poll numbers? I heard one saying Katy Gallagher’s vote has fallen by 14% or something ridiculous.
King OMalley @ #1300 Tuesday, May 10th, 2022 – 10:21 pm
Good post
I think this is the nub of the problem – how reliably you can map census level demographics to 2pp. In particular, if the key driving factor in an electorate is a well-funded charismatic independent (quite sure that factor is not measured in the census).
The overall data, from a 19,000 sample, broken down by demographics would be informative, but probably not a seat level.
The article is not up yet but the Front Page on Sky showed the headline.
@C@tmomma
Yeah, well I did a bit of that too. Mornings at UNSW library, afternoon swim at Coogee Beach, then beers and bands at Selinas (Coogee Bay Hotel). Rinse and repeat.
I could reel off all the bands I saw around the Sydney pubs in the late 70’s to 80’s but it was basically everyone in the pub rock canon.
Nope – Katy is losing skin to Pocock (but then, so is everyone else), but certainly no sense around the traps of any concern.
Michael,
Are you talking about the headline
Dutton: China’s Solomon Islands visit ‘provocative’ ahead of Federal Election
Because that seems more significant than labor senators in Canberra. I’m very confident that labor will hold it’s spots in the territory. The public service town knows who it’s favorite is.
michael @ #1292 Tuesday, May 10th, 2022 – 10:45 pm
https://www.canberratimes.com.au/story/7718844/labor-feels-squeeze-as-independents-set-to-force-gallagher-to-preferences-too/
Here’s one from May 6.
(quota is 33%)
Gallagher 27
Seselja 25
Pocock 21
GRN 11
Rubenstein 6
UAP 6
Would be heartbreaking if Katy lost her spot. I’m trying to run through the scenarios in my head but the senate counting system is too confusing
True Believer says:
Tuesday, May 10, 2022 at 10:20 pm
Cheers Upnorth, you are a true gentleman and a scholar.
My intel from NSW Libs is feds v worried about Longman internal polling, but these guys do hate scummo. Have heard it also said they want Feds out as they will run a ‘stop the wall to wall labor’ campaign for 2023. They are also hoping everyone would have broken their baseball bats on scummo.
中华人民共和国
Your welcome cobber. Obviously the NSW Libs are leaking like a sieve. PVO’s drop today and leaked texts earlier sound like they are coming from inside the Branch.
NSW Libs would rather Albo as PM to give them a chance at winning next State election. They probably realise SfM is toxic for their brand.
I think ACT Labor members were sent an email advising against tactically voting for Pocock
jt1983 @ #1294 Tuesday, May 10th, 2022 – 10:47 pm
Good to hear. Surely Zed is more at risk
michael @ #1311 Tuesday, May 10th, 2022 – 8:41 pm
Really? What was their basis for suggesting this possible? Up until now it only seemed to be a question of whether it would be 1 ALP + 1 LNP or whether Pocock maybe pulled off an upset victory to replace Zed. I’m wondering if they are thinking Pocock + Greens + Rubenstein will win 1 seat for Pocock over Labor and LNP + ONP + UAP will be enough to elect Zed?
Steve777 the Australians article had tink on 18% and labor on 30% and Zimmerman on 38%. Presumably tinks prefs are assumed to flow strongly to Zimmerman
There are ten candidates in North Sydney. Ms Tink is lumped in with two right-wing minors and two centre-left minors with 18% between them. The poll shows Mr Zimmerman suffering a 14% swing against him, a 5% swing to Labor and 3% against the Greens.
Given that the numbers are derived from a national poll adjusted by local demographics, I don’t think that the numbers are all that credible, given that unlike recent elections, there high profile independent this time. Ms Tink would be taking votes from Liberal, Labor and Green. Many regular Labor and Greens voters are voting strategically for her. Mr Zimmerman might also suffer some minor leakage on the right (a flyer being distributed says he’ll take a wrecking ball to religious schools) but there’s no religious candidate this time.
Anyway, looking at the above, I would expect Ms Tink’s preferences to ultimately split about 50-50 between the majors, while Labor and Green preferences would go solidly to her. If Labor finishes second on primaries, Zimmerman wins. If Kylea finishes second on primaries, there’s a good chance that she wins.
Steve777
As impressive as Renshaw is, there’s no way she’s running second to Zimmerman. Shows how ‘accurate’ this polling is lol. It’s a clear race between TZ and Kylea Tink. state seat of Willoughby delivered a massive swing to nearly oust that captains pick Tim James and word is they might do a number on TZ. Have heard ucomm polling the same people in North Sydney on a daily basis with Tailored questions about Kylea and Trent. This is a popcorn seat, sit back and enjoy the show, it could go either way.
After Katy Gallagher’s shock loss to David Pocock in the ACT senate race two weeks ago, Anthony Albanese’s incoming Labor government is scrambling for a new Finance Minister.
The rumoured frontrunner is Bob Katter, who unexpectedly became a member of the Labor Party as part of an 11th hour deal on election night. Labor insiders said Katter’s support had been secured in exchange for 50 new dams being built across his electorate of Kennedy, but he may now be thrust into a federal cabinet position for the first time in his decades-long political career.
The spotlight now turns to the Speaker’s position, which it had long been assumed would go to Katter in the event of a hung parliament. Rumours are swirling that it will go to Simon Earle, the ALP’s new member for Cook, instead.
https://www.facebook.com/photo/?fbid=331201482452395&set=a.321176913454852
Another doozy from the West
I find it very hard to believe Kylea Tink is a distant third in PV here in North Sydney. She’s had by far the most volunteers and the most conflutes put up in the area.
Sportsbet has just moved Zimmerman to $1.40 and Kylea Tink And Renshaw now both at $5. 10 minutes ago Kylea was paying $2.10 and Renshaw $15.
All off a dodgy poll methodology, incredible
Labor supporters in this seat are not stupid, most will vote Tink to make sure of it.
Katter as Finance Minister. I would like to see that. Some people may need to wear vests as Bob can get rather angry when under the pump.
Re True Believer @11:02.
As impressive as Renshaw is, there’s no way she’s running second to Zimmerman.
I agree it’s hard to see that happening. I plan to vote strategically for Ms Tink to vote the sitting “Liberal” off the island. I’ll vote 1 Labor 2 Green … 24 Coalition in the Senate. Hopefully next time Labor will find Ms Renshaw a winnable seat.
Upnorth @ #1320 Tuesday, May 10th, 2022 – 10:51 pm
Seems a reasonable conclusion
Multilevel regression with poststratification approach looks interesting. One requires a predictable extrapolation method to move from the sample data to population data. Heterogeneity in voting options e.g. well resourced independents in some areas have already been mentioned I note.
One can use historical voting patterns to aid extrapolation. The issue here is emergence of new voting patterns e.g. well resourced independents competing in the election for the first time i.e. it may be less sensitive to changes in voting patterns.
Also, these methods have been used for first past the post elections so far as I am aware. Preference voting would be a new challenge methinks.
Exciting!
Wow, Downer is looking good for a septuagenarian but Frydenberg is a tiny little thing (sory that’s shortlist). Prob also going mentally ‘who the frig set me up with this guy for a photo’.
True Believersays:
Tuesday, May 10, 2022 at 11:14 pm
The people doing the political bets at these agencies are not political (i.e. they don’t know what’s garbage). Political betting is very much a side issue.
Now excuse me while i put some money on Tink.
Oh it is using logistic regression to identify predictors. I wonder how they are testing for multicollinearity. VIF? Also, best to confirm a linear relationship between the independent variables and the logit of the dependant variable.
That said, the model should improve over time with increased data. With repetitive sampling over an election campaign it may be even more sensitive to swings.
You really have to scratch your head at FPTP voting. Voting for anyone other than the top two candidates is quite literally a wasted vote. What a stupid system that makes it incredibly hard for minor parties to emerge.
Of course, political parties in the US, UK etc. are used to this system and wouldn’t know anything different – but I can’t fathom it. You have leftists in the US who refuse to vote Democrat, and basically contribute to Republicans winning. Labour in the UK has the same issue winning the informed left wing vote.
Why is it even called First Past the Post? There is no quota or majority required to win the seat.
Historyintime:
“Blood is thicker than water”, as the saying goes…
A comment on Hunter Region seats (we should be a state! Bigger population and economy than Tasmania!)
OK, off my soapbox (temporarily.)
Absent any polling (I’m sure The OZ will drop suitable bombshells,) here’s my local take (I live in Maitland)…
1) One Nation doesn’t have a rock star candidate in the seat of Hunter, who polled 21% PV and dragged Fitz from a 12 to 3% margin.
2) Labor has refused to be wedged on mining/climate change. Remember the near-useless ‘gas-fired power station’ for Kurri Kurri? The Greens were furious when Labor didn’t oppose it. Instead, Labor have given conditional support – the condition being the plant rapidly transitions to using hydrogen, rather than CH4.
3) Big Dan, the candidate for Hunter, is a symbol of empathy with the mining community. He’s the one who is visible in ALP TV ads, not so much the actual sitting members.
4) There isn’t a lightning rod like ‘Adani.’
I suspect there will be a small swing back to Labor in the Hunter Valley seats.
Griff:
I would have thought the relationship between voting and demographics in a compulsory voting system would be mutual information and thus not necessarily linear
Based on the disappearance of corflutes, the comments on corflutes, the look on Seselja’s face this morning, early polling numbers, comments from young people and a guess at the flow of preferences in the ACT Senate race, Mr Seselja, a junior Minister for the Pacific debacle, is finished as a careerist senator politician in the ACT.
The young Canberra voters seem very Green, while seeing Pocock as Green, due to the stupidity of the liberals’ “green sticker”, see Pocock as “cool” and will heavily influence the Canberra Senate vote.
Gallagher will get close to a quota on first preferences and pick up some second or third or even fourth preferences and probably get over the line.
Kim Rubenstein will impact on the result depending on preferences.
Zed Seselja looks a little “black/blue sheepish” and may struggle for preferences.
The Canberra Times is a “rubbish” , “pay for rubbish” publication, not of interest for young Canberrans and has “coffee table” relevance.
Having spoken to Pocock, he appears so “raw” ,so raw, Morrison could use him in a curry and will find the political rucks, that is Federal politics, somewhat “all blackish ” in its dominance but will be a novelty flanker of interest after winning one of the two Senate seats.
The Green candidate will be the victim of the “liberal sticker” and will bleed to the imported peacock, sorry Pocock.
(but I only get to vote once)
Hazza at 11.47
We certainly won’t be putting ‘FPTP’ and ‘democracy’ in the same sentence!
hazza4257 @ #1340 Tuesday, May 10th, 2022 – 9:47 pm
Yep I scratch my head as to why most countries use first past the post, it works a bit better if you have some form of proportional representation to balance it, but still doesn’t make much sense in single member electorates.
In 2010 in the UK the Liberal Democrats in their negotiations with the Conservatives forced a referendum to vote on introducing preferential voting but lost with 67% electing for first past the post.
https://en.wikipedia.org/wiki/2011_United_Kingdom_Alternative_Vote_referendum
E. G. Theodore @ Tuesday, May 10, 2022 at 11:54 pm
“I would have thought the relationship between voting and demographics in a compulsory voting system would be mutual information and thus not necessarily linear”
Very plausible. Wouldn’t be an issue when deployed for UK and US polling. I wonder whether this has been accounted for and how.
The only people who do not support Preferential voting are people who don’t understand it.
Obviously the pommies didn’t understand it or were too lazy to educate themselves on it.
It would be like going to a restaurant, ordering your favourite meal on the menu, the waiter saying sorry we don’t have that today so now you must get up and leave, you can not chose anything else on the menu.
@ Carefully rushed.
They are a reputable polling outfit I am sure they they would have preferential voting built into their model.
I mean conventional polling methods didn’t exactly nail it in 2019 so it’s all untested this, newspoll, morgan the lot of em’ as far I am concerned.
I wouldn’t put much faith in the projections for any one seat but the final seat tallies, I certainly wouldn’t just dismiss them out of hand because it’s a very big sample over 19 000.