Polls: Essential, Morgan and ANU Indigenous Voice survey (open thread)

Two new voting intention polls have Labor keeping its nose in front, while the Australian National University unveils an extensive post-referendum survey on the Indigenous Voice.

Essential Research’s fortnightly poll records little change on federal voting intention, with the Coalition steady on 34%, Labor down one to 31%, the Greens up one to 13%, One Nation steady on 7% and the undecided component up one to 6%. The pollster’s 2PP+ reading has the Coalition as close as it has been this term to taking the lead, with Labor down a point to 48% and the Coalition steady on 47%, the remaining 6% being in the undecided category. Monthly leadership ratings have Anthony Albanese in net negative territory for the first time since the election, his approval down four to 42% and disapproval up four to 47%. Peter Dutton is up three on approval to 39% and down one on disapproval to 42%.

A monthly “national mood” reading records a deterioriation after five months of stability, with 51% now rating the country on the wrong track (up three) compared with 30% for the right track (down four). The Coalition is credited with an edge as best party to manage the economy (33% to Labor’s 25%), reduce cost of living pressures (28% to 25%) and keep prices down (ditto), though Labor leads 37% to 19% on supporting higher wages. Forty-four per cent consider social and economic equality is decreasing (one would more naturally say inequality was increasing), with only 16% holding the opposite view. The poll was conducted Wednesday to Sunday from a sample of 1151.

Also out yesterday was the weekly Roy Morgan poll, one of Labor’s better recent results with a two-party lead of 52.5-47.5, reversing a Coalition lead of 50.5-49.5 last week. The primary votes are Labor 32% (up two-and-a-half), Coalition 35% (down two-and-a-half), Greens 13.5% (steady) and One Nation 5% (down one-and-a-half). The poll was conducted last Monday to Sunday from a sample of 1379. The Australian also published further results yesterday from the recent Newspoll showing only 16% consider themselves better off than they were two years ago, compared with 50% for worse off. The 18-to-34 cohort offered the most favourable response, with 29% for better off and 37% for worse off.

The Australian National University’s Centre for Social Research and Methods also treats us to a 93-page report on the October 14 Indigenous Voice referendum, based on a survey of 4219 respondents from October 17 to 29. I haven’t absorbed this one yet, but the report is here.

Monday miscellany: Liberal preselection and NSW redistribution latest (open thread)

Dissension in the ranks among Tasmania’s federal Liberals, plus developments in Bennelong and North Sydney.

Essential Research’s fortnightly poll is due this week, although that will be less interesting than it was if it’s decided to stop publishing voting intention numbers, as was the case with last fortnight’s poll. RedBridge Group has also been in the field, with results potentially to be published this week. Other than that:

Nine Newspapers reports Gavin Pearce, Liberal member for Braddon in north-western Tasmania, is withholding his preselection nomination until his electoral neighbour, Bass MP Bridget Archer, is expelled from the party. Archer crossed the floor last week to vote against Peter Dutton’s motion for a royal commission into child sex abuse in Indigenous communities, which the hitherto indulgent Dutton described as a “mistake”. The report says four other conservative MPs are backing Pearce’s course, with one saying there was “a chance party officials would side with him”, while acknowledging this would likely mean Archer retaining her seat as an independent.

Linda Silmalis of the Daily Telegraph reports that Scott Yung, who came within 69 votes of defeating Chris Minns in Kogarah amid a backlash against state Labor among the Chinese community at the 2019 state election (UPDATE: It is noted in comments that Yung merely came within 69 votes of Minns on the primary vote, and that the final two-party margin was actually 1.8%)), is set to become the Liberal candidate for Bennelong after the withdrawal of rival nominee Craig Chung. Silmalis reports Yung had the backing of Peter Dutton, whereas Chung was favoured by moderates. Jerome Laxale gained the seat for Labor in 2022, the party’s second ever win in the seat after Maxine McKew’s famous victory over John Howard in 2007.

• Next door in North Sydney, which Trent Zimmerman lost to teal independent Kylea Tink at last year’s election, the Sydney Morning Herald’s CBD column reports that Sophie Lambert, media manager at the NSW Education Department, has nominated for Liberal preselection. Lambert’s preselection brochure says the seat was “stolen at the last election by a concerning new wave of politics”, and shows her pictured alongside conservative favourite Katherine Deves. The matter could be complicated by the current redistribution process, in which the seat could be radically redrawn or potentially abolished.

• Further to the above, responses to the call for public suggestions for the redistribution of New South Wales seats will be published on the Australian Electoral Commission site today, presumably to include the wish lists of the major parties and other interested actors.

Midweek miscellany: Morgan, JWS Research True Issues, referendum pollster performance (open thread)

A poll records the Coalition with a two-party lead for the first time since the election, but there are reasons to be dubious.

The Courier-Mail will have a Queensland state poll through at 10am local time, presumably from YouGov, which the paper has been promoting with its trademark subtlety. Alongside the usual bilge about how the results will rock the state to its foundations, we are informed the poll includes questions on how respondents might vote if Labor changed leaders.

In what’s likely to be a fallow period for federal polling post-referendum, Roy Morgan turned a few heads with its weekly voting intention result, which is the first poll this term to credit the Coalition with a two-party lead, by a bare 50.5-49.5 margin, after Labor led 54-46 a week previously. However, the result is in large part down to an anomalous flow of respondent-allocated preferences: the primary votes of Labor 32% (down three), Coalition 36% (up two), Greens 14% (steady) and One Nation 4.5% are all in the ballpark of the 2022 election result, and in fact convert to 53-47 in Labor’s favour if preference flows from the election are applied. The poll was conducted in the immediate aftermath of the referendum, from last Monday through to Sunday, from a sample of 1383.

JWS Research has released its monthly True Issues survey of issue salience, which finds 56% nominating the cost of living when asked unprompted to identify their three most important issues, shooting up from 43% in June. Housing and interest rates, health and aged care and environment and climate change are little changed in second through to fourth place, with health levelling off after a long post-pandemic slide. An index measure of the federal government’s performance is down to 48, after the four previous readings since the 2022 election came in at 52 or 53. The survey was conducted

With most of the votes from the referendum now in, here’s a ranked listing of how the pollsters performed:

Newspoll quarterly breakdowns (open thread)

Seven weeks’ aggregation of polling points to Victoria and Western Australia as areas of relative weakness for federal Labor.

The Australian has published aggregated Newspoll breakdowns from polling conducted from August 28 to October 12, encompassing the four polls conducted since Pyxis Polling took over. The overall sample is 6378, having been boosted by 2368 in the pre-referendum poll (which recorded 57% for no and 37% for yes, converting to a bang-on-accurate 60.6-39.4 after exclusion of the uncommitted).

Keeping in mind that the previous set of results, from February 1 to April 3, were conducted by a different agency, the results show Labor’s two-party lead up slightly in New South Wales (from 55-45 to 56-44) and South Australia (from 56-44 to 57-43), but down solidly in Victoria (from 58-42 to 54-46) and Western Australia (57-43 to 53-47). The Coalition is credited with a 52-48 lead in Queensland after a 50-50 result last time, and we are given the rare treat of numbers for Tasmania, where Labor leads 57-43. This suggests swings to Labor of about 4.5% in New South Wales, 2% in Queensland, 3% in South Australia and 2.5% in Tasmania, and to the Coalition of 1% in Victoria and 2% in Western Australia.

The age breakdowns do not repeat a Labor blowout last time among the 18-to-34 cohort, which has progressed over the term’s three Newspoll breakdowns from 65-35 to 69-31 to 64-35. A five-point Coalition gain on the primary vote to 26% means they do not again finish behind the Greens, who are up a point to 25%, with Labor down six to 37%. The results among the older cohorts are essentially unchanged.

Further results suggest the opening of a substantial new gender gap, or of distinctive house effects between the two polling outfits. Where last time Labor was credited with a slightly bigger lead among men (55-45) than women (54-46), its advantage is now out to 56-44 among women and in to 51-49 among men. Income breakdowns now conform with the traditional pattern, with a 57-43 Labor lead among households on annual incomes of up to $50,000 progressively receding to 50-50 among those on $150,000 or more. The previous breakdowns had Labor strongest in the two middle-income cohorts.

Weekend miscellany: redistribution and referendum latest (open thread)

Referendum results displays; progress in the federal redistribution process; party registration news.

I suspect we’re entering something of an opinion poll drought, with media polling budgets having been exhausted in the last stages of the referendum campaign. On that subject, my live results feature continue to update on a daily-or-so basis. There is also Simon Jackman’s, which includes an impressive feature allowing the user to observe relationships between booth results and various electoral and demographic measures.

Other news:

• The federal redistribution processes for Western Australia and Victoria, which will respectively increase the state’s representation from 15 seats to 16 and reduce it from 39 to 38, moved along a notch this week. Submission deadlines for suggestions have been set at November 17 for Western Australia and November 24 for Victoria; supporting information including the enrolment data that will set the quotas for enrolment (both current and projected to 2028) have been published for Western Australia and will follow for Victoria on Wednesday. The deadline for suggestions in New South Wales, which reduces from 47 to 46 seats, is this coming Friday.

• The former Liberal Democratic Party, which has lost the right to have the word “liberal” in its name following legislative changes before the last election, is seeking to register as the Libertarian Party (with a proposed logo that looks to be rather a lot like that of Queensland’s Liberal National Party). This is now its formal name in Victoria, where it boasts one seat in the Legislative Council, though it retains its old name in New South Wales, where ditto.

• The Australian reports the Joint Standing Committee on Electoral Matters will bring down its final report on the 2022 election next month. Most of the terms of reference were addressed in the interim report, the exception being “proportional representation of the states and territories in the Parliament, in the context of the democratic principle of ‘one vote, one value’”.

Indigenous Voice aftermath

Polls, podcasts and opinion pieces in the wake of another failed constitutional referendum.

Click here for full display of Indigenous Voice referendum results.

I am continuing to update the live results feature about twice daily, which should get incrementally worse for yes as late batches of postals are added. Antony Green notes that no got 56.2% on the election day vote, 64.5% on the pre-poll vote and 68.7% on the postal vote – disparities far greater than typical at elections, which have had the effect of widening the no lead as the count has progressed. Further reading (and listening) on the referendum:

• I had a paywalled piece in Crikey on Monday that observed the strength of the yes vote in Indigenous communities and teal seats, respectively contrary to suggestions advanced during the campaign by no advocates and voluminous newspaper reports on supposed internal polling.

• Nine Newspapers has further results from the Resolve Strategic poll of 4728 respondents from September 22 to October 4, which show Jacinta Nampijinpa Price viewed positively by 26%, neutrally by 22% and negatively by 16%; Nyunggai Warren Mundine respectively at 22%, 25% and 15%; and Lidia Thorpe at at 9%, 23% and 32%. All scored much higher on name recognition than yes campaigners Noel Pearson (11% positive, 20% neutral, 13% negative), Marcia Langton (9%, 15% and 12%) and Thomas Mayo (5%, 16% and 10%).

The Guardian has various attitudinal results from Accent Research and Octopus Group, the former of which is headed by the eminent Shaun Ratliff, the most interesting of which is Voice support by top news source of choice. Sky News and The Guardian reigned at opposite ends of the table; Twitter uses favoured the Voice by 54-46 while Facebook went 67-33 the other way.

• DemosAU has published results on attitudes to the referendum from 7341 respondents over the past three months, which are covered at The New Daily. Among other things, the survey found 50% of Greens and 45% of Labor voters believed “the constitution needs altering to reflect our modern nation” compared with 17% for the Coalition and 14% for One Nation, with the latter much more likely to rate that “the constitution has worked satisfactorily for over a century, we shouldn’t alter it”.

• I discussed the results of the referendum the day after the event with Ben Raue of The Tally Room on a podcast you can listen to below.

Indigenous Voice referendum live

Live coverage of results from the Indigenous Voice referendum.

Click here for full display of Indigenous Voice referendum results.

Sunday morning
Below is the output of a linear regression model that uses four demographic variables (together with controls for state-level effects) to explain 87% of the variation in the yes vote by electorate, limited for technical reasons to the 141 seats of the five mainland seats.

The choice of the four demographic variables was constrained by the need to pick ones that didn’t correlate over-much with each other. This tends to mean they could have been replaced with other variables they correlated with and still produced a robust result. To go through the four in turn:

Finished School, i.e. completed year 12. As you would expect, yes did very considerably better in seats with high educational attainment and occupational categories related with it. Such seats also tend to have high numbers of people in their twenties and thirties and renters, and few labourers.

Secular. Seats with a lot of people who identified as having no religious affiliation were significantly stronger for yes. This is a favourite variable of mine, because it reliably associates with support for post-materialist causes including a republic, same-sex marriage and the Indigenous Voice, and also with voting for Greens and teals.

Owned. Yes did worse in seats where a lot of people owned their homes, which in turn correlates strongly with the 60-plus age cohort, a measure of which might well have taken its place in the model.

Age0to19. Seats with a lot of children — or, looked at another way, mortgage-paying families — tended to do poorly for yes.

The four “state” variables tell us only that yes did better in the two bigger states than the three smaller ones, which we can tell more efficiently by looking at the results. In particular, they tell us that it did so over and above what might be expected from demographic variation between the states on the variables described above.

Saturday night

1.45am. Results updating again now, presumably with little if anything further to be added for the evening.

12.50am. There are still a few results outstanding in WA, but I’m going to have to turn off my results updating for a couple of hours. If you’ve found it in any way entertaining or useful — and it’s still the only place where you can find the results at booth level — please consider helping out with a contribution through the “become a supporter” button at the top of the site.

11.55pm. Most of the Mobile Remote Team results are in from Lingiari now, and the yes vote among them has come back to 71.6%.

10.39pm. My live results stalled for a bit there because I was hacking around trying to get Remote Mobile Team results to appear in Lingiari, which they are now doing. These are of interest because they serve largely, though not exclusively, remote Indigenous communities. With six out of 22 reporting, the results are 2908 yes (79.7%) and 742 no (20.3%).

9.34pm. Looking increasingly certain now that Victoria will also be a no. Early numbers in Western Australia confirm what you would expect there too.

8.35pm. The prospect of a yes majority in Victoria has been drifting away, with my projection of 51.7% no getting steadily closer to the raw result of 52.2%. These numbers are probably flattering yes, because postal votes in particularly are likely to be very conservative.

8.07pm. First numbers from Northern Territory are about 70-30 to no, but this could be a bit of a rollercoaster due to the peculiarity of heavily indigenous remote mobile booths — no insight I can offer on when those might report.

7.58pm. Malarndirri McCarthy on the ABC going through results from heavily indigenous booths in Queensland, which are as high as 75%. You can see the relevant booths on my pages for Leichhardt and Herbert — click activate at the bottom of the links to see the map display, and the ones with green (i.e. yes) numbers are pretty much the ones McCarthy was going through.

7.55pm. If the potential for voting to happen after the result is confirmed be deemed an issue, WA is getting a bad case of it, thanks to daylight saving and decisive results on the eastern seaboard.

7.53pm. My system has come through with the formality of calling Queensland for no, joining New South Wales, South Australia, Tasmania and the national result. The ACT is called for yes, and I have a 68% probability for no in Victoria. Nothing yet from the Northern Territory, where polls closed 23 minutes ago, or Western Australia where they do not close for over an hour.

7.31pm. My system has also been calling the national vote for no for some time, and clearly won’t be long in calling Queensland for no. Victoria remains very close, with yes having consistently been projected to be a shade below 50%.

7.24pm. Got that out a matter of seconds before Antony Green said the same.

7.23pm. My system is calling South Australia for no.

7.18pm. First results emerging from Queensland, inevitably going very heavily to no from small rural booths.

7.17pm. Behind Antony Green’s eight-ball here, but my system is now calling NSW for no.

7.07pm. I’m projecting a tight result in Victoria — raw yes vote is 47.5%, but my adjustment to account for where the votes are from gets it to 49.2%.

7.03pm. ABC calling New South Wales for no, though my system isn’t quite there yet.

6.59pm. And for what very little it’s worth, my system is calling the ACT for yes.

6.57pm. My system is calling Tasmania for no, and so apparently is Antony Green’s.

6.52pm. Among many other dubious things, the no campaign succeeded in propagating several news reports to the effect that teal seats were going to do badly for yes. But so far, yes ranges from 58% in Mackellar to 73% in Kooyong.

6.48pm. I’m calculating probabilities in a way I’m not confident enough about to include in the results pages, but they’re getting very close to calling Tasmania for no and ACT for yes — although Tasmania could potentially swing back as more Hobart booths report.

6.45pm. First booth in from South Australia — Darke Peake in Grey — records 55 votes for no and three for yes.

6.40pm. If you’re finding the results feature of any use or interest, you may perhaps care to make a contribution through the “become a supporter” buttons you’ll find at the top of this page or on the results page itself.

6.38pm. My projection for yes has improved in Victoria, from around 43% to 46.6%. But thumping no leads elsewhere.

6.27pm. So far I’m projecting very similar results in New South Wales, Victoria and Tasmania — the three states where voting has closed — ranging from 56.6% to 57.7%. Yes well ahead from a tiny count in the ACT.

6.21pm. With only yes and no to count, rather than multiple candidates never mind two-candidate preferred, the count is clearly going to progress very quickly. So far, there are seven booths with yes majorities and 44 with no.

6.20pm. The very earliest results are presumably from the most rural of areas, whereas my projection works off seat results. So if the booth results so far are conservative even by the standards of the seats they are in, as I suspect to be the case, the early projections should be unflattering for yes.

6.15pm. Small booths in from Farrer and Parkes in NSW and Wannon in Victoria, and my results seem to be working.

6pm. Polls have closed in eastern states with daylight saving. My results feature consists of a front page summarising results at national, state/territory and seat level, and results pages at House of Representatives seat level which include booth results in both table and map form (for the latter, click the activate button at the bottom of the relevant page). The seat results pages can be accessed from the drop-down menu or the “results by electorate” section at the bottom end of the main page.

The “projected” results for the national and state/territory votes make use of the seat-level estimates from Focaldata’s multi-level regression with post-stratification exercise as a baseline for measuring such results as are reported. However much they differ from Focaldata’s estimates of the relevant seats is projected on to Focaldata’s aggregated estimates. Doubtless this will be noisier than the booth-matched swings methods that can be applied at elections, but it should at least go some way towards correcting for the peculiarities of the early numbers.

Indigenous Voice: Newspoll, JWS Research and DemosAU polls

A fair bit of diversity in late polling on the margin for the Indigenous Voice, but clear unanimity on the result.

I have a live results feature standing ready for service this evening, which will presume to offer projections according to a somewhat experimental method that will be explained further below, together with individual results pages by electoral division featuring booth results displayed in table and map form (click on “activate” at the bottom of the individual seat pages for the latter). It may be the only place where the latter data will be available online, not counting the Australian Electoral Commission media feed from which it will be extracted.

Some final polls are in, with pollsters collectively offering a fairly wide spread that means the contest for post-result bragging rights is wide open:

The Australian has a Newspoll result that concurs with others in finding movement back to yes – though by an entirely insufficient three points to 37%, with no down one to 57%. The 1413 sample from this poll has been combined with 1225 sample poll from last week to produce state breakdowns with no leading 54-41 in New South Wales, 51-43 in Victoria, 65-30 in Queensland, 65-28 in Western Australia, 60-33 in South Australia and 55-38 in Tasmania. The new batch of polling means we are also treated to a second set of Newspoll federal voting intention numbers in a week, in this case putting Labor ahead 54-46 (53-47 last week) from primary votes of Labor 36% (up two), Coalition 35% (down one), Greens 12% (steady) and One Nation 6% (up one).

• A JWS Research poll in the Financial Review has no at 52% and yes at 39%, converting to 57-43 after exclusion of the undecided. The poll was conducted Friday to Monday from a sample of 922.

• The Australian related yesterday that a poll conducted from October 1 to 9 by DemosAU, whose director George Hasanakos had a polling analysis website back in the day called Poliquant, had no leading 57% to 30% in both Queensland and Western Australia.

The projection model in my results will use the seat-level estimates from Focaldata’s multi-level regression with post-stratification exercise as a baseline for measuring the results as they are reported. However much the results that are in differ from what Focaldata predicted will then be projected on to Focaldata’s overall results at state level. Doubtless this will be noiser than the booth-matched swings methods that can be applied at elections, but it should at least go some way towards correcting for the peculiarities of the early numbers.

My attention this evening will be focused on the referendum, but I will have some sort of a post up following the progress of counting in New Zealand, a dedicated thread for which is here.