Essential Research 2PP+: Labor 48, Liberal 46, undecided 7

Essential Research joins in on the tightening race narrative, although it gets there through preferences rather than primary votes. Also featured: many seat polls, one of which has independent Kate Chaney leading in Curtin, and even polls for the Senate.

The final Essential Research poll for the campaign has Labor leading by 48% to 46% on the pollster’s “2PP+” measure, in from 49% to 45% a fortnight ago, with the remainder undecided. However, both the Coalition and Labor are steady on the primary vote at 36% and 35%, the change mostly being accounted for by respondent-allocated preferences. For the minor parties, the Greens are down one to 9%, One Nation is up one to 4%, the United Australia Party is down one to 3% and independents are up one to 6%. Leadership ratings are little changed: Scott Morrison is down one on approval to 43% and up one on disapproval to 49%, Anthony Albanese is up one to 42% and steady on 41%, and Morrison’s lead as preferred prime minister moves from 40-36 to 40-37.

My reading of the state breakdowns is that the results are likely too strong for the Coalition in Victoria and too weak in New South Wales, at least in relative terms — using previous election flows, Labor leads by about 53-47 in New South Wales and 51-49 in Victoria, respectively compared with 51-49 to the Coalition and around 53.5-46.5 a fortnight ago. The Coalition is credited with a lead of about 52-48 in Queensland, a swing to Labor of over 6%, compared with about 50-50 in the previous poll. Sample sizes here would have been about 500 in New South Wales, 400 in Victoria and 300 in Queensland.

The poll was conducted Wednesday to Tuesday from a sample of 1599. It has made little difference to the BludgerTrack aggregate, since it goes off primary votes which are little changed. The more heavily weighted Newspoll and Ipsos polls that are yet to come stand more chance of moving its needle, which they will want to do in the Coalition’s favour if it is to end up with a two-party preferred reading that I personally would think plausible.

Other polling news:

The West Australian today had a poll by Utting Research showing independent Kate Chaney leading Liberal member Celia Hammond in Curtin by 52-48 on two-candidate preferred, from primary votes of 38% for Hammond, 32% for Chaney, 13% for Labor, 9% for the Greens and 3% for the United Australia Party. The poll was conducted on Monday from a sample of 514.

Katherine Murphy of The Guardian reports Labor is “increasingly bullish” about Brisbane and Higgins, and further perceives “opportunity” in Bennelong and Ryan. The report says Labor internal polling has Scott Morrison’s disapproval rating across the four seats at 62%, 65%, 57% and 58% respectively, whereas Anthony Albanese is in the mid-forties in approval and mid-thirties on disapproval.

Samantha Maiden of reports on uComms polls for GetUp! of five seats chosen because of the effects on them of fires and floods. These have Labor member Fiona Phillips leading Liberal challenger Andrew Constance by 57-43 in Gilmore, but Liberal candidate Jerry Nockles leading Labor member Kirsty McBain by 51-49 in Eden-Monaro. Labor’s Peter Cossar is credited with a 55-45 lead over Liberal National Party member Julian Simmonds in Ryan, although his primary vote of 27% is not far clear of the Greens on 22%. Labor member Susan Templeman leads Liberal candidate Sarah Richards by 56-44 in Macquarie, and Nationals member Kevin Hogan leads Labor’s Patrick Deegan by 51-49 in Page, which he won by 9.4% in 2019. I have tended to think this pollster’s results have been flattering to Labor due to a limited weighting frame encompassing only age and gender, a notion disabused here only by the Eden-Monaro result.

• Speaking of which, The Australia Institute has a now dated uComms poll of Higgins from May 2 that had hitherto escaped my notice, showing Liberal member Katie Allen will lose either to Labor by 54-46 or the Greens by 55-45. Using the results of the forced-response follow-up to allocate the initially undecided, the primary votes are Liberal 36.9%, Labor 29.8% and Greens 19.9%, in which case it would clearly be Labor that won the seat.

• Redbridge Group has published full results of its poll for Climate 200 in North Sydney which, as noted here yesterday, had Liberal member Trent Zimmerman on 33.3%, independent Kylea Tink on 23.5% and Labor candidate Catherine Renshaw on 17.8%, with 7.5% undecided. The full report has results on a follow-up prompt for the undecided, and much more besides.

Senate news:

• The Australia Institute has Senate voting intention results for the five mainland states, which suggest the Greens are looking good across the board, One Nation might be a show in New South Wales as well as Queensland, the United Australia Party are competitive in Victoria, a third seat for Labor is in play in Western Australia, and Nick Xenophon might be in a tussle with One Nation for the last place in South Australia. With a national sample of 1002 though, there are naturally wide margins of error on these results. The accompanying report also include interesting data on respondents’ levels of understanding of the Senate voting process.

The Advertiser reports an “internal party polling” – it does not say whose – of the South Australian Senate race has Labor on 34%, Liberal on 23%, the Greens on 12%, Nick Xenophon on 6%, One Nation on 4% and Rex Patrick on 3%. I imagine the most likely outcome of such a result would be two seats each for Labor and Liberal plus one for the Greens and one for Nick Xenophon. The poll had a sample of 644, with field work dates not disclosed.

• A joint sitting of the Western Australian parliament has just confirmed that Ben Small of the Liberal Party will fill his own Senate vacancy, having addressed the dual citizenship issue that forced his resignation. Small has the third position on the Liberal ticket at Saturday’s election.

Non-polling news:

Sky News has obtained a voicemail message from 2019 in which Dai Le, who is threatening Kristina Keneally’s effort to use Fowler to move from the Senate to the House of Representatives, stated she still had control of the Liberal Party’s Cabramatta branch despite having been suspended from the party after running as an independent mayoral candidate in 2016. Le has been keen to express her independent credentials in a seat where the Liberal Party has little support.

Samantha Maiden of reports Labor focus groups in Bennelong, Deakin and Pearce registered negative reactions to the Liberal Party’s policy of allowing home buyers to make use of their superannuation.

• The count for the Tasmanian Legislative Council seat of Huon has been completed, with the distribution of preferences resulting in a victory for independent candidate Dean Harriss over Labor’s Toby Thorpe at the final count by 11,840 votes (52.55%) to 10,693 (47.45%). Thorpe led on the primary vote, but Liberal preferences flowed to Harriss more strongly than Greens preferences did to Thorpe. This maintains the numbers in the chamber at four Labor, four Liberal and seven independents, although an ex-Labor member has been replaced by one with a conservative pedigree, his father having served in parliament both as an independent and with the Liberal Party.

Resolve Strategic: Coalition 34, Labor 31, Greens 14

The final Resolve Strategic poll of the campaign has Labor’s primary vote even lower than 2019, but still has them ahead on two-party preferred.

The final Resolve Strategic poll for the Age/Herald campaign provides evidence for the long-delayed narrowing, with Labor down three points on the primary vote since the mid-campaign poll to 31% and the Coalition up a point to 34%. The Greens are down a point from an improbable height last time to 14%, while One Nation are up one to 6%, the United Australia Party are down one to 4%, independents are up two to 6% and others are steady on 4%. The accompanying report provides a two-party preferred result based on historic preference flows of 52-48 in favour of Labor, compared with 54-46 last time. Anthony Albanese has nonetheless narrowed the gap on preferred prime minister, which is in from 39-33 to to 40-36, and he is up three on approval to 40% and steady on disapproval at 48%. Scott Morrison is up two on approval to 43% and up one on disapproval to 50%.

The poll was conducted Thursday through to today from a sample of 2049 compared with the usual 1400 or so, because it “added several hundred telephone responses to the customary online responses to give it a larger base”. This hasn’t made the pollster any more generous with its breakdowns, which remain limited to gender and the three largest states. By my calculation, the Coalition leads by around 51.5-48.5 in New South Wales, which is fully ten points stronger for the Coalition than the previous poll and suggests no swing compared with 2019. Labor’s 53-47 lead in Victoria is also about the same as the 2019 result, and four points weaker for Labor than the previous poll. However, the Coalition is now credited with a lead of 53-47 in Queensland, representing a roughly 5.5% swing to Labor and a shift in Labor’s favour of at least four points compared with the previous poll. The “rest of Australia” measure has shifted around four points in Labor since the previous poll, and around seven points compared with the 2019 result.

With increasing talk about women dooming the Coalition to defeat, the poll offers an interesting take on the gender gap in crediting the Greens 19% support among women compared with 9% among men, feeding into a four-point gap on two-party preferred. Morrison’s personal ratings are in fact quite a bit stronger among women than in the previous result, resulting in only a modest distinction on his net approval, although Morrison leads 45-37 among men but is tied with Albanese at 36-36 among women.

UPDATE: David Crowe of the Age/Herald offers a two-party preferred of 52-48 in one article and 51-49 in another. My calculation splits it right down the middle.

UPDATE 2: The final Essential Research poll of the campaign is probably available on The Guardian’s site by the time you read this – Peter van Onselen tweeted last night that it had Labor leading 51-49. Also just out is an Utting Research poll for The West Australian showing teal independent Kate Chaney leading Liberal member Celia Hammond 52-48 in Curtin, from primary votes of 38% for Hammond, 32% for Chaney, 13% for Labor, 9% for the Greens and 3% for the United Australia Party. The poll was conducted Monday from a sample of 514. All this and a lot more will be covered in a new post tomorrow.

Morgan: 53-47 to Labor

Roy Morgan ends its weekly campaign series finding the major parties collectively plumbing new depths, but with Labor in far the better position of the two. Plus yet more internal polling scuttlebutt, this time from Warringah, Fowler and North Sydney.

Roy Morgan has dropped its weekly federal campaign poll, which shows Labor’s two-party preferred lead narrowing from 54.5-45.5 to more believable 53-47. While this is Labor’s weakest two-party result from Roy Morgan since October, the respondent-allocated preferences measure the pollster used until last week had Labor at least one point higher when it was tied with the Coalition on the primary vote, and sometimes substantially higher. The two-party numbers are now determined by allocating preferences flows as per the result of the 2019 election.

The poll shows both major parties on what even by recent standards are remarkably low primary votes of 34% each, with Labor down 1.5% on last week and the Coalition steady. The Greens are on 13%, One Nation is on 4% and the United Australia Party is on 1%, all unchanged on last week, with independents up half to 9% and “others” up one to 5%.

The usual two-party state breakdowns have Labor leading 52-48 in New South Wales, out from 51.5-48.5 last week for a swing of around 4% compared with 2019; 57-43 in Victoria, in from 61-39, also for a swing of around 4%; 54.5-45.5 in Western Australia, in from 57.5-42.5 for a swing of about 10%; and 58-42 from the tiny Tasmanian sample. The Coalition leads 53-47 in Queensland, in from 53.5-46.5 for a swing to Labor of about 5.5%, and 51-49 in South Australia, its first lead on this small sample measure since October, and a rather stark contrast to Labor’s 62.5-37.5 lead last week (the result in 2019 was 50.7-49.3 in favour of Labor).

The poll was conducted Monday to Sunday from a sample of 1366. It will naturally be Morgan’s last of the campaign if it sticks to its usual schedule, although it may well pull something from its hat on the eve of the big day. The final Essential Research poll will reportedly be out on Wednesday, and it’s a known known that Newspoll and Ipsos each have a poll to come (it’s disappointing that we haven’t seen any state breakdowns from Newspoll, but hope springs eternal), and I assume the same will be true of Resolve Strategic. Until then:

Ten News is teasing yet another result of Liberal Party internal polling from Peter van Onselen, this time suggesting Katherine Deves is “in with a shot” of unseating Zali Steggall in Warringah, seemingly along with results from other seats including Parramatta and Bennelong.

Tom McIlroy of the Financial Review reports a Redbridge Group poll for North Sydney commissioned by Climate 200 has Liberal member Trent Zimmerman on 33.3%, independent Kylea Tink on 23.5% and Labor candidate Catherine Renshaw on 17.8%, with 7.5% undecided. The poll was conducted between May 3 and 14 from a sample of 1267.

James Morrow of the Daily Telegraph reports a poll of Fowler conducted by Laidlaw Campaigns, presumably for independent Dai Le’s campaign, has Kristina Keneally leading Le by 45% to 38% after distribution of preferences and without excluding the 17% undecided. The poll also found Le was viewed favourably by 28% and unfavourably by 10%, while Keneally was at 24% and 30%. It was conducted three weeks ago from a sample of 618.

Katharine Murphy of The Guardian notes a campaign endorsement for Katy Gallagher by Julia Gillard reflects concern that a win for independent candidate David Pocock in the Australian Capital Territory Senate race could come at the expense of Gallagher, and not Liberal Senator Zed Seselja as had generally presumed. A recent Redbridge poll suggested Pocock had gauged enough of his support from Labor to reduce Gallagher to 27%, well below the one-third quota for election

• The Australian Electoral Commission has issued a statement announcing that advertising by conservative activist group Advance Australia linking independents Zali Steggall and David Pocock to the Greens is in breach of the Electoral Act. The relevant section is section 329, banning material “likely to mislead or deceive an elector in relation to the casting of a vote”. The section is very commonly used as the basis of unsuccessful complaints about misleading political advertising, when it has been consistently been found to apply very narrowly to efforts to deceive voters into casting their ballots differently from how they intended. However, the statement suggests that the ruling made after the 2019 election over Chinese language signs encouraging votes for the Liberals in the seat of Chisholm, although dismissed, offers “a new judicial precedent” that seemingly paved the way for a more expansive interpretation.

• Ben Raue has produced highly instructive charts showing how each state’s two-party preferred vote has deviatied from the national result at elections going back to 1958.

• At the risk of getting ahead of ourselves, Andrew Clennell of Sky News reports that Josh Frydenberg is “said to have the numbers” against Peter Dutton to succeed Scott Morrison as Liberal leader should the party lose the election.

Move it or lose it

As privately conducted seat polling continues to fly thick and fast, a series of disputes have developed over whether election candidates really live where they claim.

Three items of seat polling intelligence emerged over the weekend, none of which I’d stake my house on:

Samantha Maiden of reports a uComms poll of 831 respondents for GetUp! shows independent Zoe Daniel leading Liberal member Tim Wilson in Goldstein by 59-41 on two-candidate preferred and 35.3% to 34.0% on the primary vote, with Labor on 12.5%, the Greens on 8.9% and an undecided component accounting for 4.6%. It should be noted that the campaigns are clearly expecting a closer result, and that uComms is peculiarly persisting with a weighting frame based entirely on age and gender, which was common enough before 2019 but has generally been deemed insufficient since.

Andrew Clennell of Sky News yesterday related polling conducted for the Industry Association of 800 respondents per seat showed Liberals incumbents trailing 58-42 in Robertson and 53-47 in Reid, and Labor leading by 56-44 in Parramatta and 54-46 in Gilmore, both of which the Liberals hope to win, and 57-43 in Shortland, which has never looked in prospect. The polling provided better news for the Coalition in showing the Liberals with a 57-43 lead in Lindsay and the Nationals trailing by just 51-49 in Labor-held Hunter. Clennell further related that “similar polling conducted earlier in the campaign also shows the government behind in Bennelong and Banks”. A fair degree of caution is due here as well: no indication is provided as to who conducted the polling, and its overall tenor seems rather too rosy for Labor.

• By contrast, local newspaper the North Sydney Sun has detailed results for North Sydney from Compass Polling, an outfit hitherto noted for polling conducted for conservative concerns that has then been used to push various lines in The Australian, which suggests independent Kylea Tink is running fourth and the threat Liberal member Trent Zimmerman faces is in fact from Labor’s Catherine Renshaw. The poll has Zimmerman on 34.9%, Renshaw on 25.0% and Tink on 12.4%, with Greens candidate Heather Armstrong on 15.0%. The online poll was conducted on May 6 from a sample of 507.

Speaking of staking houses, a fair bit of the electorate-level noise of the late campaign has related to where candidates claim to live for purposes of their electoral enrolment. The ball got ralling when The Australian reported that Vivian Lobo, the Liberal National Party candidate for the marginal Labor-held seat of Lilley in Brisbane, appeared not to live within the electorate at Everton Park as claimed on his enrolment. Labor’s demand that Lobo be disendorsed – always a challenging prospect after the close of nominations – was complicated on Saturday when Labor’s star candidate for Parramatta, Andrew Charlton, blamed an “oversight” for his enrolment at a Woollahra rental property owned by his wife a month after they moved to the electorate he hopes to represent.

The Liberals have naturally sought to maximise Labor’s embarrassment, with the Daily Mail quoting one over-excited party spokesperson calling on Labor to refer Charlton to the Australian Federal Police, as Lobo had been by the Australian Electoral Commission. But as the AEC’s media statement on the issue noted, Lobo has actively identified the Everton Park property as his residential address on his enrolment and nomination forms, potentially putting him in the frame for providing false or misleading information to a Commonwealth officer, punishable by a maximum 12 months’ imprisonment and $12,600 fine.

By contrast, Charlton’s is a sin of omission: failing to update his enrolment within the prescribed one month time frame after changing address, punishable only by a fine of $222 and in practice hardly ever enforced. There remains the question of why he was enrolled in Woollahra rather than at the Bellevue Hill property where he and his wife were living before their recent move to North Parramatta, but it’s difficult to see what ulterior motive might have been in play.

Now a new front has opened in the seat of McEwen on Melbourne’s northern fringe, where the Liberals are hopeful of unseating Labor member Rob Mitchell. Sumeyya Ilanbey of The Age reports today that Labor has asked the AEC to investigate Liberal candidate Richard Welch, who lives 50 kilometres from the electorate in the Melbourne suburb of Viewbank but listed an address in Wallan on his nomination form. Welch claims he had been living separately from his family in Wallan at the time of his nomination, but moved to his own property in Viewbank just days later after his landlord sold the house and terminated the lease.

Newspoll: 54-46 to Labor

Newspoll continues to present a stable picture on voting intention, crediting Labor with a two-party lead a few points stronger than ahead of the greater pollster failure of 2019.

The Australian reports on an unusually timed release from Newspoll, although the field work period from the poll of Tuesday to Friday is only slightly different from the usual Wednesday to Saturday. The poll finds no change since last week on two-party preferred, with Labor retaining a lead of 54-46 from primary votes of Labor 38% (down one), Coalition 35% (steady), Greens 11% (steady), One Nation 6% (up one) and United Australia Party 3% (down one). Scott Morrison is up a point on approval to 42% and down two on disapproval to 53%, while Anthony Albanese is down three to 38% and up two to 49%. Morrison’s lead as preferred prime minister is in from 44-42 to 43-42. The sample for the poll was 1532.

Federal election minus eight days

The latest on the ongoing Katherine Deves controversy, Wednesday night’s leaders debate, prospects for the Senate and more.

Ahead of the last weekend before the big day, and with nearly a million pre-polls plus over 750,000 returned postals already in the bag:

• In her weekly column in the Age/Herald, Nika Savva writes that Liberal internal polling shows the primary votes of the six members under challenge from teal independents have “plunged to between the mid-30s and the low 40s”, and that there is “an ever-expanding list of prime Liberal real estate tilting to Labor including Bennelong, Reid, Chisholm, Higgins, Brisbane, Ryan and Leichhardt”. Savva further accuses Scott Morrison of sacrificing these seats in a bid to “harvest votes in the regions and outer suburbs” by reviving the controversy over Warringah candidate Katherine Deves’ comments on transgender issues. According to a “well-connected Liberal” quoted by Savva, the interview on Monday in which Deves recanted her earlier apology was “set up deliberately to resuscitate the issue”.

• I had a piece in Crikey on Wednesday looking at prospects for the Senate race, and in particular for the chamber’s overall balance to tip over in favour of the left. Since the “right”, i.e. the Coalition plus One Nation, won four seats in Queensland in 2019, such an outcome would require four-left, two-right results in two states. That’s unless ACT Senate candidate David Pocock is deemed part of the left and succeeds in ousting Zed Seselja, in which case it comes down to one. However, my reading of the polls is that it’s hard to see in what state the extra seat comes from, although Tasmania, from which there is next to no credible data, cannot be ruled out. More likely is that Nick Xenophon returns and/or the Jacqui Lambie Network wins a second seat, in which case an incoming Labor government will need support from either or both in addition to the Greens to win passage for contested legislation. But there remains a worst case scenario for an incoming Labor government in which the Coalition and One Nation have half the numbers between them. An expanded post on the Senate with a dedicated thread for discussion of the Senate race is something I hope to get around to one of these days.

• The third leaders’ debate of the campaign, hosted by the Seven Network on Wednesday night and moderated by Mark Riley, was rated a clear win for Anthony Albanese by undecided voters gathered by the network in four marginal seats to subject proceedings to a “pub test”. Albanese was deemed to have won by 50% compared with 25% for Scott Morrison in both Macquarie and Solomon, and by 52% in Chisholm, against 35% for Morrison. The two were tied at 44% among the sample in Hasluck.

• A claim by Fiona Martin, the Liberal member for Reid, that her Labor opponent Sally Sitou was only contesting the seat because she had been knocked back for preselection in Fowler has prompted suggestions she had confused her opponent with Tu Le, who was passed over for Labor preselection in the seat to accommodate Kristina Keneally. Martin defended herself by pointing to a news report from 2018 that Sitou was being “touted” as a possible candidate for the corresponding state seat of Cabramatta, sufficient for her to be deemed a “failed state candidate for Cabramatta”. There appears to be no actual suggestion she was ever in prospect for Fowler.

The Australian reports on data from PowerHousing Australia identifying the 20 seats in which property prices and rents have increased the most over the past two years. Notable in respect to property prices are Gilmore at number one (55.5% higher), Bass at number six (51.7%), Lyons at number eleven (49.0%), Eden-Monaro at number sixteen (47.0%) and Dobell at number nineteen (46.2%). For rents: Robertson at number four (26.7%), Bass at number seventeen (23.9%) and Solomon at number twenty (23.6%).

YouGov MRP poll (part two): Labor 80 seats, Coalition 63, others 8

All revealed from the YouGov MRP poll that was teased yesterday, suggesting Labor is on track for a fairly comfortable parliamentary majority in its own right.

The Australian now has the full suite of projected seat results from the YouGov MRP (multi-level regression with post-stratification) poll, conducted April 14 to May 7 from a sample of 18,923. Usings its data to model results based on electorates’ demography, rather than just the relatively small number of respondents from a given electorate, it projects Labor to win a clear majority of 80 seats, with the Coalition to win 63, the Greens one and others seven.

Seats projected as Labor gains are Bennelong, Lindsay, Reid and Robertson in New South Wales, Chisholm and Higgins in Victoria, Brisbane in Queensland, Pearce and Swan in Western Australia, Boothby in South Australia, and Bass in Tasmania. However, some of these are 50-50 calls on two-party preferred that are identified as leaning fractionally to either side. This includes Bennelong and Lindsay, with Longman, Ryan and Sturt identified as remaining with the Coalition, and Corangamite as a potential Labor loss.

As noted in yesterday’s post, the model also projects Kooyong and Goldstein as teal independent gains from the Liberals, and for all existing cross-benchers to retain their seats. While Brisbane is listed as a clear Labor win, its near tie on the primary vote between Labor on 29% and the Greens on 28% suggests either could be the one to ride over the Liberal National Party on 36% with the preferences of the other.

Rolling its results into state totals produces the following:

• In New South Wales, Coalition 36.4% (down from 42.5% in 2019), Labor 37.7% (up from 34.6%), Greens 9.3% (up from 8.7%), United Australia Party 3.5% (up from 3.4%) and One Nation 4.6%.

• In Victoria, Coalition 34.7% (down from 38.6%), Labor 36.5% (down from 36.8%), Greens 12.1% (up from 11.9%), United Australia Party 4.5% (up from 3.6%) and One Nation 3.8%.

• In Queensland, Coalition 39.1% (down from 43.7%), Labor 29.7% (up from 26.7%), Greens 11.9% (up from 10.3%), United Australia Party 4.5% (up from 3.5%) and One Nation 8.8%.

• In Western Australia, Coalition 38.1% (down from 45.2%), Labor 35.7% (up from 29.8%), Greens 12.7% (up from 11.6%), United Australia Party 2.3% (up from 2.0%) and One Nation 5.3%.

• In South Australia, Coalition 35.7% (down from 40.8%), Labor 38.5% (up from 35.4%), Greens 10.0% (up from 9.6%), United Australia Party 3.2% (down from 4.3%) and One Nation 5.2%.

• In Tasmania, Coalition 30.6% (down from 34.6%), Labor 32.4% (down from 33.6%), Greens 10.2% (up from 10.1%), United Australia Party 2.2% (down from 4.8%) and One Nation 4.2%.

These numbers for the most part line up reasonably well with BludgerTrack, except in South Australia where BludgerTrack has both major parties quite a bit higher on the primary vote. The Labor numbers are also lower than in BludgerTrack from the five biggest states, ranging from a 0.7% difference in New South Wales to a 2.9% difference in Western Australia.

YouGov MRP poll (part one) and more

YouGov unveils an ambitious project to project the complete result of the federal election, as more reports emerge of grim internal polling for the Liberals.

YouGov has dropped the first results from Australia’s first ever published MRP (multi-level regression with post-stratification) poll, which aims for a detailed election prediction by surveying an expansive national sample of 18,923 and using demographic modelling to project results for each electorate. Its methods are outlined in The Australian by Campbell White of YouGov and University of Sydney data science lecturer Shaun Ratcliff. The method became something of a cause celebre when it predicted the hung parliament at the United Kingdom election in 2017 that crippled the prime ministership of Theresa May, a rare success for the British polling industry in that period. However, it did less well at the subsequent election in 2019, tipping a 28-seat Conservative majority that actually came in at 80.

It seems News Corp plans on getting some bang for its presumably considerable buck here by dealing out results piecemeal, as all we have at this stage is projected results from seats in which independent candidates (plus Rebekha Sharkie of the Centre Alliance in Mayo) might be thought competitive. Since the model works by inferring how people will vote based on demography rather than in response to their specific local circumstances, I suspect it would be more robust in traditional party-based contests.

The results nonetheless point to Liberal defeats at the hands of teal independents in Kooyong (by 53-47) and Goldstein (52-48), but not in Wentworth (56-44) or Mackellar (53-47), and particularly not in North Sydney and Curtin, where independents are projected to finish a fairly distant third behind Labor. There is also no suggestion of independent Georgia Steele being competitive in Hughes, contrary to some media reports.

Indeed, the North Sydney result is interesting in showing Labor within striking distance at 53-47, consistent with the account of Liberal internal polling by Karen Middleton in the Saturday Paper and John Howard’s presence at Trent Zimmerman’s campaign launch last week. It is also encouraging for Labor in crediting them with a 53-47 lead in Boothby, and deficits of only 52-48 in each of their little-rated prospects of Page, Casey and Flinders.

The incumbent cross-benchers are projected to retain their seats, but by a narrower margin than I would have thought likely in the case of Andrew Wilkie in Clark (61-39 over Labor, a swing against him of 10%). Rebekha Sharkie is also credited with a margin of only 52-48 in Mayo, a swing against her of 3%. Zali Steggall is projected to slightly increase her 7.2% margin in Warringah, while a lead of 53-47 is projected for Helen Haines in Indi, compared with her 1.4% winning margin in 2019. The neighbouring seat of Nicholls, where independent Rob Priestly is widely thought to be a show, is not featured.


• Peter van Onselen reported on Ten News yesterday that “an unauthorised leak of Liberal polling” showed the Liberals trailing Labor by 55-45 in Bennelong, and also behind in Reid and Robertson, together with Gilmore and Parramatta, which the party has had high hopes of gaining from Labor. Van Onselen specified that this was polling conducted by the party’s fractious New South Wales branch, and was not part of the federal party’s tracking polling of 20 target seats.

Raf Epstein of the ABC in Victoria posted results of Redbridge Group polling conducted for independent candidate Monique Ryan that showed her on 32.3% of the primary vote in Kooyong, behind Josh Frydenberg on 40.5% but narrowly ahead after preferences. The Greens are said to be on 8.4%, Labor 6.7% and the United Australia Party 5.2%.

• Day one of pre-polling on Monday drew 309,769 voters. Comparisons with 2019 are complicated by the fact that the pre-poll period has been reduced from three weeks to two: 661,225 pre-poll votes were cast in the entire first week of 2019, a figure low enough to encourage each of the main parties to agree that the first of the three weeks was not worth their bother, followed by 253,684 votes on day one of week two. Antony Green is helpfully plotting the relevant statistics, which point to a substantial upswing in postal vote applications: 13.2% of total enrolment compared with 8.1% at the equivalent point in 2019.

• Michael Gunner resigned as Chief Minister of the Northern Territory yesterday, but will remain the member for his Darwin seat of Fannie Bay. The Northern Territory News rates the favourites to replace him as Nicole Manison, Deputy Chief Minister, member for Wanguri and factional colleague of Gunner in the Right; Natasha Fyles, Health Minister, member for Nightcliff and member of the Left; and, unlikely as it may seem, back-bencher Joel Bowden, Johnston MP and former AFL player for Richmond.