The Poll Bludger

A record 542,143 voters cast their ballots on the first day of early voting on Tuesday, compared with 314,496 on the first day in 2022. Four thousand of those who voted yesterday and on Tuesday across 19 seats were surveyed as part of an “exit poll” conducted by the News Corp papers, an exercise I am a little dubious about, particularly when reported at seat level from samples of barely more than 200 each. The degree of care needed to produce properly illuminating results does not seem to have been taken: the swing figures reported at seat level do not take account of redistributions, and at national level the results are aggregated and compared to the party totals from 2022, without regard to the peculiarities of the targeted electorates.

Let’s be optimistic though and say that broadly representative voting centres were chosen, and that to the extent that they were not, the problem cancels out when results are aggregated across multiple electorates (and also that the earliest early voters are representative of the whole, which is impossible to say) (UPDATE: Adrian Beaumont points out that US experience says voters earlier in the period tend to be older, which may explain some of the weak results for the Greens). In the following analysis, I have gone to the effort of basing swing calculations off redistribution-adjusted results from 2022 at pre-poll voting centres only:

• The three Greens-held Brisbane seats of Brisbane, Ryan and Griffith were all targeted, with average results of Greens 33.7%, up 2.4%; Labor 31.8%, up 5.2%; and LNP 32.8%, down 2.9%.

• Average results in the New South Wales seats of Werriwa, Gilmore, Paterson and Bennelong were Labor 43.1%, up 4.3%; Liberal 37.2%, down 2.5%; and Greens 4.8%, down 2.8%. I have excluded the part of North Sydney that was transferred to Bennelong from my baseline calculation here, due to the complication there of teal independent Kylea Tink.

• Average results in the Victorian seats of Corangamite, Chisholm, Bruce and McEwen were Labor 42.1%, up 3.4%; Coalition 40.1%, up 4.6%; and Greens 8.6%, down 3.5%.

• Boothby and Sturt averaged Labor 44.5%, up 13.3%; Liberal 44.3%, down 0.3%; and Greens 9.0%, down 4.5%.

• In the two seats with competitive teals, Goldstein and Bradfield, there was an average Liberal vote of 43.5%, up 0.6%, and an average teal vote of 32.5%, up 2.3%. In this case I did not include areas added to Goldstein in the redistribution for the baseline, as they did not have a teal candidate in 2022.

Other seats covered by the exercise were Leichhardt, McPherson, Lyons and Solomon.

There is another item of polling in the shape of a second Ipsos poll for the Daily Mail, which provides leadership ratings but not voting intention. Anthony Albanese is up three on approval since last week to 38% and steady on disapproval at 39%. The report says only that Peter Dutton has a net rating of minus 20, unchanged on last week when he recorded 27% approval and 47% disapproval. Albanese leads 46-32 as preferred prime minister, out from 44-30. The poll was conducted Friday to Sunday from a sample of 2000.

A major campaign milestone is reached today with the commencement of early voting. The third of the campaign’s leaders’ debates will be conducted from 7:30pm this evening by the Nine Network, to be moderated by A Current Affair host Ally Langdon with questions posed by Charles Croucher of Nine, Deb Knight of 2GB and Phillip Coorey of the Financial Review. The Roy Morgan poll that normally comes out on Monday will presumably be along later today. Note that there is a new post below this one on a state poll for New South Wales.

Other than that:

• The Courier-Mail reports a DemosAU poll collectively targeting the Greens-held Brisbane seats of Brisbane, Griffith and Ryan has the Liberal National Party on 36% and the Greens and Labor on 29% each, compared with 35.7% for the LNP, 30.7% for the Greens and 26.2% for Labor at the 2022 election. Were the implied swings to occur uniformly across the three seats, the likely outcome would be Labor gaining Brisbane and the Greens retaining Ryan and Griffith. The poll was conducted “mid-April” from a sample of 1087. UPDATE: The poll further includes a finding of 56-44 two-party preferred between Labor and the LNP and 55-45 between the Greens and the LNP.

• The Financial Review reports a Nationals source as being “pessimistic about Cowper”, where Pat Conaghan is under pressure from teal independent Caz Heise, but believing the party to be ahead in Calare, where former Nationals member Andrew Gee seeks re-election as an independent and teal independent Kate Hook is again in the field after performing strongly in 2022. In the latter case, the source says “if only one independent had run, that independent would have won”.

• The News Corp papers report that One Nation is changing its how-to-vote cards in Hunter and Paterson to elevate the Coalition, having initially favoured right-wing minor parties and independents, and is reviewing the situation elsewhere. Hanson’s chief-of-staff, James Ashby, is quoted saying the party was “was restructuring its preferences in seats where the Trumpet of Patriots has put the chance of a conservative candidate’s success ‘at risk’” – though this is not in fact the case in Hunter or Paterson, where Palmer’s party has Labor last in keeping with its approach of directing against sitting members (in a related development, its candidate for Flinders is telling voters to put him last out of displeasure at his how-to-vote card favouring teal independent Ben Smith). The only seats I can think of where One Nation does not already have the Coalition ahead of everyone it could conceivably lose to are Kennedy, where Bob Katter is presumably safe, and Monash, where a second preference is recommended for Liberal-turned-independent Russell Broadbent, for whom Trumpet of Patriots has made an exception by placing him third behind another independent.

The Australian reports the weekly campaign Newspoll has Labor’s two-party lead unchanged at 52-48, from primary votes of Labor 34% (up one), Coalition 35% (steady), Greens 12% (steady) and One Nation 7% (down one). Leadership ratings are a mixed bag for Anthony Albanese: his approval rating is down two to 43% with disapproval up three to 52%, but his lead on preferred prime minister widens from 49-38 to 52-36. Peter Dutton is down two on approval to 35% and up one on disapproval to 57%. The poll also finds the Coalition favoured 34-29 on growing the economy and 35-23 on defence, but Labor favoured 31-28 on the cost of living, 42-22 on health, 33-26 on lowering taxes, 29-24 on helping first home buyers and 39-32 on “dealing with uncertainty caused by Donald Trump”. The poll was conducted Monday to Thursday from a sample of 1263.

Also, Nine Newspapers has further results from Resolve Strategic’s poll of last week as to what respondents’ “biggest hesitations or concerns” are in voting for either major party. Labor’s are “a lack of action on the cost of living” and economic management, while the Coalition’s is “Peter Dutton’s personality as leader”.

Further recent posts you might care to take note of: one analysing the how-to-vote cards and their likely impact, one summarising recent reports of internal polling and a guest post from Adrian Beaumont on next week’s Canadian election.

With early voting to begin on Tuesday, details of how-to-vote cards are starting to accumulate. So far as the Senate is concerned, the ABC is making life easy by gathering them in one place, while the eternally patient among you can view lower house how-to-vote cards for Labor candidates and Liberal candidates one by one on individual candidate pages. I tend to consider the subject not worth much of the news space devoted to it, because only major party voters follow them in substantial numbers, and in most cases they make the final count and their preferences are thus not distributed — though the number of exceptions is of course much higher than it used to be.

The Victorian Electoral Commission provided a rare insight into the matter when it made available full ballot paper data for seven seats at the 2022 state election, which is normally available only for upper house counts and in New South Wales, where its applicability is limited by optional preferential voting. The rate at which Labor votes exactly followed the how-to-vote card was consistently around 30% (helpfully, there were large numbers of candidates in these seats, so few are likely to arrived upon the requisite order by happenstance). The rate of adherence among Liberal voters varied widely according to the amount of effort the party was putting into a given seat — 57.0% in Brighton, where it was fighting off a teal independent, and 53.9% in Hawthorn, where it was successfully challenging a Labor incumbent, but only 29.4% in Preston, where the fight was between Labor and the Greens.

As for minor parties, whether the Greens actively preference Labor (as they are doing in every seat at this election) or offer no recommendation (which is as far as they ever go in repudiating Labor) makes about five points’ worth of difference to the percentage received by Labor, which is typically upwards of 80%. The impact is likely even less for smaller parties, which lack the volunteer base needed to disseminate how-to-vote cards on a large scale at polling places. As such, it would not pay to put too much weight on the decision by Clive Palmer’s Trumpet of Patriots to direct preferences against most sitting members. It was, indeed, news to me that his United Australia Party had done much the same in 2022, a fact related by Kos Samaras of RedBridge Group in a report in the News Corp papers yesterday. Out of 108 seats where the relevant data exists, I calculate that Labor got 41.7% of UAP preferences in seats where they had sitting members in 2022 and 35.5% where the Coalition did – the opposite of what would have been expected if their voters had been following the how-to-vote card in non-trivial numbers.

The same approach was adopted by One Nation during its early heyday at the Western Australian state election in 2001, in this case involving a party with genuine mass support and not simply the artefact of a multi-million dollar advertising campaign. This is sometimes said to have been a key factor in Labor’s unexpectedly clear-cut win at the election over the two-term Coalition government of Richard Court. To test this, I have just blown the dust of my print copy of the statistical returns from that elections and identified 24 seats where One Nation candidates were the last excluded in the preference count and their votes distributed between Labor and the Coalition. The distinction in this case was at least in the right direction, but nonetheless very modest — Labor averaged 53.2% of the preference transfers in Coalition-held seats and 49.7% in Labor-held seats. I assumed there would be a relationship between Labor’s vote share in a given seat and the strength of its preference flow — the likeliest explanation for the counter-intuitive finding just noted for the United Australia Party — but there was in fact little evidence of this.

So with all that in mind, the following points worth noting have emerged from news reportage, in lieu of my own lack of motivation to investigate the matter too deeply myself:

• Macnamara is the only seat where Labor is not directing preferences, which it is doing there at the initiative of incumbent Josh Burns in recognition of the local Jewish community’s concerns about the Greens. It is entirely possible that Burns will run third, in which case Labor preferences will come into play in the race between the Liberals and the Greens. Had it played out that way in 2022, the Greens would have won handily on a flood of Labor preferences, just as Burns easily defeated the Liberals after receiving fully 90.25% of the preferences from the third-placed Greens. Macnamara partly corresponds with the state seat of Brighton, where the behaviour of Labor preferences in 2022 was noted above. The 10,126 Labor votes in the seat included 3231 who followed the card, 5468 who favoured the Greens over the Liberals independently, and 1427 who had the Liberals ahead of the Greens. If those who followed the how-to-vote card had split in the same proportion as those who did, the flow of Labor preferences to the Greens would have fallen from 85.9% to 79.3%. Making a reasonable assumption that the Labor primary vote in the seat will be about 30%, this suggests Labor’s open ticket could contribute about 2% to a swing of upwards of 10% that the Liberals would need if it came down to them and the Greens.

• Next door in Goldstein, Liberal candidate Tim Wilson has teal incumbent Zoe Wilson second last ahead of the Greens, leaving her behind One Nation as well as Labor. The Age notes that Wilson said in 2019 that he had “a longstanding view that we should put One Nation and their despicable acolytes last”. While this may be of interest on an optical level, Wilson’s preferences are of little electoral consequence as he seems assured of making the final count.

• Conversely, the Liberals in Western Australia have put teal independent Kate Hulett ahead of Labor in third place in Fremantle (though here too they favour One Nation, who are second), a fairly remarkable turnaround on the March state election when they had her last but for the Greens.

• Labor are favouring Liberal over the Nationals in the three-cornered contest in the new Western Australian seat of Bullwinkel. However, an apparent improvement in Labor’s fortunes suggests the issue here is likely to be whether Nationals candidate Mia Davies or Liberal candidate Matt Moran makes it to the final count against Labor’s Trish Cook.

• The Liberals may have scotched the chances of independent Peter George by putting him behind Labor in Franklin. George is a former ABC journalist running with support from Climate 200 and in opposition to salmon farming, none of which would have endeared him to the Liberals. With 19-year-old Greens candidate Owen Fitzgerald having conceded he would be unable to sit in parliament due to a dual citizenship, Matthew Denholm of The Australian reported the party was expected to throw its support behind George, who had already been endorsed by former party leader Bob Brown.

• Liberal-turned-independent Russell Broadbent has interestingly opted to place teal independent Deb Leonard all-but-last in Monash. Leonard was presumably counting on getting ahead of him and benefiting from the strong preference flows that typically apply between independents.

• Paul Karp of the Financial Review reports Ziad Basyouny and Ahmed Ouf, who are respectively challenging Tony Burke in Watson and Jason Clare in Blaxland with support from The Muslim Vote, are directing preferences to Labor ahead of Liberal “after failing to strike a preference deal with the opposition”. The Liberal how-to-vote cards have them both dead last, which does not surprise, but will nonetheless make life difficult for them. A review article in the Sun-Herald today says Labor believes it could face swings in the seats of “more than 6%”, with Watson of greater concern.