Preference flows and by-elections (open thread)

A look at preference flow data from the 2019 and 2022 elections, and the latest on looming by-elections in the Northern Territory, Tasmania and (sort of) Western Australia.

Something I really should have noted in last week’s post is that the Australian Electoral Commission has now published two-candidate preferred preference flow data from the election, showing how minor party and independent preferences flowed between Labor and the Coalition. The table below shows how Labor’s share increased for the four biggest minor parties and independents collectively (and also its fraction decrease for “others”) from the last election to this and, in the final column, how much difference each made to Labor’s total share of two-party preferred, which was 52.13%.

Note that the third column compares how many preference Labor received with how many they would have if preference flows had been last time, which is not the same thing as how many preferences they received. Labor in fact got nearly 2% more two-party vote share in the form of Greens preferences at this election because the Greens primary vote was nearly 2% higher this time.

State and territory by-election:

• Six candidates for the August 20 by-election in the Northern Territory seat of Fannie Bay, in ballot paper order: Brent Potter, described in a report as a “government adviser, army veteran and father of four”, for Labor; independent George Mamouzellos; independent Raj Samson Rajwin, who was a Senate candidate for the United Australia Party; Jonathan Parry of the Greens; independent Leah Potter; and Ben Hosking, “small business owner and former police officer”, for the Country Liberals.

• Following the resignation of Labor member Jo Siejka, a by-election will be held for the Tasmanian Legislative Council seat of Pembroke on September 10. Siejka defeated a Liberal candidate by 8.65% to win the eastern Hobart seat at the periodic election in 2019. There will also be a recount of 2021 election ballots in Franklin to determine which of the three unelected Liberals will replace Jacquie Petrusma following her resignation announcement a fortnight ago. As Kevin Bonham explains, the order of probability runs Bec Enders, Dean Young and James Walker.

• Still no sign of a date for Western Australia’s North West Central by-election.

Liberal with the truth

A look back on what internal Liberal polling appeared to be saying ahead of the May election, and the related matter of the Katherine Deves controversy.

Last week I took a big picture look at how the main public pollsters performed in their immediate pre-election polling. Today I offer a necessarily incomplete account of the only partly knowable subject of internal party polling – specifically that of the Liberal Party, and how it played out against the backdrop of bitter conflict over its strategy of the campaign of pursuing culturally conservative constituencies at a time when those under threat from the teal independents needed every socially liberal vote they could get.

Much of this story relates to the controversy surrounding Warringah candidate Katherine Deves, which Scott Morrison appeared to consider the key to unlocking enough Labor-held seats in the outer suburbs and regions to balance defeats in inner metropolitan seats, at least to the extent of allowing him to hold on to power in a minority government. The notion that this strategy might have been hitting its mark was not the exclusive preserve of Liberal Party optimists. Shortly after the controversy first emerged early in the campaign, Phillip Coorey of the Financial Review wrote that “in the suburbs, the regions and the religious communities, the government – and Labor – believes the Deves issue is going gangbusters in Scott Morrison’s favour, messy as it may be”. A week later, Chris Uhlmann cited a Labor strategist in the Age/Herald who believed the issue was playing “90/10 in Deves’ favour” in the suburbs and the regions. Cameron Milner, a former Queensland Labor state secretary now all too comfortable in a new perch on The Australian’s op-ed page, described the Liberals’ exploitation of the controversy as “brilliant foghorn politics” that would yield a bumper crop of Hanson and Palmer preferences.

When Deves recanted her initial apology for her comments a fortnight out from the election, Niki Savva in the Age/Herald cited a Liberal source saying this had been “set up deliberately to resuscitate the issue”. Complicating the notion of a divide between what Uhlmann called “the inner-city bubble” and mainstream opinion further afield, Lanai Scarr of The West Australian reported that some were “even tipping Deves could pull off her own ‘miracle’ win and insulate other conservative electorates nationally in the process”, potentially saving the Liberals in such difficult contests as the Perth seat of Swan.

Needless to say, none of this looks terribly prescient now that the election’s unknowns are known. The possibility that the Liberals were acting on faulty intelligence is intriguingly raised by a report from Peter van Onselen on Ten News four days out from the election, which related that Liberal polling had Katherine Deves trailing Zali Steggall by only 53-47 – quite a lot closer than Steggall’s eventual winning margin of 61-39. Lest it be thought that this was some kind of Liberal Party psyop, it formed part of a batch of polling that was otherwise disastrous for the Liberals, with two-party preferred scores inclusive of an uncommitted component showing them trailing 50-43 in Bennelong and 50-41 in Parramatta (worse than their actual losing margins) and 49-48 in Reid (better), with particularly large deficits among women.

This happened to be the second batch of Liberal seat polling that van Onselen had been able to report late in the campaign, the first of which emerged as a bone of contention post-election in the party’s deepening culture war over the teal independent seats and whether they should be cut loose in favour of a more populist approach that took its cues from Donald Trump. This had the Liberal primary vote at 43% in Kooyong, 37% in Goldstein and 44% in Higgins, which bore up quite well against respective final results of 42.7%, 40.4% and 40.7%. Shortly after the election, Sharri Markson of The Australian recorded the following reaction to the leak inside the Liberal camp:

Senior Liberal figures scratched their heads, wondering where it had originated. The precise numbers did not reflect what was emanating from the party’s official poster, Crosby Textor. An internal probe discovered that (Senator Andrew) Bragg had submitted expenses to the NSW Liberal division of about $35,000 to $40,000 to conduct his own alternative polling in many NSW seats. There is no suggestion that Bragg leaked the polling to van Onselen, which he denies. It was not in his interest to depress the prospects of candidates he was fighting hard to help win. It’s not even clear whether the polling Bragg commissioned was the same polling broadcast on Ten. However, Morrison’s team believed it was.

Bragg had circulated the polling he commissioned to many Liberals – an action one source described as “sloppy” – and the suggestion is a recipient subsequently leaked it to the media. Questioned about the research for this article, Bragg admits he commissioned alternate polling and is scathing about the way Liberal headquarters and Crosby Textor treat Liberal candidates, who he says are kept in the dark about how they are faring.

“The Liberal Party and Crosby Textor treat the candidates like absolute shit and don’t give them the information they need,” Bragg says. “The candidates, who are often members of parliament, all they are given is a phone briefing and if they’re lucky they might get a piece of paper. Crosby Textor omit key things like the favourability of the leader because they’re worried that will leak to the media. If you know the party leader is massively unpopular you’ll differentiate so you can hang onto the seat. But if you’re not told that how are you supposed to know? It’s conflicts galore.”

Echoes of Bragg’s criticism were to be heard outside the tent from Kos Samaras, who as one of the principals of the Redbridge Group had provided polling and strategic advice to Climate 200 (with which, as per the disclosure notice at the bottom of this site’s sidebar, I was involved myself):

Why did the teals win? Many reasons. But at the centre of the campaign was an absolute commitment to the data. There were no games with what the internal polling said. There were no favourites shown, whereby resources are sent into a seat, even though the polling painted a different picture … (The Liberals) poured resouces into one seat, Kooyong, at the expense of others, even though their data was indeed showing a grim picture. That picture of course was never told, as the constant backgrounding into the media was akin to a story-telling session, skunk drunk, at a pub. The Liberal decision-making was riddled with bias and subjectivity, fuelled by an internal factional structure that made it impossible for data to be utilised correctly.

If early indications are anything to go by, the tension between the Liberal Party’s determination to tack to the right on cultural issues and electoral imperatives to win the favour of more liberally minded voters could be set to play out again at the Victorian state election in November. Stay tuned.

The polls and the sum of the parts

An overdue appraisal of the pollsters’ performance at the May federal election.

The 2022 federal election was a much happier experience for the polling industry than 2019, with each of five pollster producing election eve primary vote numbers broadly suggestive of the actual result. However, there was a collective error in favour of Labor, whose actual primary vote came in 2.3% below the pollster consensus while the Coalition landed 0.4% higher. While not at the standard Australian consumers had come to expect before 2019, such errors were fairly moderate by historic standards, particularly in the international context.

The pollsters in the national voting intention game before the election: Newspoll, conducted by YouGov for The Australian; Ipsos, for the Financial Review; Resolve Strategic, for the Nine Newspapers; Essential Research, for The Guardian; and Roy Morgan, which tacks voting intention on to its market research survey for its own amusement and, in the pre-election period at least, regularly published its results. The charts below show their final pre-election poll results for the Coalition, Labor and the Greens as black dots that lie at the centre of the span of their margins of error, with the relevant party’s actual result shown as a thick vertical line.

In the case of the three who are members of the Australian Polling Council (Newspoll, Ipsos and Essential Research), what is shown are the pollsters’ effective margins of error, which account for the fact that their results are weighted to emphasise or de-emphasise demographic cohorts who are over-represented or under-represented in their samples. Since Resolve Strategic and Roy Morgan don’t provide this detail, the margins of error have been calculated from their raw sample sizes, making them somewhat smaller than they would be otherwise.

Some caveats must be applied here: the final survey periods ranged from over a week before the election in the case of Roy Morgan (from May 9 to 15) to the course of the final week (Saturday to Wednesday from Ipsos, Thursday to Thursday from Newspoll), and pollsters may always plead they were caught out by late shifts in voting intention that it was beyond their power to foresee. There’s also a certain injustice in evaluating a pollster’s performance entirely by its pre-election poll, as the unavoidable randomness of the exercise means an element of luck is involved in who gets the honours. The best we can do is keep that in mind in the analysis that follows.

Essential Research did not have a great result, having systematically understated the non-major party vote (results here exclude a 7% undecided component). Other than that, three results are clearly outside the margins of error: Newspoll’s 36% and Ipsos’s 35.8% (after exclusion of 5% uncommitted) for Labor, whose actual result was 32.6%, and Resolve Strategic’s 14% for the Greens, whose result was 12.3%. None of these final poll results look like outliers for the pollster concerned. Newspoll’s four previous polls, which were published weekly during the campaign period, had Labor one to three points higher than the final result; Ipsos’s polling consistently had Labor clear of their actual result even without excluding their uncommitted component; and Essential Research’s numbers bounced a few points at a time within a consistent range throughout 2022.

Resolve Strategic’s evident inflation of Greens support was peculiar to its last two polls, which had it at 15% and 14% compared with 10% to 12% in its earlier polling. It should be noted here that margins of error are tighter for parties with lower vote shares, and that it’s conceivable that bad luck with rounding put the final poll outside the error margin at 14% rather than within it at 13%. State breakdowns show the inflation of the Greens vote arose from New South Wales and Queensland, which was respectively balanced by unduly low results for Labor and the Coalition. The latter was a peculiarity of the final poll, which had the Coalition at 31%, down from 41% at the previous poll a fortnight earlier and comparing with an actual result of 39.6%.

Essential Research’s issues were consistent across the board, landing too high for the Coalition and Labor and too low for the Greens in each state. The Victorian numbers were particularly far off the mark, perhaps reflecting their unusually high reading of 11% undecided in this state. It should be noted that the results for Western Australia and South Australia are three-week rolling averages, which means they ought not be regarded as strictly election eve results.

By process of elimination, it might be thought that the honours belong to Roy Morgan, whose reputation never really recovered from a series of poor federal election performances two decades ago. However, state breakdowns suggest a certain amount of luck was involved here, in that every one of its results during 2022 put Labor much higher in Victoria than they managed at the election. The element of luck was that the least askew of these results came in the pre-election poll, which had Labor’s lead in Victoria at a relatively narrow 57-43 compared with an actual result of 54.6-45.4. UPDATE: Adrian Beaumont also points out that they always had the United Australia Party implausibly low, the final result coming in at 1% compared with an election result of 4.1%, and this was matched by an excessive reading for independents/others.

Unfortunately there are state breakdowns available for Newspoll, for which we disappointingly did not see any of the usual aggregated breakdowns during the campaign period, and Ipsos, which did not maintain the finely detailed breakdowns it offered for its first few polls a the business end of the election. I’ll have more to say though about pollster performance at the election during quiet moments over the weeks or months to come, so stay tuned.

Final TPP: 52.1-47.9 to Labor

The final national two-party preferred total from the election is now settled, give or take a few hundred votes.

Click here for full display of House of Representatives election results.

The Australian Electoral Commission has completed its Labor-versus-Coalition two-party preferred counts for what it calls “non-classic” contests, namely those in which independent or minor party candidates made it to the final exclusion in the preference distribution – notwithstanding two minor errors which, by my estimate, will add an all-important 0.01% to Labor’s total when corrected. That will produce a final result of 52.14% and Coalition 47.86%, for a swing to Labor from the 2019 election of 3.67%. I have revised the entry guide to my results feature, linked to above, which had previously been projecting 51.9-48.1 to Labor based on preference estimates.

As well as a national figure, the two-party preferred counts have also yielded a number of interesting results at electorate level. Mayo tipped over to Labor for the first time by 1.6%, the Liberal margin having shrunk over the three previous election from 12.5% in 2013 to 5.4% when Rebekha Sharkie won the seat in 2016 to 2.5% in 2019. Labor also won the two-party vote in Ryan by 2.4% after a swing of 8.4%, in Brisbane by 4.4% after a swing of 9.3%, and in Griffith by 11.1% after a swing of 8.2%, for all the good it did third-placed Terri Butler.

In the teal independent seats, Liberal two-party margins over Labor were reduced to 1.3% in North Sydney by a swing of 8.0%; 8.6% in Mackellar by a swing of 4.6%; 5.9% in Wentworth after a swing of 3.9%; 4.8% in Goldstein after a swing of 3.0%; 4.2% in Kooyong after a swing of 2.2%; and 5.6% in Curtin after a swing of 8.3%. Despite/because of the absence/presence of Tony Abbott/Katherine Deves, the two-party swing to Labor in Warringah was only 0.7%, reducing the Liberal margin to 1.4% after a 9.0% swing in 2019 reduced it to a then-unprecedented 2.1%. In Bradfield, where Nicolette Boele rode the teal independent wave to make the final count and reduce Liberal member Paul Fletcher to a winning margin of 4.2%, Labor picked up a 10.0% two-party swing to reduce the margin on that measure to 6.5%. The Liberals’ previous lowest winning margin over Labor in Bradfield was 13.5% in 2007, when it was held by Brendan Nelson.

On the other side of the coin, Kristina Keneally suffered an 8.3% two-party swing in the course of losing Fowler to independent Dai Le, reducing the margin there to 5.7%, breaking a previous record in that seat of 8.8% in 2010. In what might have been seen as a warning to Labor, the 2010 result came off the back of a 13.8% swing after Labor used the seat to accommodate Chris Hayes. Hayes’ existing seat of Werriwa had in turn been used to accommodate Left faction powerbroker Laurie Ferguson, whose existing seat of Reid had been made a lot less safe after being effectively merged with its abolished neighbour, Lowe.

The AEC has also published a zip file of full preference distributions for all 151 seats, pending the data entry required to make them available in a more accessible format on the site. Groom was a late addition to the AEC’s non-classic contests lists when the preference distribution made it apparent that independent candidate Suzie Holt had received enough preferences from minor party and independent candidates to close a gap of 18.73% to 8.26% on the primary vote. The Liberal National Party member, Garth Hamilton, ended up within a winning margin over Holt of 6.89% at the final count. If you observe the booth results map at the bottom of my results page, you can observe that support for the two candidates was finely balanced in the electorate’s dominant population centre of Toowoomba, with rural and small town votes tipping the balance in favour of Hamilton.

A lot of explaining to do

An overview of an Australian National University study based on surveys of the same voters at the start of the election campaign and immediately after it.

The Australian National University’s Centre for Social Research and Methods has published a study entitled Explaining the 2022 Australian Federal Election Results, which seeks to live up to its name by analysing two surveys, one conducted from April 11 (the day after the election was called) to April 24, the other from May 23 (two days after the election) June 5. The sample for the latter was 3556, of whom 3350 completed the former survey, providing an invaluable insight into how voting intention and related attitudes changed over the course of the campaign. Excitingly, the data from the May survey “will be available for download through the Australian Data Archive at the end of June 2022” – the April survey data is already available here.

The analysis uses a statistical model to analyse relationships between various voter characteristics and vote choice, together with straightforward breakdowns of vote choice by various demographic variables. Notable findings from the report:

• As well as associating positively with conservative voting, age was significant in that younger Coalition voters from 2019 were more likely to defect.

• Consistent with what polling breakdowns were showing, women as well as younger voters voted further to the left than men, which in both cases manifested in lower support for the Coalition, higher support for the Greens and little difference for Labor.

• Voters with university degrees likewise defected from the Coalition in greater proportion than those with less education. Income did not have a statistically significant impact on vote switching independent of education, and being in the bottom fifth on household income continued to associate positively with voting Labor (and also with voting for “others”, namely everyone but Coalition, Labor and Greens, nearly half of which was One Nation and United Australia Party).

• Of those who changed their minds between the start of the campaign and election day, the largest movement was from Labor to the Greens, accounting for 4.0% of all voters, with 2.4% going the other way. The difference in flows between the two major parties fell well short of statistical significance, with 3.0% going from Coalition to Labor and 2.0% vice-versa. A similar exercise from the 2019 election found the biggest flow during that campaign to be from “others” to the Coalition.

• The statistical model found that speaking a language other than English associated negatively with non-Coalition voting. While this may seem counter-intuitive, the model measured the effect independently of many other explanatory variables, notably those relating to income and living in “the most disadvantaged areas”. So while non-English speaking voters were undoubtedly less likely to vote Coalition in aggregate, they seem more likely to have done so than English-speaking voters with the same demographic profile. It would be interesting to have seen the results with Chinese language speakers, who are typically more conservative but seem to have swung heavily to Labor at this election, considered separately from those of other backgrounds.

• An explanatory variable confusingly identified as “lives in another capital city” recorded a strong negative association with voting for Labor and the Greens. It would make all sorts of sense if this should in fact say “lives outside of a capital city”.

Senate results finalised

One last indignity for the Coalition as it loses its third Victorian Senate seat to the United Australia Party, while results in New South Wales and Western Australia play according to script.

All three outstanding Senate results were concluded today, leaving us with a new Senate of 32 Coalition members (including six Nationals, two elected on Queensland’s Liberal National Party ticket at one from the Country Liberal Party in the Northern Territory), Labor 26, Greens 12, One Nation two, Jacqui Lambie Network two, United Australia Party one and one independent. Today’s results in descending order of interest:

• Clive Palmer has not emerged completely empty-handed after Ralph Babet of the United Australia Party won the last seat in Victoria, unseating Liberal incumbent Greg Mirabella for a result of Labor two, Coalition two, Greens one and UAP one. I originally thought Mirabella likely to win based on how preferences flowed in 2019, but came to think Babet more likely to after seeing how preferences were flowing in other states. Consistent with the model I produced after using preference flows from the Queensland result, Mirabella actually fell to third place behind both Babet and Labor’s third candidate, Casey Nunn, after One Nation excluded, with Babet on 9.2%, Nunn on 7.9% and Mirabella on 7.2%. Babet’s lead then widened after the exclusion of Mirabella to 11.9% to 9.8%.

• The result in Western Australia was three Labor, two Liberal and one Greens, which always seemed highly probable, although I entertained vague notions towards the end that tight right-wing preference flows might result in Labor’s third seat going to One Nation instead. At the second last exclusion, third Liberal incumbent Ben Small went out with 6.5% of the vote to 10.2% for Labor’s Fatima Payman and 8.7% for Paul Filing of One Nation, but the distribution of his preferences left the gap between Payman and Filing essentially unchanged at 12.2% to 10.6%.

• As always seemed clear, the result in New South Wales was Coalition three, Labor two and Greens one. After the exclusion of Legalise Cannabis, incumbent third Liberal Jim Molan was elected with 12.3% of the vote, ahead of One Nation’s Kate McCulloch on 9.9%.

And for the sake of completeness, a summary of the earlier results:

Queensland returned two Labor, two Liberal National, one Greens and one One Nation Senator, the latter being Pauline Hanson, who won out over the incumbent third candidate on the LNP ticket, Amanda Stoker. When the exclusion of Legalise Cannabis left three remaining candidates chasing two seats, Stoker held 10.3% of the vote against 14.2% for Pauline Hanson and 13.9% for Labor’s Anthony Chisholm, who were duly elected in that order. Hanson substantially outperformed my projection based on 2019 preference flows, which only got her to 12.1% compared with 14.2% for Chisholm, with Stoker on 10.9%.

South Australia returned three Liberals, two Labor and one Green. The third Liberal, Kerrynne Liddle, finished with 12.4%, ahead of Jennifer Game of One Nation on 9.5%. At the previous count, third Labor candidate Trimann Gill was excluded with 8.0% to Liddle’s 9.5% and Game’s 8.7%, though Liddle would have won even if Gill had stayed ahead of Game.

• The result in Tasmania was two Labor, two Liberal, one Green and one Jacqui Lambie Network. Tammi Tyrell of the JLN was elected with a full quota after the exclusion of the incumbent third Liberal, Eric Abetz, whose preferences pushed her to 14.9% ahead of One Nation’s Steve Mav on 8.9%, with another 1.2% of the Abetz vote remaining undistributed at the point where Tyrell passed the threshold of a quota. Abetz was excluded with 6.6% in the previous round, behind Tyrell on 12.7% and Mav on 7.2%. Abetz’s campaign for below the line votes had little impact: he scored 4.27%, whereas the second Liberal, Wendy Askew, had 13.2% after the distribution of the surplus of the first Liberal, Jonathan Duniam.

• In the Australian Capital Territory, independent David Pocock was elected at the expense of Liberal incumbent Zed Seselja with 36.3% at the final count to Seselja’s 28.6%. Labor’s 33.37% share of the vote, fractionally over a quota, ensured there was no chance that Katy Gallagher was going to lose her seat to Pocock rather than Seselja, as one campaign poll suggested she might.

• The Northern Territory had an orthodox result of one seat for Labor and one for the Country Liberal Party, who fell just short of a quota on the primary vote with 32.97% and 31.70% respectively without any other party doing well enough to be threatening.

Late counting: more Senate buttons pressed

Pauline Hanson returned in Queensland, leaving on the Victorian and New South Wales results to be confirmed.

Saturday, June 18

Some discussion on Twitter in the wake of the Queensland result inspired me to consider the possibility that right-wing preference flows might yet deprive Labor of a third seat in Western Australia, a result that would appreciably weaken the new government’s hand in the Senate.

The answer is that they could if preferences indeed behave as they did in Queensland, which I have illustrated with one of two new sheets on my Senate projection spreadsheet, identified as “WA — Qld prefs”. This is because a strong flow of preferences to One Nation would give them an even chance of passing the third Liberal to make it to the final count, at which point One Nation would take the last seat on Liberal preferences. However, since One Nation got less than half the vote share in WA that they did in Queensland, it seems intuitively likely that they will also get a weaker flow. I have also included the same exercise using preference flows from South Australia, identified as “WA — SA prefs”, where the parties’ vote shares more closely resembled Western Australia. This too suggests One Nation has a strong chance of making the final count, but in this case they would fall well short of taking the seat from Labor.

It should be noted here that there is a dramatic difference between the strong preference flow to One Nation upon the LNP’s exclusion in the Queensland example and the weak flow upon the Liberals’ exclusion in South Australia — partly for the reason just noted, but also because the Liberal how-to-vote card in South Australia did not recommend a preference to One Nation, whereas the LNP card had them second. Western Australia is an intermediate case in this respect, since the Liberal card directed preferences to One Nation in nine seats but not in the other six. But in the event that minor party preferences flowed to One Nation only as strongly as they did in South Australia — and assuming this was still enough to put them ahead of the Liberals and into the final count — the flow to One Nation upon the Liberals’ exclusion would have to be fully as strong as in Queensland for them to then overhaul Labor, which hardly seems likely.

Friday, June 17

The fifth Senate button press has just been conducted in Queensland, and it’s the first that relates to a result that I considered in any way in doubt. As I thought highly probable but not quite certain, Pauline Hanson has held her seat at the expense of Amanda Stoker, the incumbent third candidate on the Liberal National Party ticket. The other seats have gone two LNP (James McGrath and Matt Canavan), two Labor (Murray Watt and Anthony Chisholm) and one Greens (Penny Allman-Payne). I’ll have more to offer on this when the preference distribution and ballot paper data are published.

Most interesting of those still to come is Victoria, where the last seat promises to be a very tight race between the incumbent third Liberal, Greg Mirabella, and Ralph Babet of the United Australia Party (a lot more on that here), which the AEC announces will happen at 10am on Monday. New South Wales, which looks certain to be three Coalition, two Labor and one Greens, will be conducted half-an-hour earlier. Still no word yet on Western Australia, which looks like Labor three, Coalition two and Greens one. No word yet on when those might be expected.

UPDATE: The Queensland Senate distribution is now up on the AEC site. It turned out that Amanda Stoker was not seriously in contention: when the exclusion of Legalise Cannabis left three remaining candidates chasing two seats, Stoker held 10.3% of the vote against 14.2% for Pauline Hanson and 13.9% for Labor’s Anthony Chisholm, who were duly elected in that order. Hanson substantially outperformed my projection based on 2019 preference flows, which only got her to 12.1% compared with 14.2% for Chisholm, with Stoker on 10.9%. This is perhaps consistent with what was seen in South Australia, where preferences among right-wing minor parties were tighter this time with less leakage to the Coalition. As noted in the previous post, this suggests Ralph Babet has a solid chance of poaching the last seat in Victoria from Greg Mirabella, contrary to what my model was suggesting.

Late counting: first Senate buttons pressed

Final resolutions of the ACT and NT Senate counts imminent, as the AEC also gets to work sorting out the final two-party preferred result.

Thursday, June 16

UPDATE: Tasmanian Senate result confirmed.

The next cab off the rank for the Senate is Tasmania at 3pm today, which should confirm a result of two each for Liberal and Labor and one each for the Greens and the Jacqui Lambie Network. Yesterday’s result in South Australia (see below) did not surprise, but my analysis of the ballot paper data has: the United Australia Party did a lot better on preferences than in 2019, sufficient to suggest that their candidate in Victoria, Ralph Babet, is well in the hunt for a final seat that I had hitherto thought most likely to go to the third Liberal, Greg Mirabella.

On this spreadsheet you will find my determination of how the various minor players’ preferences split between Labor, Liberal, Greens and UAP for the Senate in South Australia at both this election and in 2019. Most notably, the UAP only got 30.2% of One Nation preferences ahead of the other three in 2019, but this time they got 58.3%. The same pattern is evident in lesser degree among most other parties, particularly on the right. The one glaring exception is the Liberal Democrats, the results for which show how important ballot paper order is for this particular party. In 2019 they were well to the right of the ballot paper, polled 0.67%, and sent only 35.4% to the Liberals on four-party preferred. This time they drew the column A, got 2.20% and sent 61.2% to the Liberals, who were right nearby in column C.

To get a sense of what this might mean for Victoria, my Senate projection spreadsheet now contains a new sheet called “Vic 2”, which as much as possible replaces the preference data from Victoria in 2019 with the new results in South Australia. Note that I have left the distribution for the LDP undisturbed rather than swing it dramatically to the Liberals as per the South Australian result. If Antony Green is correct in his assessment that “lockdowns and changes in party registration rules” might mean more LDP preferences for the UAP at the expense of the Liberals, this assessment will actually be conservative with respect to their chances of overhauling Mirabella.

Whereas the existing projection gives the UAP only a 0.2% boost over the Coalition when One Nation preferences are distributed, the new one makes it 2.1% in “Scenario 1”, where Legalise Cannabis are excluded before One Nation, those two parties being closely matched at the previous exclusion. Competition from Legalise Cannabis for One Nation preferences in “Scenario 2” reduces this to 1.6%, but the difference comes back to them when Legalise Cannabis is excluded in later counts.

There are, in effect, three scenarios laid out here, depending on who drops out at close exclusions in the final stages. As noted, one involves Legalise Cannabis dropping out before One Nation, another vice-versa. On the first of these, I now have Mirabella dropping out before both UAP and Labor, and his preferences then deciding the result for the UAP. The second sets up another tight exclusion at the next round, with either Legalise Cannabis or the third Labor candidate going out next.

The former case, Scenario 2a, is essentially the same as Scenario 1, the only difference being the order of exclusion between Legalise Cannabis and One Nation. But in Scenario 2b, Labor’s exclusion unlocks what seems to me a surprisingly strong flow of preferences to the Liberals, precious few for the UAP, and not enough for Legalise Cannabis. On this scenario, Mirabella makes it over the line — just.

Wednesday, June 15

As noted below, the button will be pressed on the South Australian Senate result at 3pm today.

UPDATE: The result in South Australia, as expected, was 1. Simon Birmingham (Liberal), 2. Penny Wong (Labor), 3. Andrew McLachlan (Liberal), 4. Don Farrell (Labor), 5. Barbara Pocock (Greens), 6. Kerrynne Liddle (Liberal). The full preference distribution is here. Liddle ended short of a full quota at the final count with 140,008 votes (12.4%), ahead of Jennifer Game of One Nation on 107,672 (9.5%). At the previous count, third Labor candidate Trimann Gill was excluded with 89,740 (8.0%) to Game’s 97,755 (8.7%) and Liddle’s 107,705 (9.5%), though Liddle would have won at the final count either way.

The AEC now announces the button will be pressed at 3pm on Tasmania, which I consider a foregone conclusion of two each for Liberal and Labor and one each for the Greens and the Jacqui Lambie Network.

Tuesday, June 14

The Australian Electoral Commission has announced the buttons will be pushed on the Senate counts today for the Australian Capital Territory at 10am and the Northern Territory at 11am. There seems little doubt about the former result and none about the latter, but it will be interesting to see exactly how minor party and independent preferences flow through to what I am presuming will be a win for independent David Pocock over Liberal incumbent Zed Seselja in the ACT. It also seems likely that the resolution in South Australia and Tasmania is not far off.

For the lower house, the AEC is now conducting Coalition-versus-Labor counts in the 26 seats where the two-candidate preferred counts include independents or minor parties, a process that is a little over 10% complete. This will finally provide us with a definitive two-party preferred and swing results for the election as a whole. The count so far has been systematically favourably for the Coalition because they are starting with declaration votes, and in particular with postal votes, which account for over 80% of those counted so far. It is for this reason that they point to a collective swing to the Coalition of around 9%, which will assuredly not be the case after the ordinary votes are added.

My displays of the lower house results can be found here, but for the two-party preferred results you will need the AEC site.

UPDATE: Counts for the Northern Territory and the Australian Capital Territory resolved as anticipated. The preference distributions will be published later today. The button will be pressed on South Australia at 3pm local time tomorrow. Liberal candidate Andrew Constance has requested a recount after his defeat by 373 votes in Gilmore.

UPDATE 2: The distributions have been posted for both the Australian Capital Territory and Northern Territory. The ACT result was not close: at the final distribution, David Pocock had 103,618 votes (36.3%) and Zed Seselja had 81,485 (28.6%).

UPDATE 3: I had made the following calculations using the ballot paper data of how the various candidates’ preferences were apportioned between a) Labor, Liberal and David Pocock, b) Liberal and David Pocock, and c) Labor and Liberal. Since Labor polled almost exactly a quota, the question of preferences either to them or from them was academic so far as the count was concerned, but instructive with respect to voter behaviour.