Category: Federal Election 2019

Essential Research is continuing to provide The Guardian with polling on a fortnightly basis, but is still limiting itself to issue polling in the wake of the great debacle of last month. This week’s poll is concerned with nuclear power, after a push by Queensland MPs James McGrath and Keith Pitt for a parliamentary inquiry into lifting Australia’s nuclear power ban (showing rather unfortunate timing, in view of the runaway success of HBO’s television series Chernobyl). The poll finds a slight majority of 44% to 40% in favour of Australia having nuclear power plants, compared with a 40-40 tie when Essential last posed the question in 2015 – the kicker being that only 28% said they would be comfortable living near one, with 60% disagreeing. Among the other findings, 47% per cent rated that nuclear would be better than coal-fired power for the environment.

In election counting news, the button was pressed yesterday on the New South Wales Senate result, which, foreseeably, produced three seats for the Coalition (Liberals Hollie Hughes and Andrew Bragg, and Perin Davey of the Nationals), two for Labor (Tony Sheldon and Tim Ayres) and one for the Greens (Mehreen Faruqi). Above-the-line votes accounted for 93.1% of the total, which included more than two quotas each for the Coalition and Labor (albeit just barely in the latter case). This meant the top two candidates on the Coalition and Labor tickets were elected immediately, leaving two seats to be determined by the remainder of the preference distribution. The chart below shows how this proceeded as the last eight candidates were excluded, and also shows how the main candidates were placed after the surpluses of the first four elected candidates were distributed (Count 4).

Under the old system, the entirety of the vote was effectively divided between the sixth elected candidates and the unelected seventh, who was left with what is known as the “wastage quotas”. Now that it’s possible for votes to exhaust, it becomes possible for the count to fail to deliver quotas to six candidates, in which case the final seats go to whoever comes nearest at the final count. Such was the case with the last two seats in New South Wales – 0.39 quotas exhausted, and the final three quotas were distributed between three candidates in such a way as to leave all of them short of a full quota. Two of these candidates, Davey of the Nationals and Faruqi of the Greens, finished just short with 0.97 and 0.96 quotas respectively, causing them each to be elected well ahead of Kate McCulloch of One Nation on 0.68.

The chart illustrates exactly how far Jim Molan, shown in blue, fell short of winning the third seat through the strength of his below-the-line support, notwithstanding conservative excitement that he achieved the highest below-the-line vote in Senate history – in terms of aggregate votes, which is naturally a significant qualification when considering a result from New South Wales. Molan’s total share of the first preference vote was 2.92%, some distance behind a number of recent results in Tasmania, where the rate of below-the-line voting is particularly high. His exclusion unlocked a flood of preferences to Davey that closed the gap between her and Faruqi, who were all but level for the remainder of the count.

However, a good many of Molan’s preferences flowed out of the Coalition ticket and further to the right, with 20% going to McCulloch compared with 71.5% for other Coalition candidates. McCulloch also received a strong flow of preferences when Shooters Fishers and Farmers were the last party excluded two counts later. However, this was well short of what she needed to put her in the hunt for the last two seats, for which her share of the total vote would have had to have been about 2% higher. For more details on preferences, Ross Leedham has determined four-party preferred preference flows along the same lines as I provided in yesterday’s post on the Tasmanian result, observing how small party preferences split between the Coalition, Labor, the Greens, One Nation and exhaustion.

To get a sense of how the result might have played out under the old system, I’ve had a play with Antony Green’s Senate calculator from 2013, using the results from this election where possible and judiciously allocating the residue from new parties to old ones. This suggests One Nation would have won the fifth seat at the expense of either the Coalition and the Greens, who would have been in a very tight race for the last seat. One Nation preference feeders would have included not only Shooters and Fishers, Liberal Democrats, Christian Democrats, the Democratic Labour Party and Australian Conservatives (nee Family First), but also leftist concerns such as Animal Justice, thanks to Glenn Druery-inspired preference networks that had nearly every micro-party preferencing each other ahead of the main three.

The button will apparently be pressed on the Western Australian result this morning and Victoria tomorrow, both of which will assuredly produce results of three Liberal, two Labor and one Greens. Not sure when Queensland and South Australia will be done.

A summary of what remains to be resolved of election counting:

• The button is yet to be pressed on five of the eight Senate counts, with Tasmania, the Australian Capital Territory and Northern Territory completed and fully published. More on the Tasmanian result below.

• The Coalition-versus-Labor two-party preferred preference count for Farrer is 54% complete, with the remainder presumably to be knocked over today. Only then will we have a definitive total for the national two-party preferred, but the remaining uncertainty is relevant only to the second decimal place: to the first, the Coalition will finish with 51.5%, a swing of either 1.1% or 1.2%.

• Preference distributions for lower house seats are yet to be published, though in some cases they have assuredly been conducted. As noted previously, only with the distribution could the theoretical (though not practical) possibility of One Nation winning Hunter from Labor be ruled out.

I will be taking a deep dive into each Senate result as they are reported. As discussed here, none of the results are seriously in doubt, with the highly arguable exception of Queensland.

The chart below shows how the late stages of the preference distribution for Tasmania proceeded, after the election of the first three candidates and the elimination of lower order candidates and parties (the latter included independent Craig Garland, who managed a disappointing 3475 votes, compared with the 6633 he polled at last year’s Braddon by-election). The first three were the top two on the Liberal ticket, Richard Colbeck and Claire Chandler, and the first on Labor’s, Carol Brown. Both Liberal and Labor polled clear of two quotas (the primary vote totals can be found here), but owing to Tasmania’s high rate of below-the-line voting (28% in this case), neither scored over two quotas on above-the-line votes alone. However, Chandler was promptly elected after Colbeck as most of his below-the-line votes proceeded straight down the Liberal ticket.

The situation for Labor was more complicated owing to Lisa Singh, who again had to campaign for below-the-line votes to retain her seat after the party placed her fourth on the ticket. This she was able to accomplish at the 2016 double dissolution, when she won Labor’s fifth seat from number six on the ticket. This time though she had the effectively impossible task of winning one of two Labor seats from number four. Singh scored 5.68% of the first preference vote, slightly down on her 6.12% in 2016. This meant she remained in the count longer than the candidate one place above her, who on both occasions was John Short, but she was well behind the second candidate on the Labor ticket, Catryna Bilyk, who received all the above-the-line votes remaining after the election of Brown.

As the chart demonstrates, the race for the last three seats was not close – Labor was always going to win a second seat; Liberal and Labor were both only slightly in excess of two quotas; and the respective vote shares of 12.57% for the Greens and 8.92% for the Jacqui Lambie Network guaranteed them both a seat. Nick McKim of the Greens edged over the line to take the fourth seat after the preferences of various minor parties were distributed. Bilyk and Lambie were both pushed over a quota at the point where Singh was excluded, very slightly behind One Nation candidate Matthew Stephen, although it would have made no difference if Stephen had gone out first. The result was thus clear-cut enough that all elected candidates achieved quotas in their own right, which is not guaranteed under the new Senate electoral system under which some votes can exhaust.

The table below records “four-party preferred” preference splits for those parties that failed to win seats (including Craig Garland as “Group O”).

The Australian Electoral Commission is now conducting Coalition-versus-Labor preference counts in seats where its indicative preference counts included minor party or independent candidates – or, if you want to stay on top of the AEC’s own jargon in these matters, two-party preferred counts in non-classic contests.

Such counts are complete in the seven seats listed below; 94% complete in Warringah, where the current count records a 7.4% swing to Labor, 78% complete in New England, where there is a 1.2% swing to the Coalition; at a very early stage in Clark (formerly Denison, held by Andrew Wilkie); and have yet to commence in Farrer, Indi, Mayo and Melbourne. Labor have received unexpectedly large shares of preferences from the independent candidates in Kooyong, Warringah and Wentworth, to the extent that Kevin Bonham now reckons the final national two-party preferred vote will be more like 51.5-48.5 in favour of the Coalition than the 52-48 projected by most earlier estimates.

We also have the first completed Senate count, from the Northern Territory. This isn’t interesting in and of itself, since the result there was always going to be one seat each for Labor and the Country Liberals. However, since it comes with the publication of the full data file accounting for the preference order of every ballot paper, it does provide us with the first hard data we have on how each party’s preferences flowed. From this I can offer the seemingly surprising finding that 57% of United Australia Party voters gave Labor preferences ahead of the Country Liberals compared with only 37% for vice-versa, with the remainder going to neither.

Lest we be too quick to abandon earlier assessments of how UAP preferences were behaving, this was almost certainly a consequence of a ballot paper that had the UAP in column A, Labor in column B and the Country Liberals in column C. While not that many UAP votes would have been donkey votes as normally understood, there seems little doubt that they attracted a lot of support from blasé voters who weren’t much fussed how they dispensed with preferences two through six. There also appears to have been a surprisingly weak 72% flow of Greens preferences to Labor, compared with 25% to the Country Liberals. It remains to be seen if this will prove to be another territorian peculiarity – my money is on yes.

Note also that there’s a post below this one dealing with various matters in state politics in Western Australia.

I had a paywalled Crikey article on Friday on the religion factor in the election result, drawing on results of the Australian National University’s Australian Election Study survey. Among other things, it had this to say:

The results from the 2016 survey provide some support for the notion, popular on the right of the Liberal Party, that Malcolm Turnbull brought the government to the brink of defeat by losing religious voters, who appear to have flocked back to the party under Morrison. Notably, the fact that non-religious voters trusted Turnbull a lot more than they did Abbott did not translate into extra votes for the Coalition, whereas a two-party swing to Labor of 7% was recorded among the religiously observant.

The charts below expand upon the survey data featured in the article, showing how Labor’s two-party preferred has compared over the years between those who attend religious services several times a year or more (“often”), those who do so less frequently (“sometimes”), and those who don’t do it at all (“never”).

Some other post-election observations:

• Rosie Lewis of The Australian reports the looming Joint Standing Committee on Electoral Matters inquiry into the election will examine the three-week pre-polling period and the extent of Clive Palmer’s campaign spending. There is not, it would seem, any appetite to explore the debilitating phenomenon of fake news proliferating on social media, for which Australia arguably experienced a watershed moment during the campaign through claims Labor had a policy to introduce a “death tax”. This is explored in depth today in a report in The Guardian and an accompanying opinion piece by Lenore Taylor. That said, not all of the mendacity about death taxes was subterranean, as demonstrated by this official Liberal Party advertisement.

• As best as I can tell, all votes for the House of Representatives have been counted now. There was a fall in the official turnout rate (UPDATE: No, actually — it’s since risen to 91.9%, up from 91.0% in 2016), which, together with the fact that not all votes had been counted at the time, gave rise to a regrettable article in the Age-Herald last week. However, as Ben Raue at the Tally Room explores in depth, the turnout rate reflects the greater coverage of the electoral roll owing to the Australian Electoral Commission’s direct enrolment procedures. This appears to have succeeded to some extent in increasing the effective participation rate, namely votes cast as a proportion of the eligible population rather than those actually enrolled, which by Raue’s reckoning tracked up from 80.0% in 2010 to 83.2% – an enviable result by international standards. However, it has also means a larger share of the non-voting population is now on the roll rather than off it, and hence required to bluff their way out of a fine for not voting.

• The rate of informal voting increased from 5.0% to 5.5%, but those seeking to tie this to an outbreak of apathy are probably thinking too hard. Antony Green notes the shift was peculiar to New South Wales, and puts this down to the proximity of a state election there, maximising confusion arising from its system of optional preferential voting. The real outlier in informal voting rates of recent times was the low level recorded in 2007, which among other things causes me to wonder if there might be an inverse relationship between the informal voting rate and the level of enthusiasm for Labor.

The first pollster to put its head above the parapet post-election has been Essential Research, though it’s sensibly refraining from treating us to voting intention results for the time being. As reported in The Guardian yesterday, the pollster’s fortnightly survey focused on what respondents did do rather than what they would do, finding 48% saying their decision was made well in advance of the election, 26% saying they made up their mind in the weeks before the election, and 11% saying they made up their mind on polling day. Lest this seemingly high rate of indecision be cited as an alibi for pollster failure, the historical results of the Australian National University’s Australian Election Study – which you can find displayed on page 18 here – suggest these numbers to be in no way out of the ordinary.

The poll also found those who decided in the final weeks came down 40% for the Coalition and 31% for Labor. However, assuming the sample for this poll was as per the Essential norm of between 1000 and 1100 (which I hope to be able to verify later today), the margin of error on this subset of the total sample would have been over 5%, making these numbers statistically indistinguishable from the almost-final national primary vote totals of 41.4% for the Coalition and 33.3% for Labor. This goes double for the finding that those who decided on election day went Coalition 38% and Labor 27%, remembering this counted for only 11% of the sample.

Perhaps notable is a finding that only 22% of respondents said they had played “close attention” to the election campaign, which compares with results of between 30% and 40% for the Australian Election Study’s almost equivalent response for “a good deal of interest in the election” between 1996 and 2016. Forty-four per cent said they had paid little or no attention, and 34% some attention. These findings may be relevant to the notion that the pollsters failed because they had too many politically engaged respondents in their sample. The Guardian reports breakdowns were provided on this question for voters at different levels of education – perhaps the fact that this question was asked signifies that they will seek to redress the problem by weighting for this in future.

Also featured are unsurprising findings on issue salience, with those more concerned with economic management tending to favour the Coalition, and those prioritising education and climate change favouring Labor and the Greens.

In other post-election analysis news, the Grattan Institute offers further data illustrating some now familiar themes: the high-income areas swung against the Coalition, whereas low-to-middle income ones went solidly the other way; areas with low tertiary education swung to the Coalition, although less so in Victoria than New South Wales and Queensland.

Another popular notion is that Labor owes its defeat to a loss of support among religious voters, as a hangover from the same-sex marriage referendum and, in what may have been a sleeper issue at the cultural level, the Israel Folau controversy. Chris Bowen said in the wake of the defeat that he had encountered a view that “people of faith no longer feel that progressive politics cares about them”, and The Australian reported on Saturday that Labor MPs believed Bill Shorten blundered in castigating Scott Morrison for declining to affirm that he did not believe gay people would go to hell.

In reviewing Labor’s apparent under-performance among ethnic communities in Sydney and Melbourne, Andrew Jakubowicz and Christina Ho in The Conversation downplay the impact of religious factors, pointing to a precipitous decline in support for Christian minor parties, and propose that Labor’s promised expansion of parental reunion visas backfired on them. Intended to capture the Chinese vote in Chisholm, Banks and Reid, the actual effect was to encourage notions of an imminent influx of Muslim immigrants, “scaring both non-Muslim ethnic and non-ethnic voters”.

However, I’m not clear what this is based on, beyond the fact that the Liberals did a lot better in Banks than they did in neighbouring Barton, home to “very much higher numbers of South Asian and Muslim residents”. Two things may be said in response to this. One is that the nation’s most Islamic electorate, Watson and Blaxland, recorded swings of 4% to 5% to the Liberals, no different from Banks. The other is that the boundary between Banks and Barton runs right through the Chinese enclave of Hurstville, but voters on either side of the line behaved very differently. The Hurstville pre-poll voting centre, which serviced both electorates, recorded a 4.8% swing to Labor for Barton, and a 5.7% swing to Liberal for Banks. This may suggest that sitting member factors played an important role, and are perhaps of particular significance for Chinese voters.

So far as the outcome on seats is concerned, two questions from the federal election remain to be answered: who wins Macquarie, which could potentially deliver the Coalition a 78th seat, or – more likely – a 68th for Labor; and who gets the last Senate seat in Queensland. No new numbers have been added to the count in Macquarie since Wednesday, apparently because they’ve been gathering everything together for one last heave. Labor leads by 282; I make it that there are about 950 votes outstanding; the Liberals will need nearly two-third of them to close the gap. Their more realistic hope, if any, is that an error shows up during the preference distribution, but that’s highly unlikely after all the checking that’s been done already.

Out of the other lower house seats, I’ll be particularly interested to see the results of the preference distribution in Joel Fitzgibbon’s seat of Hunter, where there is a chance the One Nation candidate might draw ahead of the Nationals candidate to make the final count. The Nationals have 23.5% of the primary vote to One Nation’s 21.6%, but by applying Senate preference flows from 2016 to allocate the minor parties, I get this narrowing to 27.1% to 26.3%. If nothing else, One Nation making it to second will provide us with hard data on how Coalition preferences divide between Labor and One Nation, a circumstance that has never arisen before at a federal election. The result in the seat of Mirani at the Queensland election in 2017 suggests it should be a bit short of 80%. If so, Fitzgibbon should emerge with a winning margin of about 2%, compared with his 3.0% lead in the Labor-versus-National count.

As discussed here last week, I feel pretty sure Labor’s second Senate candidate in Queensland will be pipped to the last seat by the Greens, though God knows I’ve been surprised before. That will mean three seats for the Coalition and one apiece for Labor, One Nation and the Greens. We probably won’t know the answer for about a fortnight, when the data entry should be completed and the button pressed.

There are other questions we’re still a while away from knowing the answer to, like the final national two-party preferred vote. All that can be said with certainty at this point is that it will be nowhere near what the polls were saying, but the most likely result is around 52-48 to the Coalition. The AEC’s current count says 51.6-48.4, but this doesn’t mean much because it excludes 15 seats in which the two-candidate counts are “non-classic”, i.e. not between the Coalition and Labor. Only when separate Coalition-versus-Labor counts are completed for those seats will we have a definitive result.

We will also have to wait until them for a definitive answer on exactly how many United Australia Party and One Nation preferences flowed to the Coalition. This has been a contentious question for the past year, since pollsters recognised recent federal election results were unlikely to provide a reliable guide to how they would flow this time, as per their usual practice. As Kevin Bonham discusses at length, this was one of many questions on which certain pollsters exhibited an unbecoming lack of transparency. Nonetheless, their decision to load up the Coalition on preferences from these parties has been more than vindicated, notwithstanding my earlier skepticism that the split would be as much as the 60-40 used for both parties by Newspoll.

A full display of the election results, with complete booth figures, swings and probability estimates, can be found here.

Monday, June 3

The four day break in counting in Macquarie ended with all one-way traffic for Labor: postals broke 154-125, out-of-division pre-polls 103-87 and absents 88-66, putting Labor’s lead out from 282 to 348. Antony Green has called it for Susan Templeman, and Templeman has claimed victory. A few hundred votes still to mop up, presumably tomorrow.

Friday, May 31

Still nothing from Macquarie.

Thursday, May 30

No counting was conducted in Macquarie today. I believe the few remaining scraps are likely to be tidied there today, with there not being enough there to overturn the 282 vote Labor lead. The only hope for the Liberals now is a serious error turning up in the full preference count.

Wednesday, May 29

Happily for Labor, my supposition that there wouldn’t be too many absents left in Macquarie was misplaced – a new batch today broke a handy 402-259 their way. The latest batch of out-of-division pre-polls also surprised in breaking 316-170 for Labor. This extends Labor’s lead from 67 to 282, and there wouldn’t be much more still out there than 500 or so pre-polls and 300 postals – unless I’m still wrong about absents, in which case Labor’s lead should widen further.

If any doubt remained in Cowan, it was dealt with by today’s 1456-1061 break to Labor on absents, along with 106-87 on the latest postals. This pushes the Labor lead from 825 to 1239, which means Labor’s lead here is actually greater than it is in Eden-Monaro and Lilley.

Tuesday, May 28

Labor finally took the lead in Macquarie today, emerging with a 67 vote lead after trailing by 39 votes yesterday. This was thanks to a stronger batch of absents than yesterday’s (505-438 to Labor), a slight gain on out-of-division pre-polls (483-477) and rechecking of ordinary votes, which cost the Liberals 27 votes and boosted Labor by six. However, there can’t be too many more absents left in the can, and the Liberals could hope to claw back about 40 votes on remaining postals, of which I would guess there are about 300. That leaves the result in the hands of maybe 1000 further out-of-division pre-polls. These have slightly favoured Labor so far, and did likewise in 2016, but batch results can vary considerably depending on where they are sourced from.

Macquarie is the only seat still seriously in doubt, as Labor’s Anne Aly stubbornly maintains her lead in Cowan. Today it went from 810 to 825, as a batch of out-of-division pre-polls favoured her 896-881. Their might be another 2000 absents and another 2000 out-of-division pre-polls to come, of which the former have favoured Labor while the latter have split exquisitely evenly. The Liberals would need at least a 55-45 split in their favour.

Monday, May 27

Macquarie remains close in every way, with today’s count dominated by a 493-471 split in favour of Labor on out-of-division pre-polls, and absents going 471-469 to Liberal. Together with rechecking, the net effect is to reduce the Liberal lead from 57 to 40. The result hinges mostly on perhaps 1800 outstanding absents, which can vary significantly in their behaviour from batch to batch. In Cowan, out-of-division pre-polls give the Liberals only a very slight boost, reducing the Labor lead from 833 to 810. The Liberals will need about 57% out of maybe 6000 outstanding votes, few of which are postals, the only vote type on which they have come close to doing that well.

Saturday, May 25

Only minor additions to the count today, but that’s enough to be significant in Macquarie, where the Liberal lead is now down from 71 to 46. A batch of declaration votes broke 108-73 in favour of Labor, offsetting a net Liberal gain of 10 from rechecking. Labor’s lead in Cowan is up from 813 to 833, mostly due to a batch of absents that broke 383-352. In Bass, out of division pre-polls broke 167-123 to Labor, reducing the Liberal lead from 699 to 656 (there were also tiny changes from rechecking). In Lilley, Labor’s lead went from 879 to 901 due to rechecking and out of division pre-polls, the latter of which broke 449-427 their way.

Friday, May 24

The small number of provisional votes were counted today in Cowan, and they behaved typically in giving Labor a slight boost, of 211-184. However, the advantage was outweighed by rechecking, with the Labor lead ending the day at 813, down from 839. But with only handfuls of postal voting yet to come, the Liberals are going to have to do unusually well on absents and out-of-division pre-polls.

Macquarie could very easily go either way, but only rechecking was conducted today, the effect of which was to reduce the Liberal lead from 131 to 71. Postals to continue to widen the Liberal lead in Bass, now out from 561 to 699, while absents have moved Labor further ahead in Lilley, from 817 to 879.

Thursday, May 23

Macquarie looks like going right down to the wire, with the first batch of absents favouring Labor 476-444, and the latest batch of postals reversing the earlier tide in breaking 259-194 to Labor. That cuts yesterday’s Liberal lead of 196 to 131. In Cowan, Labor’s lead is out from 748 to 839 as the first absents break 941-843 their way. Conversely, the first absents from Bass have broken 836-787 to Liberal, which, together with rechecking, pushes the Liberal lead out from 497 to 561. In Lilley, Labor’s lead slips slightly from 842 to 817 as the latest postals break 875-787 to the LNP, outweighing a rather hefty 275-197 Labor advantage on the first absents.

Wednesday, May 22

Only rechecking today in Cowan, where Labor ended the day 748 ahead, compared with 762 yesterday. Elsewhere:

Chisholm. Another batch of postals breaks 1064-905 to the Liberal, increasing their lead from 1220 to 1379.

Macquarie. The latest postals have broken 524-396 to Liberal, exactly the same proportion as those already in the count, increasing their lead from 68 to 196.

Bass. The Liberal lead nudges from 453 to 497, with the latest postals breaking 517-483 to Liberal together with some slight ordinary vote adjustments on rechecking.

Lilley. A big batch of postals breaks very much like the first, going to 2551-2093 to the LNP, which reduces the Labor lead from 1288 to 842, but doesn’t change the impression that Labor should be able to hold on.

Tuesday, May 21

My election results facility, linked to above, has ended the day less buggily than it began. Developments from today’s count:

Chisholm. Another 3963 postals have gone similarly the first 5413, and in doing so have increased the Liberal lead from 591 to 1220. The trend of absent votes in 2016 suggests Labor should only be able to claw back about 250 there.

Macquarie. Only rechecking done today, nudging the Liberal lead from 50 to 68. Labor should only make slight gains on absent and out-of-division pre-polls, which I think more likely than not to be outweighed by the Liberal gain on outstanding postals.

Bass. I thought the first batch of postals surprisingly strong for Labor, but it turns out postals behaved no differently from ordinary votes in 2016 as well, where usually they lean conservative. Today’s batch, however, went 767-706 in favour of the Liberals, increasing the lead from 392 to 453. Absent votes were likewise bang on the ordinary votes in 2016; out-of-division pre-polls favoured Liberal. So Labor would need to pull a rabbit out of the hat here.

Lilley. A rare bit of good news for Labor, in that it looks an error had been made in the Geebung booth that had it favouring the LNP 1046-830, but which has now shows as 1033-862 in favour of Labor. That boosts their lead from 901 to 1288, which you’d think would be enough.

Cowan. Labor’s lead has reduced from 1006 to 762 on the back of a second batch of postals, which went 1223-1040 to the Liberals – slightly less favourable for them than the first, of which the Liberals got 56.9% rather than 54.0% – and ordinary vote rechecking, which boosted them by 61. However, absents and out-of-division pre-polls in 2016 behaved very much like ordinary votes, and there shouldn’t be a huge mass of postals outstanding, so I would think it likely Labor will hang on.

Monday, May 20

As you can see above, I now have an election results facility in business, albeit still with a few bugs to be ironed out. With that more-or-less accomplished, I should be able to follow the final stages of the count in more detail. It now appears clear that the Coalition has secured a majority, the most likely result being 77 seats out of 151. Counting of postal votes is still at a fairly advanced stage, and these reliably lean conservative, so the trend of yesterday’s counting was in their favour. However, no absent votes have been counted, and these can sometimes go the other way. Furthermore, Kevin Bonham believes he has observed a tendency of the first batches of postals to be more conservative than later ones. With that in mind, here’s the latest mail from those undecided seats where counting progressed yesterday, i.e. all of them other than Indi and Boothby, where I’m probably being overly cautious in not calling them for independent and Liberal respectively.

Chisholm. Liberal candidate Gladys Liu extended her lead yesterday from 166 to 591. Postals have so far recorded a smaller swing to Labor, of 0.8%, than ordinary votes, which swung 2.5%. Still in doubt.

Macquarie. The Liberals now lead here by 50 votes, after trailing by 312 yesterday. Postals have so far recorded a 2.2% swing to the Liberals, not much different from ordinary votes. Still in doubt.

Bass. Better news for Labor here, with the Liberal lead narrowing from 437 to 392, and postals surprisingly swinging slightly in Labor’s favour after ordinary votes swung over 6%. Still in doubt.

Lilley. Labor’s lead narrowed yesterday from 1110 to 901. Postals haven’t swung much differently from ordinary votes so far, so Labor seem likely to hold on.

Wentworth. Kerryn Phelps conceded defeat to Liberal candidate Dave Sharma yesterday as a strong trend on postals blew the lead out from 1751 to 2864.

Sunday, May 19

This post will be used to provide regularly updated coverage of late counting in seats that remain in doubt, of which I count seven: the marginal Liberal seat of Chisholm in Melbourne, where newcomer Gladys Liu leads by 166 votes (0.11%); Macquarie on Sydney’s western fringe, where Labor incumbent Susan Templeman is 312 votes in front (0.18%); Bass, where Liberal candidate Bridget Archer holds a lead of 437 votes (0.36%) over Labor incumbent Ross Hart; Indi, where independent candidate Helen Haines holds a 2781 lead (1.6%) over the Liberals in her bid to succeed retiring independent Cathy McGowan; Labor-held Lilley in Brisbane, where Labor’s Anika Wells hold a 1110 vote lead (0.69%) as she seeks to succeed the retiring Wayne Swan; and, stretching it a little further, Wentworth, where Liberal candidate Dave Sharma now holds a 1751 lead (1.16%) over independent incumbent Kerryn Phelps, and Boothby, where Liberal incumbent Nicolle Flint leads Labor by 2183 votes (1.18%). Hopefully tomorrow I will finally get the time to fix the bugs in a results reporting facility that will report results and swings at booth level.

From my paywalled article in Crikey yesterday:

In the wake of its most unambiguous failure at a federal election since at least 1980, Australia’s polling industry is licking its wounds.

The Nine/Fairfax papers have announced the Ipsos poll series will be put on ice, and those pollsters who do return to the field shortly will face catcalls whether they persist in recording a Labor lead we now know doesn’t exist, or only now start detecting a Coalition lead that eluded them through the entirety of the past parliamentary term.

Despite it all though, the pollsters’ performance hasn’t been without its defenders.

Spoiler alert: the latter refers to David Briggs and Nate Silver. But Peter Brent can now be added to the list, up to a point, following a review of the issues raised by the polling failure in Inside Story. Specifically, Brent observes that the primary vote miss was less severe than the two-party preferred; that the difference arose from a stronger-than-anticipated flow of minor party and independent preferences to the Coalition; that herding was less apparent on the primary vote (most markedly in the case of Ipsos’s reading of the balance of support between Labor and the Greens); and that the result was, if nothing else, no worse than the Victorian state election.

Another point noted is the strange consistency with which polls have pointed to extravagant gains for Labor in Queensland before and during election campaigns, only for them to fall away at the end. On this occasion, the falling away as recorded by pollsters wasn’t remotely on the scale needed to predict the result, with statewide polling published towards the end of the campaign landing at least 7% shy of what looks like being the Coalition two-party vote in the state.

The question of geographic variability in the pollster failure seemed worth exploring, so I have put together a table of state and electorate level polling published in the last fortnight or so of the campaign, available below the fold at the bottom of the post. Almost all of this polling was conducted by YouGov Galaxy, whether under its own name or as Newspoll. The only exception was a set of state-level two-party preferred totals from Ipsos, published at the tail end of the campaign by the Age-Herald (which performed rather poorly).

Below all this is a list of “average bias” figures, consisting of straight averages of the observed errors, be they positive or negative, rather than the absolute errors. This means combinations of positive and negative results will have the fact of cancelling out — although there were actually very few of those, as the errors tended to be consistently in the one direction. The national and state-level two-party results are estimates provided to me by Nine’s election systems consultant David Quin. With no Coalition-versus-Labor figures available from 15 electorates, this inevitably involves a fair bit of guess work.

A few points should be observed. Given that poll trends pointed to a clear long-term trend to the Coalition, pollsters may be excused a certain amount of Labor bias when evaluating polling that was in many cases conducted over a week before the election. This is particularly true of the Newspoll state aggregates, which cover the full length of the campaign.

Another issue with the Newspoll state aggregates is that One Nation was a response option for all respondents in the early part of the campaign, despite their contesting only 59 out of 151 seats. Their vote here accordingly comes in too high, and as Peter Brent notes, at least part of their failure could be explained by stranded One Nation supporters breaking in unexpectedly large quantities to the Coalition, rather than other minor party targets of opportunity like Clive Palmer.

In seat polling though, where the issue did not arise, the polls were remarkable in having understated support for One Nation, and overstated it for the United Australia Party. This was one face of a two-sided polling failure in Queensland, of which the other was a serious imbalance towards Labor in support recorded for the major parties. While Queensland has caught most of the attention on this score, the polls were just as far out in measuring the primary votes of the major parties in Western Australia. Things were less bad in Victoria, but Coalition support was still significantly underestimated.

The only bright spots in the picture are New South Wales and South Australia, where Newspoll just about nailed the Coalition, Labor and Greens primary votes, and got the big things right in four seat polls. While Labor’s strength was overstated in Macquarie, it does now appear Labor will pull through there – for more on that front, stay tuned to the late counting thread.

Continue reading “Of swings and misses: episode three”