Analysis and discussion of elections and opinion polls in Australia
This post will be progressively updated to follow the late counting in the undecided seat of Petrie.
Election night. Kevin Rudd’s concession speech boast notwithstanding, Labor’s is slightly behind the eight-ball in two of its Queensland seats, with an ABC-projected swing of 3.0% against a margin of 2.5%. There is essentially a full election night’s complement of results, not counting those cast at the Brisbane town hall.
Breaking news: Abbott wins.
Nothing actually further to add at this stage, but the new day requires a new thread. Please observe the plethora of new posts below for those wishing to follow late counting, and especially the fairly detailed one attempting to review the Senate situation. These extend on to the second page; I’ll do something to make them more accessible in the morning.
And they’re off …
You’ll be able to hear my own dulcet tones this evening on ABC Radio’s live coverage, along with those of Fran Kelly, Mark Colvin, Sandy Aloisi and Senators John Faulkner and Michael Ronaldson, which is being carried on all ABC local, regional and national stations and of course online. Below you’ll find embedded Crikey’s contribution to this evening’s festival of democracy, which I might find time to contribute to in a quiet moment. For those of you who like it both ways, an old-fashioned comments thread is open below. Moderation will inevitably be light to non-existent, so play nice everybody.
We’ve got exit polls, from which the only clear message to emerge is that exit polling still has a way to go, in this country at least. Morgan has been publishing results of SMS polling throughout the day which has consistently had the Coalition ahead 52-48, with primary vote figures at 4pm of 33.5% for Labor, 42.5% for the Coalition, 11.5% for the Greens and 5% for the Palmer United Party (hourly updates here. But Newspoll has results suggesting Labor will be very lucky to get off that lightly, with Labor to lose 14 seats in New South Wales, seven in Queensland, three in Victoria (with Labor to gain Melbourne and Sophie Mirabella to retain Indi), but a status quo result in South Australia and Tasmania apparently not covered. So at the very least, we have a credibility race on between Newspoll and Morgan this evening. For what it’s worth, Morgan’s exit polling wasn’t far off last time. Many thanks to the PB comments community for doing my research for me here.
Three more big-sample polls for those of you still wondering what the next three years might have in store. Alternatively, you could just wait a couple of hours.
The last of the major polls go as follows:
• A Newspoll survey conducted on Wednesday and Thursday from over 2500 respondents has Labor on 33%, the Coalition on 46% and the Greens on 9%, for a commanding Coalition lead of 54-46 on two-party preferred. Full breakdowns here.
• Nielsen concurs with Newspoll on both major parties’ primary votes in its poll of 1431 respondents conducted on Wednesday and Thursday, but has the Greens two points higher at 11%. Both The Age and the Sydney Morning Herald have exasperatingly declined to provide breakdowns, but from what I can gather from the printed copy, the poll has the Coalition ahead 56-44 in New South Wales and behind 51-49 in Victoria, while in Queensland Labor’s primary vote is on just 27% (under two Senate quotas, for those of you with an eye on that kind of thing).
• Morgan has a poll of 4937 respondents conducted by SMS, online and live interview phone polling which has Labor at just 31.5%, with the Coalition on 44%, the Greens on 10.5% and the Palmer United Party on 6.5%. This pans out to 53.5-46.5 on respondent-allocated preferences, but to 54.5-45.5 on the previous election preferences method used by Nielsen and Newspoll.
BludgerTrack has been updated with all of the above, and it continues to offer a rosier assessment for Labor than the betting markets in particular would suggest (though note that I’ve knocked on the head my idea of revising the preference model to grant Labor a bigger share of the Palmer United Party vote in Queensland, which has made two seats’ difference). As I’ve noted a number of times, this is mostly down to the consistent tendency of electorate-level polling to produce worse results for Labor than that national and statewide polling that are the bread and butter of BludgerTrack. To illustrate this point, and also for your general convenience, I offer below a complete listing to all such polls published during the campaign. Averages are also provided for the swings in each state, and by each pollster. What this suggests is that the automated phone polling by Galaxy, which has generally produced highly plausible results, has not been too far out of line with national polling, and it has generally offered highly plausible results. The live interview phone polling of Newspoll looks to have performed similarly, but that’s because its sample includes the unusual cases of New England and Lyne. Beyond that three automated phone pollsters who are relatively new to the game, and whose consistent findings of huge Coalition swings should accordingly be treated with caution.
Key: NP=Newspoll, RT=ReachTEL, Gal.=Galaxy, Lon.=Lonergan, JWS=JWS Research.
NEW SOUTH WALES N ALP L-NP GRN 2PP SWING Dobell/RobertsonNP 13/8 505 35 50 8 46 -7 Lindsay Lon. 14/8 1038 32 60 3 36 -15 Lyne NP 14/8 504 26 51 7 41 +3 New England NP 14/8 504 24 53 5 34 +1 Kingsford Smith RT 15/8 610 38 47 10 48 -7 McMahon RT 15/8 631 45 50 2 47 -11 Blaxland RT 15/8 636 50 47 3 52 -10 Bennelong RT 15/8 631 28 64 8 35 -12 Macquarie JWS 15/8 710 35 51 8 45 -4 Lindsay JWS 15/8 578 35 57 3 39 -12 Greenway JWS 15/8 570 44 46 1 51 0 Banks JWS 15/8 542 43 50 4 47 -4 Werriwa Gal. 20/8 548 41 48 5 48 -9 Reid Gal. 20/8 557 38 50 9 47 -6 Parramatta Gal. 20/8 561 44 45 4 50 -4 Lindsay Gal. 20/8 566 41 50 3 46 -5 Greenway Gal. 20/8 585 45 46 3 49 -2 Barton Gal. 20/8 551 44 44 9 52 -5 Banks Gal. 20/8 557 40 47 6 48 -3 Barton Gal. 20/8 575 44 44 9 52 -5 Banks Gal. 20/8 575 40 47 6 48 -3 K-S/Page/E-M NP 26/8 601 37 47 11 48 -7 ALP marginals* NP 26/8 800 34 52 7 43 -9 McMahon JWS 28/8 482 44 52 3 47 -11 Average swing -6.1 * Parramatta/Reid/Banks/Lindsay/Greenway. VICTORIA N ALP L-NP GRN 2PP SWING Deakin RT 15/8 619 36 50 13 47 -4 Corangamite RT 15/8 633 36 54 10 44 -6 Melbourne RT 15/8 860 35 24 35 Indi RT 15/8 611 18 47 6 Corangamite JWS 15/8 587 36 48 10 47 -3 Aston JWS 15/8 577 29 59 8 37 -12 La Trobe Gal. 20/8 575 36 45 12 49 -3 Corangamite Gal. 20/8 575 35 52 9 44 -6 Chisholm Gal. 20/8 575 46 45 7 48 -8 ALP marginals* NP 28/8 800 34 47 13 47 -4 McEwen JWS 28/8 540 35 47 6 45 -14 Bendigo JWS 28/8 588 40 40 9 51 -9 Average swing -6.9 * La Trobe/Deakin/Corangamite QUEENSLAND N ALP L-NP GRN 2PP SWING Griffith RT 05/8 702 48 43 8 46 -12 Forde RT 08/8 725 40 48 4 46 -2 Forde Lon. 15/8 1160 34 56 4 40 -8 Forde JWS 15/8 568 33 54 4 40 -8 Brisbane JWS 15/8 607 36 50 9 46 -3 LNP marginals* NP 20/8 1382 32 54 5 40 -8 Forde NP 20/8 502 38 48 5 46 -2 Griffith Lon. 21/8 958 38 47 11 48 -10 Griffith NP 22/8 500 37 48 12 48 -10 Lilley JWS 28/8 757 40 48 5 46 -7 Griffith JWS 28/8 551 48 40 7 57 -1 Blair Gal. 29/8 604 39 40 8 50 -4 Dawson Gal. 29/8 550 34 48 4 43 -5 Griffith Gal. 29/8 655 41 37 12 54 -4 Herbert Gal. 29/8 589 36 47 6 45 -3 ALP marginals** NP 30/8 800 38 42 8 49 -4 Average swing -5.9 * Brisbane/Forde/Longman/Herbert/Dawson/Bonner/Flynn/Fisher ** Moreton/Petrie/Lilley/Capricornia/Blair/Rankin/Oxley WESTERN AUSTRALIA N ALP L-NP GRN 2PP SWING Brand Gal. 29/8 660 42 42 10 52 -1 Hasluck Gal. 29/8 553 34 46 10 45 -4 Perth Gal. 29/8 550 47 35 13 58 +2 Average swing -1.2 SOUTH AUSTRALIA N ALP L-NP GRN 2PP SWING Hindmarsh Gal. 22/8 586 41 44 10 50 -6 Wakefield Gal. 26/8 575 44 35 7 55 -6 Adelaide Gal. 29/8 571 40 39 12 54 -4 Average swing -5.0 TASMANIA N ALP L-NP GRN 2PP SWING Bass RT 22/8 541 30 52 8 42 -15 Braddon RT 22/8 588 36 51 4 43 -14 Denison RT 22/8 563 19 24 11 Franklin RT 22/8 544 30 39 16 51 -10 Lyons RT 22/8 549 30 47 11 44 -18 Bass RT 03/9 659 28 54 10 41 -16 Average swing -14.7 AVERAGE SWING BY POLLSTER N SWING Galaxy 22 -4.3 Newspoll 10 -4.7 JWS Research 13 -6.8 ReachTEL 16 -8.6 Lonergan 3 -11.3
Two days out from the only one that counts, three more polls.
Three more polls have emerged over the past 24 hours, though some of them already seem like old news. ReachTEL in particular will shortly be superseded when the results of the third such poll in consecutive days are revealed on Seven Sunrise at 6am. All three polls, together with new state breakdowns, have been thrown into the BludgerTrack mix. In turn:
• A Galaxy phone poll of 1303 respondents has the Coalition leading 53-47 on two-party preferred from primary votes of 35% for Labor, 45% for the Coalition, 9% for the Greens and 5% for the Palmer United Party. Full results including attitudinal questions from GhostWhoVotes.
• ReachTEL has Labor’s primary vote at 32.7%, compared with 35.3% on the result from the day before, with the Coalition down from 44.2% to 43.6% and the Greens up from 9.7% to 10.0%. The Palmer United Party meanwhile charges onward from 4.4% to 6.1%. This poll too has the Coalition leading 53-47 on two-party preferred. UPDATE: No need to amend the headline, because today’s poll is apparently 53-47 as well.
• Essential Research has an online poll of 1035 respondents with Labor on 35% of the primary vote (steady on Monday’s result), the Coalition on 43% (down one) and the Greens on 10% (steady). I’m also told the poll has the Palmer United Party on 4%, as did Monday’s result. On two-party preferred the Coalition lead is at 52-48, down from 53-47 on Monday.
UPDATE: Morgan has a poll of 3939 respondents conducted last night and the night before by SMS, phone (live interview I assume) and online which has Labor on 31.5%, the Coalition on 45%, the Greens on 9.5% and the Palmer United Party on 6.5%. The published two-party preferred figure is 53.5-46.5 to the Coalition, which I presume to be respondent-allocated. State breakdowns are promised this afternoon, and there will be a “final poll” both conducted and released this evening.
UPDATE 2: The Guardian has a Lonergan Research automated phone poll of 862 respondents showing the Coalition lead at a narrow 50.8-49.2, with primary votes of 34% for Labor, 42% for the Coalition, 14% for the Greens and a relatively modest 10% for “others”. It also features, for what it’s worth (not much in my experience), Senate voting intention: 29% Labor, 40% Coalition, 16% Greens and 8% “others”. This seems consistent with the general pattern of Senate polling to inflate the vote for the Greens (and, back in the day, the Democrats).
UPDATE 3: Channel Nine reports tomorrow’s Nielsen poll has the Coalition leading 54-46.
Some musings on Senate prospects for micro-parties, plus a few recent updates to the seat-by-seat election guide.
I’m running the above headline essentially because I have no new poll to trumpet for the following assortment of bits-and-pieces. The latest addition is yesterday’s large-sample ReachTEL poll, which was a relatively good result for Labor taking into account the past lean to the Coalition in this series. Its inclusion caused a 0.6% shift in Labor’s favour without affecting the seat projection, mostly because the improvement was concentrated in Victoria where there are few marginal seats. This isn’t the first time recently that the addition of a ReachTEL result has caused BludgerTrack to move in Labor’s favour, which raises the possibility that the series is not as pro-Coalition as it used to be. If so, the addition of the result with out-of-date bias adjustments attached might be causing the present BludgerTrack numbers to flatter Labor slightly. There has apparently been, for the second evening running, a poll conducted overnight by ReachTEL which will have been unveiled on Seven Sunrise by the time most of you are reading this.
(UPDATE: A less good result today for Labor, and another good one for the Palmer United Party. Labor’s primary vote is down to 32.7% and the Coalition’s up to 43.6%, with the Greens on 10.0% and Palmer on 6.1%. Two-party preferred is 53-47 to the Coalition. ReachTEL also has a very ugly result for Labor from the Tasmanian seat of Bass, courtesy of the Launceston Examiner, with Liberal candidate Andrew Nikolic on 51.8% and Labor member Geoff Lyons on 26.6%.)
Now to those bits and pieces. First, I address what looks to be one of the election’s most significant imponderables: the share of the vote that will go to micro-parties in the Senate. Much hinges on the answer, given the tightness of the preference arrangements between micro-parties and the extremely limited value of polling as a guide to the smaller details of Senate voting patterns. Tim Colebatch of Fairfax has run reports over the past week based based on what Antony Green’s Senate election calcalators come up with when seemingly plausible vote share scenarios are plugged into them, which have been partly inspired by simulations conducted by Poll Bludger commenter Truth Seeker (who details them on his own blog).
One particularly headline-grabbing observation was that Pauline Hanson might succeed in her bid for a New South Wales Senate seat at the expense of Arthur Sinodinos, who has the number three position on the Coalition ticket in New South Wales. Since Labor, the Coalition and the Greens all have Hanson last on their preference order, this can only happen if she and the various parties feeding her preferences collectively amount to more than a quota (14.3%). Colebatch argues that this is highly plausible: “In 2010, 29 micro-parties won 14 per cent of the vote between them. This time there will be 41 of them, and disillusioned Labor supporters could swell their collective vote to 20 per cent – easily enough for a Senate quota.”
This appears to assume that collective vote share of micro-parties will continue to expand as more of them enter the field. Evidence from the last three elections, which provide a common footing in that the Democrats and One Nation had faded from minor to micro-party status, provides some support for this. Excluding the unusual circumstance of South Australia in 2007, when Nick Xenophon polled a full quota in his own right, there are 17 state-level observations for modelling the relationship between the number of Senate groups and the vote share for micro-parties (which I take to mean everyone other than Labor, the Coalition and the Greens). The model I have derived is 0.243+(0.283*A)+(0.681*B), where A is the number of Senate groups and B is the “others” vote in the House of Representatives from the state in question. This has an R-squared of 0.517 and a p-value of 0.006, which is to say that the model explains 51.7% of the variation in these 17 results and has a 99.4% chance of being better than no model at all.
With unprecedented numbers of Senate groups at this election ranging from 23 in Tasmania to 44 in New South Wales, this suggests “others” votes ranging from 12.9% to 20.5% (going off the BludgerTrack projections for the lower house “others” vote), which is well in line with Colebatch’s expectations. However, there’s a considerable theoretical problem with the model in that it presumes the relationship to be perfectly linear. If this were so, the major party vote would disappear altogether if only enough micro-parties took the field. In reality, the rate of increase has to taper off, and the meagre sample of observations available offers no insight as to point at which it does so. My own guess though is that it kicks in fairly sharply before we reach the stage where we can start talking of an aggregate micro-party vote approaching 20%.
To offer some historic guidance as to the sorts of numbers you should be punching into the Senate calcalators, the table below displays the vote for micro-parties of various kinds in each state. “Religious” includes the Democratic Labour Party, although they no doubt occupy something of a grey area. The “right” category is exclusive of the “religious” one. “Left” is defined broadly to incorporate the Democrats and all environmentalist concerns, even ostensibly conservative ones. There were also parties and independents that were deemed not to fall into any of these categories, so the “total” column is not simply an aggregate of the other three.
2010 Relig. Right Left Total NSW 3.63 5.55 3.37 13.82 Victoria 5.35 3.83 3.28 13.2 Queensland 4.31 7.59 3.61 16.43 WA 3.71 2.66 2.81 9.92 SA 5 2.65 2.55 11.11 Tasmania 1.69 2.24 0.66 5.36 TOTAL 4.16 4.81 3.18 13.13 2007 NSW 3.83 3.35 2.44 10.17 Victoria 3.77 1.24 3.16 8.72 Queensland 2.76 6.73 3.04 13.08 WA 3.57 1.1 1.84 7.04 SA 3.97 1.68 1.68 22.25 Tasmania 2.67 0.19 0.78 4.38 TOTAL 3.48 2.97 2.57 10.71 2004 NSW 3.17 3.73 3.88 12.17 Victoria 4.16 1.55 4.18 10.98 Queensland 3.37 9.34 4.09 18.05 WA 2.73 2.82 3.05 9.22 SA 3.98 1.53 3.95 10.02 Tasmania 3.03 0.16 0.82 7.04 TOTAL 3.42 2.93 3.84 12.22
Now to some scattered bits of news for around the traps that I have recently used to supplement the seat-by-seat election guide:
Indi (Liberal 9.0%):Liberals have been telling journalists of serious concerns for Sophie Mirabella’s hold on Indi, where she faces a well-organised challenge from independent Cathy McGowan. The Guardian reports on widespread opinion polling being conducted in the electorate; the Weekly Times reports that Labor are campaigning strongly to boost McGowan; and The Australian reports some in the Liberal Party have been urging Tony Abbott to visit the electorate. The contest is another source of friction between the coalition parties, with former state Nationals MP Ken Jasper among those who are throwing their weight behind McGowan.
Melbourne (Greens 6.0% versus Labor): The Greens have been spruiking a poll of 400 respondents conducted for them by Galaxy showing Adam Bandt’s primary vote up 4% since the 2010 election, with “as many as four in 10” Liberal voters in the seat planning to ignore the direction of their party’s how-to-vote card that voters should favour Labor ahead of the Greens in their preference allocation. This is actually in line with the 35% rate of leakage in inner Melbourne when the Liberals likewise directed preferences against the Greens at the 2010 state election, which nonetheless wasn’t high enough to win them any of the seats they were anticipating. But taken together with the purported primary vote swing, it suggests a very close result.
McMahon (Labor 7.8%): The Liberal candidate for Chris Bowen’s western Sydney seat, Liverpool area police superintendent Ray King, has been defended by a series of police figures and corruption investigators after Labor claimed he had a “close friend” in Roger Rogerson, the notorious detective who was imprisoned in 1990 for perverting the course of justice. The claim has been denied by Rogerson as well as King, with retired assistant commissioner Geoff Schuberg complaining of a “grubby, baseless smear campaign”.
Forde (Liberal National 1.6%):The Australian reports that Forde MP Bert van Manen, who is fighting off a challenge from Peter Beattie, was the half-owner and recently resigned director of a financial planning firm which owed creditors more than $1.5 million when it collapsed last year. The report says administrators KPMG had told creditors of “unreasonable director-related transactions” behind the collapse. A Liberal spokesperson was quoted saying van Manen had personally settled with the main credtior, Westpac, but no comment was offered on $325,000 owed to three further creditors.
Greenway (Labor 0.9%): The Sydney Morning Herald observes a “systemic” silence among Liberal candidates in Sydney, “with multiple examples emerging of candidates pulling out of events or interviews”. The low profile assumed by Greenway MP Jaymes Diaz has been particularly widely noted, after he failed to show for a candidates forum in Blacktown last week.
Herbert (Liberal National 2.2%) and Dawson (Liberal National 2.4%): Sid Maher of The Australian identifies marginal seats on the central Queensland coast as the main targets for the Coalition’s promised curtailing of marine protected areas, a pitch at commercial and recreational fishers. A similar promise before the 2010 election was “credited with delivering the seat of Dawson”, by persons unidentified.
With most assessments rating Bass and Braddon as likely Labor losses, the central Tasmanian seat of Lyons is thought to the state’s next most likely Liberal gain.
As part of my ongoing endeavour to bring the seat-by-seat election guide up to speed on potentially interesting seats that had previously fallen through the net, today I review the last seat which had remained unattended to in the potential Labor disaster zone of Tasmania.
Known until 1982 as Wilmot, Lyons covers what remains of Tasmania after the north-west coast (Braddon), north-east coast (Bass), central Hobart (Denison) and Hobart’s outskirts (Franklin) are naturally ordered along community of interest lines. It thus includes small towns on either side of Tasmania’s pronounced north-south divide, including New Norfolk outside Hobart and the southern outskirts of Launceston, along with fishing towns and tourist centres on the east coast and rural territory in between. The seat has moved with the state’s distinct electoral rhythms over the decades, being held by the Liberals from 1975 to 1993 and Labor ever since.
As Wilmot the seat was won for Labor in 1929 by the man whose name it now bears, Joseph Lyons, who had been Tasmanian Premier up until his minority government’s defeated in 1928. After assuming the position of Postmaster-General in Jim Scullin’s newly elected Labor government, Lyons and his followers split from the party in a dispute over economic policy in response to the Depression. Joining with the opposition to become the leader of the new conservative United Australia Party, Lyons emerged Prime Minister after a landslide win at the December 1931, and remained so until his death in 1939.
Labor briefly resumed its hold on the seat by winning the by-election held after Lyons’s death, but it was for the United Australian Party by Allan Guy at the next general election the following year. It next changed hands against the trend of a swing away from Labor at the 1946 election, when Guy was unseated by Labor’s Gil Duthie, who would hold the seat for nearly three decades until all five Tasmanian seats went from Labor to Liberal in 1975. The 9.9% swing that delivered the seat to Liberal candidate Max Burr the seat in 1975 was cemented by an 8.0% swing at the next election in 1977, and the Franklin dam issue ensured the entire state remained on side with the Liberals in 1983 and 1984. The realignment came when Burr retired at the 1993 election, when the loss of his personal vote combined with a sharp statewide backlash against John Hewson’s proposed goods and services tax to deliver a decisive 5.6% swing to Labor.
The member since has been Dick Adams, a former state government minister who had lost his seat in 1982 whose background is with the Left faction Liquor Hospitality and Miscellaneous Workers Union, for which he once worked as an organiser. Adams survived a swing in 1996 before piling 9.3% on to his margin in 1998, enough of a buffer to survive a small swing in 2001 and a large one in 2004 when northern Tasmania reacted against Mark Latham’s forest policies, which had been bitterly opposed by Adams. The reaction against Labor in 2004 was reversed in 2007, when Adams picked up a 5.1% swing, and another strong performance by Labor in Tasmania at the 2010 election boosted his margin a further 4.0%. The Liberals’ candidate for the second election running is Eric Hutchison, a wool marketer with Tasmanian agribusiness company Roberts Limited.
A ReachTEL automated phone poll of 549 respondents conducted two-and-a-half weeks out from the election had Eric Hutchison with a commanding lead: 54% to 34% on the primary vote after exclusion of the undecided, and 55.8-44.2 on two-party preferred.
A new ReachTEL poll offers Labor some vague encouragement, and concurs with Morgan and Essential in having Clive Palmer’s party at 4% nationally.
This morning’s Seven Sunrise (which the Liberal Party is carpet-bombing with advertising) has results from a ReachTEL automated phone poll, reporting primary votes of 35% for Labor, 45% for the Coalition and 4% for the Palmer United Party (remarkable unanimity on that figure from pollsters lately). (UPDATE: Full results here. The Coalition vote turns out to round to 44%, not 45%, and the Greens are on 9.7%.) The Coalition’s two-party preferred lead is at 52-48, down from 53-47 a week ago. Tony Abbott leads Kevin Rudd 53-47 on ReachTEL’s all-inclusive preferred prime minister rating, and 51% of respondents reported they favoured abolishing the carbon tax against 34% opposed.
In an otherwise quiet day on the polling front yesterday, AMR Research has published its third online poll of federal voting intention, conducted between Friday and Monday from a sample of 1101, showing Labor on 34%, the Coalition on 44%, and the Greens on 10%.
Finally, to give you something to look at, I’ve extended yesterday’s exercise of providing a state-level BludgerTrack chart for Queensland across all mainland states, with two-party preferred shown along with the primary vote. Once again, black represents the combined “others” vote. Note that the data gets “noisier” as sample sizes diminish for the smaller states. This is not as bad as it looks though with respect to the trendlines, as the outliers are generally from the smallest samples and the model is weighted to limit the influence.
