Newspoll: 51-49 to Coalition

The second federal poll since the election finds the Coalition back where it started after an apparent post-election bounce in the previous poll three weeks ago.

Newspoll’s first result in three weeks, and second since the election, turns up a surprise in recording a shrinking in the Coalition’s lead from 53-47 to 51-49 – which, if meaningful, would mean an end to the honeymoon period and a return to where things stood at election time. On the primary vote, the Coalition is on 42%, down two points on the last poll and up 0.6% on the election result; Labor is on 34%, up one point and 0.7%; the Greens are on 11%, steady and up 0.6%; and One Nation are on 4%, up one point and 0.9%.

Leadership ratings are likewise consistent with the fading of a post-election sugar hit, with Scott Morrison down three on approval to 48% and up six on disapproval to 42%. Anthony Albanese’s ratings also seem to be trending from mediocre to respectable, with his approval up two to 41% and disapproval down to 34%, leaving him shading Morrison by a point on net approval. However, this hasn’t translated to preferred prime minister for some reason, on which Morrison holds a healthy lead of 48-30, out from 48-31 last time.

The poll was conducted by online and automated phone surveying from a sample of 1623, from Thursday to Sunday. Full report from The Australian here. As before, we remain in the dark as to how the pollster’s methods have been adjusted since the election failure, if at all. However, the size of the movements, and the lack of anything obvious to explain them, suggests the poll has not had been subjected to the smoothing method that Newspoll must have been using before the election to give it its uncanny and, as it turned out, misleading consistency.

Various stuff that’s happening

Sarah Henderson reportedly struggling in her Senate preselection comeback bid, plus yet more on the great pollster failure, and other things besides.

Newspoll’s no-show this week suggests last fortnight’s poll may not have portended a return to the familiar schedule. Amid a general post-election psephological malaise, there is at least the following to relate:

• The great pollster failure was the subject of a two-parter by Bernard Keane in Crikey yesterday, one part examining the methodological nuts and bolts, the other the influence of polling on journalism and political culture.

Richard Willingham of the ABC reports former Corangamite MP Sarah Henderson is having a harder-than-expected time securing Liberal preselection to replace Mitch Fifield in the Senate, despite backing from Scott Morrison, Josh Frydenberg and Michael Kroger. According to the report, some of Henderson’s backers concede that Greg Mirabella, former state party vice-president and the husband of Sophie Mirabella, may have the edge.

• The Joint Standing Committee on Electoral Matters has invited submissions for its regular inquiry into the 2019 election, which will be accepted until Friday, September 2019. Queensland LNP Senator James McGrath continues to chair the committee, which consists of five Coalition, two Labor and one Greens member.

Daniella White of the Canberra Times reports Labor is struggling to find candidates for next October’s Australian Capital Territory election, said by “some insiders” to reflect pessimism about the government’s chances of extending its reign to a sixth term.

• The Federation Press has published a second edition of the most heavily thumbed tome in my psephological library, Graeme Orr’s The Law of Politics: Election, Parties and Money in Australia. A good deal of water has passed under the bridge since the first edition in 2010, most notably in relation to Section 44, which now accounts for the better part of half a chapter.

Call of the board: Melbourne

More gory detail on the result of the May 18 federal election, this time focusing on Melbourne, where an anticipated election-winning swing to Labor crucially failed to materialise.

Time for part four in the series that reviews the result of the May 18 election seat by seat, one chunk at a time. As will be the routine in posts covering the capital cities, we start with a colour-coded map showing the two-party preferred swing at polling booth level, with each booth allocated a geographic catchment area by means explained in the first post in this series. Click for an enlarged image.

Now to compare actual election results to those predicted by a demographic linear regression model, to help identify where candidate or local factors might be needed to explain the result. I now offer a new-and-improved form of the model that includes interaction effects to account for the differences in demographic effects between the cities and the regions. The utility of the change, if any, will become more apparent when I apply it to regional seats, which confounded the original version of the model. The coefficients and what-have-you can be viewed here – the table below shows the modelled predictions and actual results for Labor two-party preferred, ranked in order of difference between the result and the prediction of the model.

The main eyebrow-raisers are that the model anticipates a stronger performance by Labor in nearly every Liberal-held seats, to the extent that blue-ribbon Higgins and Goldstein are both rated as naturally highly marginal. While this could prove a portent of things to come in these seats, it might equally reflect a model leaning too heavily on the “secular/no religion” variable to cancel out the association between income and Liberal support in the inner cities.

As in Sydney, the numbers provide strong indications of incumbency advantages, with both Labor and Liberal members tending to outperform the model and thus appear at opposite ends of the table. I suspect this reflects both the obvious explanation, namely personal votes for sitting members, and a lack of effort by the parties into each other’s safe seats. A tendency for parties to perform more modestly when a seat is being vacated is not so overwhelming as to prevent strong results relative to the model for Labor in Jagajaga and Liberal in Higgins.

With that out of the way:

Aston (Liberal 10.1%; 2.7% swing to Liberal): Aston attracted a lot of discussion after the 2004 election when the Liberals recorded a higher two-party vote than they did in their jewel-in-the-crown seat of Kooyong. Now, for the first time since then, it’s happened again, and by a fairly substantial margin (the Liberal-versus-Labor margin in Kooyong having been 6.7%). As illustrated in the above table, the swing places Alan Tudge’s margin well beyond what the seat’s demographic indicators would lead you to expect.

Bruce (Labor 14.2%; 0.1% swing to Labor): Located at the point of the outer suburbs where the Labor swing dries up, cancelling out any half-sophomore effect that may have been coming Julian Hill’s way after he came to the seat in 2016.

Calwell (Labor 18.8%; 0.9% swing to Liberal): Among the modest number of Melbourne seats to swing to the Liberals, reflecting its multiculturalism and location at the city’s edge. Maria Vamvakinou nonetheless retains the fifth biggest Labor margin in the country.

Chisholm (Liberal 0.6%; 2.3% swing to Labor): Labor’s failure to win Chisholm after it was vacated by Julia Banks was among their most disappointing results of the election, but the result was entirely within the normal range both for Melbourne’s middle suburbs and a seat of its particular demographic profile. The swing to Labor was concentrated at the northern end of the electorate, which may or may not have something to do with this being the slightly less Chinese end of the electorate.

Cooper (Labor 14.6% versus Greens; 13.4% swing to Labor): With David Feeney gone and Ged Kearney entrenched, the door seems to have slammed shut on the Greens in the seat formerly known as Batman. After recording high thirties primary votes at both the 2016 election and 2018 by-election, the Greens crashed to 21.1%, while Kearney was up from 43.1% at the by-election to 46.8%, despite the fact the Liberals were in the field this time and polling 19.5%. In Labor-versus-Liberal terms, a 4.2% swing to Labor boosted the margin to 25.9%, the highest in the country.

Deakin (Liberal 4.8%; 1.7% swing to Labor): While Melburnian backers of the coup against Malcolm Turnbull did not suffer the retribution anticipated after the state election, it may at least be noted that Michael Sukkar’s seat swung the other way from its demographically similar neighbour, Aston. That said, Sukkar’s 4.8% margin strongly outperforms the prediction of the demographic model, which picks the seat for marginal Labor.

Dunkley (LABOR NOTIONAL GAIN 2.7%; 1.7% swing to Labor): Together with Corangamite, Dunkley was one of only two Victorian seats gained by Labor on any reckoning, and even they can be excluded if post-redistribution margins are counted as the starting point. With quite a few other outer urban seats going the other way, and a part-sophomore effect to be anticipated after he succeeded Bruce Billson in 2016, it might be thought an under-achievement on Chris Crewther’s part that he failed to hold out the tide, notwithstanding the near universal expectation he would lose. However, his performance was well beyond that predicted by the demographic model, which estimates the Labor margin at 6.6%.

Fraser (Labor 14.2%; 6.1% swing to Liberal): Newly created seat in safe Labor territory in western Melbourne, it seemed Labor felt the loss here of its sitting members: Bill Shorten in Maribyrnong, which provided 34% of the voters; Maria Vamvakinou in Calwell, providing 29%; Tim Watts in Gellibrand, providing 20%; and Brendan O’Connor in Gorton, providing 16%. The newly elected member, Daniel Mulino, copped the biggest swing against Labor in Victoria, reducing the seat from first to eleventh on the national list of safest Labor seats.

Gellibrand (Labor 14.8%; 0.3% swing to Liberal): The city end of Gellibrand followed the inner urban pattern in swinging to Labor, but the suburbia at the Point Cook end of the electorate tended to lean the other way, producing a stable result for third-term Labor member Tim Watts.

Goldstein (Liberal 7.8%; 4.9% swing to Labor): Tim Wilson met the full force of the inner urban swing against the Liberals, more than accounting for any sophomore effect he might have enjoyed in the seat where he succeeded Andrew Robb in 2016. Nonetheless, he maintained a primary vote majority in a seat which, since its creation in 1984, has only failed to do when David Kemp muscled Ian Macphee aside in 1990.

Gorton (Labor 15.4%; 3.0% swing to Liberal): The swing against Brendan O’Connor was fairly typical of the outer suburbs. An independent, Jarrod Bingham, managed 8.8%, with 59.2% of his preferences going to Labor.

Higgins (Liberal 3.9%; 6.1% swing to Labor): One of many blue-ribbon seats that swung hard against the Liberals without putting them in serious danger. Nonetheless, it is notable that the 3.9% debut margin for Katie Allen, who succeeds Kelly O’Dwyer, is the lowest the Liberals have recorded since the seat’s creation in 1949, surpassing Peter Costello’s 7.0% with the defeat of the Howard government in 2007. Labor returned to second place after falling to third in 2016, their primary up from 14.9% to 25.4%, while the Greens were down from 25.3% to 22.5%. This reflected a pattern through much of inner Melbourne, excepting Melbourne and Kooyong.

Holt (Labor 8.7%; 1.2% swing to Liberal): The populous, northern end of Holt formed part of a band of south-eastern suburbia that defied the Melbourne trend in swinging to Liberal, causing a manageable cut to Anthony Byrne’s margin.

Hotham (Labor 5.9%; 1.7% swing to Labor): The swing to third-term Labor member Clare O’Neil was concentrated at the northern end of the electorate, with the lower-income Vietnamese area around Springvale in the south went the other way.

Isaacs (Labor 12.7%; 3.4% swing to Labor): What I have frequently referred to as an inner urban effect actually extended all along the bayside, contributing to a healthy swing to Mark Dreyfus. The Liberal primary vote was down 7.4%, partly reflecting more minor party competition than in 2016. This was an interesting case where the map shows a clear change in temperature coinciding with the boundaries, with swings to Labor in Isaacs promptly giving way to Liberal swings across much of Hotham, Bruce and Holt.

Jagajaga (Labor 6.6%; 1.0% swing to Labor): Jenny Macklin’s retirement didn’t have any discernible impact on the result in Jagajaga, which recorded a modest swing to her Labor successor, Kate Thwaites.

Kooyong (Liberal 5.7% versus Greens): Julian Burnside defied a general Melburnian trend in adding 2.6% to the Greens primary vote, and did so in the face of competition for the environmental vote from independent Oliver Yates, whose high profile campaign yielded only 9.0%. Labor was down 3.7% to 16.8%, adrift of Burnside’s 21.2%. But with Josh Frydenberg still commanding 49.4% of the primary vote even after an 8.3% swing, the result was never in doubt. The Liberal-versus-Labor two-party margin was 6.7%, a 6.2% swing to Labor.

Lalor (Labor 12.4%; 1.8% swing to Liberal): The area around Werribee marks a Liberal swing hot spot in Melbourne’s west, showing up as a slight swing in Lalor against Labor’s Joanne Ryan.

Macnamara (Labor 6.2%; 5.0% swing to Labor): Talked up before the event as a three-horse race, this proved an easy win for Labor, who outpolled the Greens 31.8% to 24.2%, compared with 27.0% to 23.8% last time, then landed 6.2% clear after preferences of the Liberals, who were off 4.6% to 37.4%. The retirement of Michael Danby presumably explains the relatively weak 5.0% primary vote swing to Labor in the seven booths around Caulfield and Elsternwick at the southern end of the electorate, the focal point of its Jewish community. The result for the remainder of the election day booths was 9.7%.

Maribyrnong (Labor 11.2%; 0.8% swing to Liberal): Nothing out of the ordinary happened in the seat of Bill Shorten, who probably owes most of his 5.0% primary vote swing to the fact that there were fewer candidates this time. Typifying the overall result, the Liberals gained swings around Keilor at the electorate’s outer reaches, while Labor was up closer to the city.

Melbourne (Greens 21.8% versus Liberal; 2.8% swing to Greens): The Greens primary vote in Melbourne increased for the seventh successive election, having gone from 6.1% in 1998 to 22.8% when Adam Bandt first ran unsuccessfully in 2007, and now up from 43.7% to 49.3%. I await to be corrected, but I believed this brought Bandt to within an ace of becoming the first Green ever to win a primary vote majority. For the second election in a row, Bandt’s dominance of the left-of-centre vote reduced Labor to third place. On the Labor-versus-Liberal count, Labor gained a negligible 0.1% swing, unusually for a central city seat.

Menzies (Liberal 7.2%; 0.3% swing to Labor): Very little to report from Kevin Andrews’ seat, where the main parties were up slightly on the primary vote against a smaller field, and next to no swing on two-party preferred, with slight Liberal swings around Templestowe in the west of the electorate giving way to slight Labor ones around Warrandyte in the east.

Scullin (Labor 21.7%; 2.1% swing to Labor): Third-term Labor member Andrew Giles managed a swing that was rather against the outer urban trend in his northern Melbourne seat.

Wills (Labor 8.2% versus Greens; 3.2% swing to Labor): The Greens likely missed their opportunity in Wills when Kelvin Thomson retired in 2016, when Labor’s margin was reduced to 4.9%. Peter Khalil having established himself as member, he picked up 6.2% on the primary vote this time while the Greens fell 4.3%. Khalil also picked up a 4.2% swing on the Labor-versus-Liberal count, strong even by inner urban standards, leaving him with the biggest margin on that measure after Ged Kearney in Cooper.

EMRS: Liberal 38, Labor 30, Greens 16 in Tasmania

The Liberals maintain the lead in a new Tasmanian state poll, but Hare-Clark and a high minor party vote obscure the picture.

It seems the only state that can rely on a regular polling series these days in Tasmania, where local outfit EMRS has been plugging away with state voting intention polling on a more-or-less quarterly basis for as long as this site can remember.

Its latest numbers raise the possibility that state Labor has suffered collateral damage from the federal election, with their vote down four to 30%, while the Liberals are steady at 38% and the Greens up three to 16%. Nonetheless, Labor’s Rebecca White remains in the unusual position of leading the incumbent, Will Hodgman, by 45-40 as preferred premier, up from 43-41.

Kevin Bonham’s take on the voting intention numbers is that the Liberals would hold on to a bare majority, with the qualification that a) the Greens and especially “others” numbers seem higher than plausible, true to the pollster’s form, and b) Sue Hickey’s rebellious activities raise doubts about the implied second seat for the Liberals in Denison/Clark.

The poll was conducted last Monday to Wednesday from a sample of 1000. The series used to be branded as a phone poll, but these days the survey method is not divulged. Kevin Bonham’s pollster field guide suggests it now covers the gamut of landline, mobile and online panel polling.

Essential Research: Newstart, robodebt, social media

More evidence that voters favour social democratic policy options, right up until polling day.

The fortnightly Essential Research poll, which is still yet to resume results for voting intention, focuses largely on questions around social security. Among its findings are that the Newstart rate is deemed too low by 58%, about right by 30% and too high by 5%. Forty-four per cent expressed strong support for an increase from $280 per week to $355, a further 31% said they somewhat supported it, and only 18% said they were opposed, 7% strongly.

I don’t normally make anything out of breakdowns published in average sample polls, but it’s interesting to note that the “too low” response increases progressively across the three age cohorts to peak at 66% among the 55-and-over. There was also a relationship between age and correct answers to a question in which respondents were asked to identify the weekly Newstart payment, the overall result for which was 40%, up from 27% when it was previously asked last June. Only 29% of Coalition voters expressed strong support for an increase compared with 55% for Labor supporters, but the difference was narrower when combined with the “somewhat” response, at 84% to 68%.

On the Centrelink “robodebt” debt recovery program, 58% supported calls for it to be shut down compared with 32% opposed. Twenty-two per cent said they had heard a lot about the program and 30% a little, while 18% said they had not heard any details and 30% that they were not aware of it at all.

The one question not relating to social security covers social media companies’ collection of personal information, with 80% expressing concern about the matter and the same number wanting tighter regulation. The affirmative response for both questions progressively increased across the three age cohorts.

Also noteworthy from the poll is that Essential Research has taken to publishing “base” figures for each cohort in the breakdown, which evidently reflect their proportion of the total after weightings are applied. This is at least a step in the direction of the transparency that is the norm in British and American polling, in that it tell us how Essential is modelling the overall population, even if it doesn’t divulge how much each cohort’s responses are being weighted to produce those totals.

The poll was conducted Thursday to Sunday from an online sample of 1102 respondents.

Federal election preference flows

New figures from the AEC confirm the Coalition’s share of Hanson and Palmer preferences was approaching two-thirds, a dramatic increase on past form.

We now have as much in the way of results out of the federal election as we’re ever going to, with the Australian Electoral Commission finally publishing preference flow by party data. The table below offers a summary and how it compares with the last two election. They confirm that YouGov Galaxy/Newspoll was actually too conservative in giving the Coalition 60% of preferences from One Nation and the United Australia Party, with the actual flow for both parties being nearly identical at just over 65%.

The United Australia Party preference flow to the Coalition was very substantially stronger than the 53.7% recorded by the Palmer United Party in 2013, despite its how-to-vote cards directing preferences to the Coalition on both occasions. A result is also listed for Palmer United in 2016, but it is important to read these numbers in conjunction with the column recording the relevant party’s vote share at the election, which in this case was next to zero (it only contested one lower house seat, and barely registered there). Greens preferences did nothing out of the ordinary, being slightly stronger to Labor than in 2016 and slightly weaker than in 2013.

The combined “others” flow to the Coalition rose from 50.8% to 53.6%, largely reflecting the much smaller footprint of the Nick Xenophon Team/Centre Alliance, whose preferences in 2016 split 60-40 to Labor. This also contributes to the smaller share for “others”, with both figures being closer to where they were in 2013. “Inter-Coalition” refers to where there were both Liberal and Nationals candidates in a seat, some of whose preferences will have flowed to Labor rather than each other. The “share” result in this case records the combined Coalition vote in such seats as a share of the national formal vote.

While we’re here, note the blog’s other two recent posts: Adrian Beaumont’s account of Brecon & Radnorshire by-election, and my own in-depth review of the legal challenges against the election of Josh Frydenberg in Kooyong and Gladys Liu in Chisholm.