Morgan: 58.5-41.5 to Coalition

The latest Morgan face-to-face poll, combining the results of the last two weekends’ polling, adds to a picture of Labor’s slight gains over the previous month or two being reversed. The Coalition has picked up two points on the primary vote directly at Labor’s expense, with the Coalition on 49.5 per cent and Labor on 32.5 per cent. The Greens are down a point to 12 per cent. On respondent-allocated preferences, the Coalition lead is up from 56.5-43.5 to 58.5-41.5; on the previous election measure it’s up from 53.5-46.5 to 55.5-44.5.

I’ve had quite a lot to say about the persistent gap between Morgan’s two two-party preferred measure, which is a fairly recent phenomenon. The chart below shows how Labor’s share of respondent-allocated preferences has tracked in the three poll series that publish results for both measures, namely Morgan’s frequent face-to-face polls, their less frequent phone polls, and the monthly Nielsen results. All three began the year more or less in the territory of the 2010 election result, which delivered Labor 65.8 per cent of all non-major party preferences.

Up to a month ago, all three seemed to agree that this had declined by about 10 per cent. Since then, we’ve seen individual Nielsen and Morgan phone poll results (which may prove to be aberrations) showing a revesal of that trend, while Morgan face-to-face has the Labor share lower than ever. Morgan face-to-face has also been consistent in giving Labor the lowest share of the three series throughout this year. I should note as always that the previous election measure has a better track record for predicting the election result in any case.

In other poll news, a fortnightly Port Macquarie-based publication called The Port Paper has published results from an automated phone poll conducted by ReachTEL in Rob Oakeshott’s electorate of Lyne showing support for Rob Oakeshott at just 14.8 per cent, against 55.3 per cent for the Coalition and 17 per cent for Labor. This has raised eyebrows on a number of counts. Firstly, the question on voting intention was the last of three put to respondents, after attitudinal questions on carbon tax and pokies reforms (both of which were strongly opposed), which is commonly recognised in the polling caper as the wrong way to get an accurate response. Secondly, the principals behind The Port Paper are very strongly associated with the Nationals. And thirdly, Bernard Keane in Crikey today relates that ReachTEL “proudly announced it was an associate member of Clubs Qld, which has this year been campaigning aggressively against the Andrew Wilkie-led poker machine reform push. The Port Paper story fails to disclose that.”

The Port Paper also published a poll from the corresponding state electorate of Port Macquarie before the election in March, and while it was not brilliantly accurate, its errors were not in a direction that would inspire doubts about its motives. The vote for soon-to-be-defeated independent incumbent Peter Besseling was about right (34 per cent, compared with 36.5 per cent at the election), but the Nationals were too low (40 per cent against 52.2 per cent) and Labor too high (14 per cent against 5.7 per cent).

Newspoll: 57-43 to Coalition

GhostWhoVotes reports Newspoll has the Coalition lead up from 56-44 to 57-43, from primary votes of 27 per cent for Labor (down two), 47 per cent for the Coalition (steady) and a solid 14 per cent for the Greens (up two). Taken together with personal ratings, the figures point to a still worsening disaffection with both options. After a slight recovery last time, Julia Gillard is back in the doghouse with 29 per cent approval (down four) and 61 per cent disapproval (down three), but Tony Abbott is also down three on approval to 36 per cent and up three on disapproval to 55 per cent. Abbott has maintained his one point lead as preferred prime minister, with both down a point to 39 per cent and 38 per cent.

Today’s Essential Research had Labor up a point on two-party preferred (to 56-44 from 57-43) and also on the primary vote, to 32 per cent, with the Coalition and the Greens steady on 50 per cent and 10 per cent. In other findings, 24 per cent support the health package finalised by government last month against 9 per cent opposed, with the great majority either indifferent (31 per cent said it would have little or no impact) or ignorant (28 per cent said they had heard nothing, 36 per cent little). Forty-seven per cent supported David Cameron’s suggestion that access to Twitter and Facebook be blocked during periods of civil unrest, with support varying as you would expect according to age and social media usage.

In further poll news, Phillip Coorey of the Sydney Morning Herald related yesterday that a JWS Research automated phone poll had the Liberals leading 60-40 in beleagured Labor MP Craig Thomson’s seat of Dobell. The Weekend Australian also had Newspoll results on a range of issues to do with health policy, from which GhostWhoVotes presents the tables here, here, here and here.

UPDATE: Tasmanian gentleman psephologist Kevin Bonham offers the following historical context in comments:

OK so we have Gillard net sat -32 Abbott net sat -19. I did my bit by giving them both negatives.

The combined net sat of -51 for both leaders is now their worst and the =30th worst on record. The 29 worst readings are:

* Two Hawke-Peacocks from just before the 1990 election
* Four Keating-Hewsons from before the 1993 election
* Twenty Keating-Hewsons from after the 1993 election
* Three Keating-Downers

The record is -76 by Keating-Hewson in Sep 1993.

The “best” net sat of -19 is their worst and the =22nd worst on record. The 21 worst readings are:

* One Hawke-Peacock
* Six Keating-Hewsons from before the 1993 election
* Thirteen Keating-Hewsons from after the 1993 election
* One Keating-Downer

The record is -30 by Keating-Hewson in June 1993

There are only eight cases on record of the PM having a net sat of -32 or worse and a LOpp having a net sat of -19 or worse. These were all Keating-Hewsons (two before the election and six after).

Howard’s worst net sat in his successful term as Leader of the Opposition was -12.

Of boats and votes

Nielsen struck a blow for transparency yesterday by releasing comprehensive data for their polling on asylum seekers, featuring detail on the questions and how they were asked, breakdowns by state, location, gender, age and voting intention, and no fewer than eight tables cross-tabulating various results for the eight questions asked. They even went so far as to include the raw numbers they reached after weighting the responses for age, gender and location, not that this particularly tells us much.

The poll also deserves credit for posing thoughtfully crafted questions on a complex and contentious subject. No doubt taking inspiration from Murray Goot and Ian Watson’s recent paper on public opinion and asylum seekers, which noted that results had been heavily influenced by “the way questions are framed, the kinds of questions that precede these questions (and) the range of possible responses the questions allow”, the Nielsen report offered the following:

It is important to note that the results of opinion polls on this issue are more sensitive to the wording of the questions asked than for many other topics. This is because the issues are often emotional for some and complicated for all. Respondent knowledge on this subject is never complete. The task of adequately condensing complex options into fair but meaningful questions is also a difficult one.

The questions in this poll were stripped of their political context as much as possible. For example the ‘sent to another country to be assessed’ option was not offered in the context of deterrence, nor was any human or financial cost alluded to. It was not offered as Labor or Coalition policy (e.g. by calling it the ‘Malaysian solution’ or the ‘Pacific solution’).

The Fairfax papers asserted that the poll showed voters “at odds with both Julia Gillard and Tony Abbott and the perception that attitudes have hardened against asylum seekers”, and certainly the figures point to a more liberal attitude than the tenor of political debate would suggest. However, The Age gilded the lily a little with a graphic showing 60 per cent believed those assessed as genuine refugees should be allowed to stay in Australia permanently. It takes a bit of digging to appreciate that this excludes the 15 per cent who didn’t believe the asylum seekers should be assessed at all, having preferred that they be “sent out to sea”. The number supporting settlement in Australia was nonetheless a very solid 49 per cent, although there remained a combined 44 per cent in favour of the less liberal options of temporary protection visas (29 per cent support) and sending boats back out to sea (15 per cent). The same issue occurs with The Age’s figures for whether boat arrivals should be held in detention (64 per cent) or allowed into the community (32 per cent): putting the aforementioned 15 per cent back in (together with the 4 per cent “other/don’t know“), the results come down to 52 per cent and 26 per cent.

Regarding the treatment of asylum seekers on arrival, the results can be broken down thus:

22% – Allowed to live in the Australian community
12% – Detained in Australia, excluding children
17% – Detained in Australia, including children
4% – Sent to another country, allowed to live in community there
23% – Sent to another country and detained there
4% – Assessed for refugee status, no opinion on detention
15% – No assessment for refugee status: sent back out to sea
4% – Other/don’t know

And on their treatment after being assessed for refugee status:

49% – Settled in Australia
29% – Granted temporary protection visas
2% – Returned to country of origin
15% – No assessment for refugee status: sent back out to sea
5% – Other/don’t know

To those who are ready to junk the orthodox view on this subject, I would offer a few notes of caution. Certainly there was no majority in favour of assessing refugee status in Australia at the time of the Tampa episode, when Nielsen and Morgan polls had between 68 per cent and 77 per cent in favour of turning boats away. It is hardly plausible that so many of these respondents have had changes of heart that only 15 per cent now remain. What it likely shows is how the finer point of public opinion on this issue are shaped by the terms of the debate at the time. The symbolism in August/September 2001 involved boats being either allowed to land or held at bay by the military – only as the Howard government scrambled to effect its “Pacific solution” was the public alerted to the fact that the latter course only constituted half a policy. This may have led to a change in questions posed and answers given in opinion polls, but it doesn’t follow that there was a shift in underlying attitudes.

This leads to a point that occurs to me about the wording of Nielsen’s “sent to another country to be assessed” option: for many respondents, Nauru might not register as “another country” in the sense that Malaysia does, as it is perceived either as a dependency of Australia or too insigificant to qualify as a “country”. This option may accordingly have been interpreted by some as an invitation to sign on for the Malaysia solution. If Nielsen had at least added enough political context to allow for the restoration of the Pacific solution as a response option, the poll may have told a somewhat different story.

UPDATE (22/8): Crikey reports the latest Essential Research has Labor up a point on two-party preferred (to 56-44 from 57-43) and also on the primary vote, to 32 per cent, with the Coalition and the Greens steady on 50 per cent and 10 per cent. In other findings, 24 per cent support the health package finalised by government last month against 9 per cent opposed, with the great majority either indifferent (31 per cent said it would have little or no impact) or ignorant (28 per cent said they had heard nothing, 36 per cent little). Forty-seven per cent supported David Cameron’s suggestion that access to Twitter and Facebook be blocked during periods of civil unrest, with support varying as you would expect according to age and social media usage.

UPDATE 2: Full Essential Research report here.

Nielsen: 58-42 to Coalition

The Fairfax broadsheets report this month’s Nielsen result has the Coalition’s two-party lead at 58-42, from primary votes of 28 per cent for Labor (up two), 48 per cent for the Coalition (down three) and 12 per cent for the Greens (up one). Although a bad result for Labor by any measure, this is nonetheless an improvement on their 61-39 from Nielsen the previous month, and it maintains a trend evident throughout this year of Nielsen being a few points worse for Labor than all other pollsters. It accordingly sits quite well with the 56-44 Newspoll and what I am interpreting as a 57-43 result from the substantial Morgan phone poll released on Friday.

Julia Gillard’s personal ratings have risen slightly from the canvas: her approval rating is up four to 38 per cent with disapproval down five to 57 per cent, while Tony Abbott is down four on approval to 43 per cent and up four on disapproval to 52 per cent. Abbott maintains a 47-44 lead as preferred prime minister, down from 51-40 last time. Michelle Grattan’s report tells us Labor has a 52-48 two-party lead in Victoria, compared with a 55-45 deficit in last month’s poll, and that the Coalition lead in Queensland is 65-35, down from 68-32 last time. It should be remembered here that state-level results are from small samples. Further from Grattan:

Victoria … is also where Ms Gillard has a big lead as preferred PM – she is ahead by a hefty 51-40 per cent; in New South Wales she is ahead by 46-43 per cent. By contrast, in Queensland … Ms Gillard is behind as preferred PM 36-55 per cent. In Western Australia, she is behind Mr Abbott 33-57 per cent. Voters are disillusioned with the current leaders as economic managers. Almost three in 10 (29 per cent each) think former leaders Kevin Rudd or Malcolm Turnbull would be ”best to manage another economic crisis if one occurs”. Mr Abbott was rated as best by 21 per cent, compared with 15 per cent for Ms Gillard. A total of 58 per cent prefer a leader other than the current leaders. People remain strongly against the government’s carbon price, with opposition to it steady on 56 per cent and support at 39 per cent. Backing for the carbon price is highest among the Greens (79 per cent) and ALP voters (68 per cent); overwhelmingly, Coalition voters are opposed (82 per cent). More than a quarter of Labor voters are against the carbon price, and one in five Green voters. Regional voters are more likely to oppose the carbon price (62 per cent) than city voters (53 per cent).

UPDATE: Gordon Graham on Twitter:

#Nielsen best to manage another economic crisis if one occurs: Rudd 29%, Turnbull 29%, Abbott 21%, Gillard 15%

UPDATE 2: Full results from Nielsen here. The Coalition two-party vote is 58 per cent in New South Wales (down one on last month), 48 per cent in Victoria (down seven), 65 per cent in Queensland (down three), 61 per cent in South Australia/Northern Territory (steady) and 61 per cent in Western Australia (down two), remembering that the smaller states especially come from small samples. Labor has a better overall result on respondent-allocated preferences (56-44, a five-point improvement) than on the previous-election measure, and while I don’t recommend reading much into this, it’s interesting to note how different this is from Morgan, which has consistently had Labor doing worse on respondent-allocated preferences throughout this year.

UPDATE 3: Essential Research has the Coalition lead unchanged at 57-43 on two-party preferred, Labor has gained a point on the primary vote to 31 per cent, but the Coalition and the Greens are steady on 50 per cent and 10 per cent respectively. As with Nielsen, Julia Gillard’s personal ratings have rebounded from a diabolical result a month ago: most encouragingly for her, this is the first poll since June 14 (Newspoll and Essential results from the same day) in which she has led Tony Abbott as preferred prime minister, now leading 38-36 after trailing 37-39 last month. Gillard’s approval is up six to 35 per cent and her disapproval down seven to 55 per cent, while Tony Abbott is down two to 37 per cent and up one to 50 per cent.

Tellingly, 47 per cent of respondents say they think it “likely” there will be “another global financial crisis similar to the one that occurred in 2009” against 39 per cent who think it “about 50/50”, with only 8 per cent opting for “not very likely”. In that event, 40 per cent would more trust the Liberals to deal with it against 31 per cent for Labor and 20 per cent no difference, while 36 per cent would favour stimulus spending in response against 39 per cent who would not. For all that, 54 per cent believe the government has handled the economy well in recent years against 39 per cent who rate it as poor.

Morgan: 56.5-43.5 or 53.5-46.5 or 57-43 or 54.5-45.5 to Coalition

Roy Morgan has released two sets of poll results simultaneously, by way of confusing the hell out of everybody who doesn’t pay more attention than they ought to. One combines the results of the last two weekends’ face-to-face polling; the other is a phone poll conducted on Wednesday and Thursday nights from a big sample of 1006. Furthermore, Morgan as always publishes separate two-party results using both respondent-allocated preferences and preferences as directed at the previous election, and these continue the recent trend of being highly divergent.

For mine, the most significant of the resulting four sets of figures is the previous-election two-party measure from the phone poll, as this has been conducted with the same methodology and from a similar sample size as Newspoll. Unfortunately, this particular result does not make sense to me. Whereas the primary vote figures are slightly better for the Coalition than this week’s Newspoll – 49 per cent against 29.5 per cent for Labor and 12 per cent for the Greens, compared with 47 per cent, 29 per cent and 12 per cent – the previous-election two-party result is a fair bit worse: 54.5-45.5 compared with Newspoll’s 56-44. Applying the preference flows from the previous election (with 79 per cent of Greens preferences and 42 per cent of all other minor party and independent preferences going to Labor) produces a result of 57-43. That, as it happens, is the result Morgan has listed for its respondent-allocated measure – which is not to suggest they have run them the wrong way around.

The phone poll also comes with attitudinal questions, finding global warming scepticism at a plateau of 37 per cent after a steady increase over the previous three years; opinions on the carbon tax more or less unchanged since a month ago with support at 38 per cent and opposition at 58 per cent; and support for the Coalition’s policy of overturning the carbon tax down three points to 45 per cent with opposition up three to 48 per cent. There is also a flawed question on asylum seekers which invites respondents to choose between allowing boat arrivals to apply for immigration or subjecting them to the Malaysia solution, with no further options available. This finds 52 per cent appearing to support the Malaysia solution, contrary to last week’s Essential Research, but this is almost certainly because it’s the “tougher” of the only two alternatives presented.

The face-to-face poll shows essentially no change on the previous published result from the weekends of July 16-17 and July 23-24. Labor’s primary vote is steady on a relatively healthy 34.5 per cent, the Coalition is up half a point to 47.5 per cent and the Greens are steady on 12 per cent. The respondent-allocated two-party result is unchanged on 56.5-43.5, while the previous-election result is up from 53-47 to 53.5-46.5. This time, the latter figure is exactly where I would expect it to be.

In other news, draft federal boundaries for South Australia were published today: see the post below.

Matters South Australian

Three thereof:

• Draft boundaries have been published for a redistribution of South Australia’s federal electoral boundaries. Antony Green reviews the damage. This is not one of the more momentous redistributions of recent history: the changes are fairly minor, Labor has no seats on margins of less than 5 per cent, and one doubts the direction of the electoral tide at the next election will be such as to endanger the Liberal margins of Boothby and Sturt. This is just as well for the Liberals in the former case, as three amendments have cut the margin from 0.8 per cent to 0.3 per cent, while Christopher Pyne in Sturt has gained an extra 0.4 per cent buffer on his 3.4 per cent margin. Labor has had a 0.4 per cent free kick in its most marginal seat of Hindmarsh, pushing Steve Georganas’s margin from 5.7 per cent to 6.1 per cent, while Kate Ellis’s margin in Adelaide has been garnished from 7.7 per cent to 7.5 per cent. It’s a measure of Labor’s extraordinarily strong performance in South Australia last time that Kingston, Makin and Wakefield, which were all in Liberal hands late in the life of the Howard government, now have double-digit Labor margins. The redistribution hasn’t disturbed this, although an exchange of northern Adelaide Labor heartland for parts of the Barossa Valley has cut Nick Champion’s margin in Wakefield from 12.0 per cent to 10.3 per cent.

• The Advertiser has taken advantage of the leadership transition to conduct one of its occasional self-conducted polls of state voting intention. Conducted in the wake of Mike Rann’s semi-involuntary retirement announcement, this has the Liberal lead narrowing from 58-42 in late June to 54-46, from primary votes of 32 per cent for Labor (up seven) and 44 per cent for the Liberals (down five). However, given a margin of error of around 4.5 per cent, and suspicions that the paper’s polling expertise might not be all that great, caution should be exercised before diagnosing any kind of durable Labor revival. Nonetheless, a substantial 57 per cent of respondents say they expect better from incoming Premier Jay Weatherill than they have been getting from the incumbent.

• The date for the Mike Rann-Jay Weatherill leadership transition, in case you missed it, has been set for October 20.