Turnbull PM: Not Likely Any Time Soon (Poll Roundup)

2PP Aggregate: 53.1% To Labor (-0.9 since mid-May)
This article will be updated through late June to about mid July.  Scroll to the bottom for the latest additions.

======================================================
Just enough results of interest have accumulated to start another roundup of the federal polling information that is out there.  But there is still not a vast amount to look at since the previous instalment, so this article doubles as my thoughts on something else that was causing rather a lot of media flutter last week.  First, the basic polling:

Is The Government Recovering?

This week's polls might be taken as some mild sign that the government is starting to recover from the bad polling recorded following the federal budget.  Newspoll is back a point to 53:47 and this week's Morgan sample (54.5 by last-election preferences, 55.5 respondent-allocated) is similar once due allowance is made for Morgan's house effect.  Essential failing to move (it's still on 54) of course means very little.  (In passing, I note that it is interesting to see the massive 17% support for Others in this Newspoll being described as "support for independents and micro-parties".  Palmer United are certainly not a micro-party anymore, and much of the Coalition's recent loss of primary support seems to be going directly to PUP.)



However, there's a catch in the above, and it's in the "this week's Morgan sample" bit.  With the exception of the aftermath of the Budget, Morgan have normally released combined data for two weekends.  In this case, they've chosen to split the data thus producing two separate polling results.  Last week's was rather drastic: 59% 2PP to Labor by last-election preferences (60.5% - that is not a misprint - respondent-allocated.) If the two weekly samples were merged then Labor would be at either 56.5 or 57 for the fortnight, which even accounting for Morgan skew would be enough to cancel out Newspoll on the question of whether the Coalition are improving.

The Morgan press release attributes the 2PP going down four points in a week then up five points the next to the damaging impact of leadership speculation surrounding Malcolm Turnbull.  The much more likely explanation is that this was mostly random sample error, and that last week's sample was overly generous to Labor.  Indeed, even after back-adjusting my estimate of the trend to take into account Morgan's result last week, and correcting the reading for Morgan's house effect (with my estimate thereof increased because of this result), the Morgan result for last week still comes out slightly further off trend than its margin of error, making the result a probable (albeit marginal) rogue.

It is simply not the case that leadership speculation is always damaging in the polls - for example Labor rose both in the immediate lead-up to Julia Gillard's removal of Kevin Rudd and in the leadup to Rudd's failed challenge in early 2012.  There are two very good reasons for speculation to not be very damaging in the current situation - firstly there is no evidence of actual Coalition disunity over the leadership, and secondly given that Turnbull is far more popular than Abbott, media speculation about him taking over is hardly likely to scare voters away.

Even probably-rogue data are still data, and back-including last week's Morgan sample in last week's aggregate blows last week's final figure out to 54.9% to Labor, but it hasn't taken the Newspoll and the new Morgan sample long to take that back down to 53.9%, the current value.  Apart from the one-point flicker upwards caused by the Morgan subsample, there's been no real change over the past five weeks.  Here's the spiky (unsmoothed) tracking graph:

And here's the smoothed version:

I have made some minor changes to my aggregate methods to cater for (i) evidence now indicating that the size of Morgan's house effect is more like 1.5 points again and (ii) the inclusion of Morgan multi-mode polls taken over one week with smaller sample sizes than normal.

If the Morgan sample was just completely rogue (or if there really was a spike last week that is now gone) then following weeks may show improvement for the Coalition as its influence washes out of the system.  However, while many sources are taking the one-point improvement in Newspoll and Morgan over a fortnight as indications that the worst is over, my take is it's still too soon to tell.

There is certainly no sign of improvement for Prime Minister Abbott; his latest Newspoll netsat (-31, 30% satisfied, 61% dissatisfied) is his worst as Prime Minister and his worst personally since recording identical figures in late November 2012.  Something has been made of these figures being so close to the -34 (28-62) recorded by Julia Gillard just prior to her removal; what is forgotten here is that Gillard recorded ratings worse than -31 eighteen times over a span of two years before being removed.

In the history of Prime Ministerial dissatisfaction scores, figures in the high 50s and low 60s aren't all that rare but ratings of 63 and above are.  Here is a graph showing how often each dissatisfaction score has appeared in Newspoll history.  Abbott's current rating is highlighted in red.

Yes, these are bad results for the PM, but he's not quite cracked the elite zone of elector dislike yet.

Roy Morgan also released one of their rare leadership phone polls, showing Abbott with a net rating of -25 (by comparison, in this infrequent series, Gillard got down to -38 and Abbott's worst as Opposition Leader was -35).  But perhaps more attention came from the two polls in the last few weeks to canvass voter preferences for the two party leaderships.  And that brings us to Mr Turnbull:

Turnbull Still Vastly Preferred ...

In the last few weeks the political question I've been asked most often is whether I think Malcolm Turnbull is about to try to roll Tony Abbott and become Prime Minister.  I say this because I have been asked it twice! One always hesitates to give a definite "no" to questions like this, because that would involve reading the minds of politicians and worse still parties .  Nonetheless, this article argues that there would be no tactical sense in making Turnbull PM right now.

The case for belief that Turnbull might replace Abbott soon rests on the following facts:

1. The government is unpopular. For the tracking see above. The recent Budget was the second-worst received in the last 28 years and its impact is still being felt in polling more than a month later.

2. Tony Abbott is currently very unpopular.  Also as tracked in the previous article, Abbott's Newspoll net satisfaction recently fell 23 points in six weeks (and from a negative base as that); there are only nine cases in Newspoll history of a greater fall in a similar period.  Two recent Newspolls have seen Abbott with a netsat in the bottom 9% of all PM ratings in the Newspoll era, and Nielsen recorded him with one of the worst results for an incumbent PM ever.  Abbott's unpopularity carries over into preferred Prime Minister polling, with Bill Shorten (whose own ratings have been generally nondescript at best) leading him on all recent polls in spite of the large skew of most such polling in the incumbent's favour.

3. Tony Abbott has never been popular.  Unlike John Howard, who experienced some short periods of severe unpopularity but had a career median netsat of +8, Abbott has not shown any ability to be lastingly popular either as Prime Minister or before that as Leader of the Opposition.  This is important because there is a strong and apparently causal link between the rating of the PM and the government's two-party preferred vote.

4. Malcolm Turnbull continues to be very popular.  Preferred leader polling while Abbott was Opposition Leader consistently showed that more voters nominated Turnbull than Abbott as their preferred Coalition leader, and at times even that more Coalition voters did so.  Smorgasbord-style leader polling last week by Morgan showed not all that much has changed; Turnbull leads Abbott 44 to 15, with Abbott's 15% only a point above the score recorded by Julia Gillard just before she was removed from office.  Essential has the gap at 31:19.  (I note in passing that although Morgan's list differs from Essential's only by including Barnaby Joyce, who scores 5%, Essential has a "someone else" score of 19% while Morgan's is only 1%. I would bet that most of Essential's 19% couldn't actually say who else they had in mind.)

5. Malcolm Turnbull has recently denied leadership speculation in very strident terms.  As we know from Yes Minister a leadership challenge isn't serious until it is officially denied, and we saw plenty of this from both Gillard and Rudd during the last term.  To some, any denial Turnbull cares to offer is taken as confirming his intention to run for the job, as would be any confirmation he might offer, and also any refusal to comment.

Together with the known facts there are more subjective and ambitious claims made: general claims that the Coalition are just not in control of their fate under Abbott, views that voters are locked in to negative perceptions of this Government, and suggestions Abbott is seen generally as especially un-"Prime-Ministerial" or offensive to basic Australian values.  I'd say it's still way too soon to say about all of those things.

Leadership speculation came to a head with not only the above-noted polling, but also comments by Turnbull attacking Andrew Bolt and Alan Jones for talking up possible leadership fighting.

...But So What?

These are the reality-checks that I think should be imposed on any attempt to translate Turnbull's recent good polling (and Abbott and his government's poor polling), plus the recent speculation, into a real chance of an imminent Turnbull takeover or even formal challenge.

1. The government's polling isn't actually that awful yet.  Since the mid-1940s an Opposition lead of around 54:46 or greater has been seen in exactly 50% of parliaments.  Of these, seven out of 13 have ended with victory to the incumbent government.  Furthermore with the ancient exception of the Chifley government in 1949 (which was never far behind in polling at all), all governments that have lost were at some stage much further behind than the Coalition are now (as also were a few governments that won).  So if this is as bad as it gets for this government, it's likely it will be re-elected.

2. The government's polling must be seen in context.  The government has chosen to introduce some very harsh budgetary measures with the intention of taking polling pain early in its term but being able to point to benefits of its actions by the time of the next election.  Whatever might be said about whether this sacrifice of voting intention is brilliant or bungled, it is not the same thing as a government losing support because of a scandal or incompetence.  A hit in the polls is the expected result of such a strategy and it could have easily been bigger.  To execute a strategy that was expected to lead to being well behind in the polls, to then be well behind in the polls, and to argue that this meant the strategy didn't work and it is time to change leader, doesn't make any sense at all - unless you never believed in it in the first place.  If what the Coalition is attempting with this Budget has indeed failed, then it will be some time before they know it.

3. Turnbull is not the preferred leader of Coalition supporters. When it comes to preferred-leader questions between candidates from the same party, the normal advantage for incumbents can become reversed.  Supporters of the opposing parties are quite likely to support anyone but the current leader, largely because they will often have developed a tribal dislike of the current leader (especially if that leader is not perceived as moderate.)  However, a government should really only care about the opinions on leadership of those voters who are open to voting for it (but not certain to do so).  We can't easily identify from these polls which way those voters whose votes actually hang on the leadership question have gone, but most of those ALP and Greens supporters who prefer Turnbull wouldn't vote Coalition under any leader.

4. Turnbull's polling reflects rose-coloured glasses.  We saw this movie before when Kevin Rudd was reinstalled. The Labor 2PP jumped around six points, Rudd's leadership ratings were mildly positive, and he led Tony Abbott as preferred PM by 22 points.  By election day voters had been reminded of much of what caused them to fall out of love with Rudd in the first place.  While the replacement of Abbott with Turnbull would in itself lead to many voters who dislike Abbott being greatly relieved, Turnbull would still be a Liberal Prime Minister and would not be able to make policy for the whole party.

5. Turnbull isn't anywhere near having the numbers yet. I'd be surprised if the number of Liberal MPs who would support Turnbull in a leadership ballot right now was even in double figures.  This is not just because of the disappointments of Turnbull's spell as Opposition Leader (including the Ozcar bungle) but also because of his position to the left of the party.

6. The Liberals don't roll incumbent PMs.  Well, they did in 1971, but that was different.  John Gorton wasn't the Liberals' first-choice leader but was forced on them by the Country Party.

7. The timing is just wrong. This, I think, is the big one.  Even if it made complete sense to install Turnbull as leader for the next election with Abbott clearly incapable of winning it, it would be better to offer a new leader as a relatively fresh prospect with, say, eight months in the job, rather than installing him now and allowing two years for voters to become disillusioned with the switch.  It would certainly be necessary to give Abbott enough time in the job for it to be beyond doubt that Abbott had failed.

It is possible to make some of these points a little too strongly.  For instance, Turnbull's infamous loss of popularity over the Ozcar bungle was partly because it was an unfair attack on Kevin Rudd, who was still very popular at the time.  Also, while much is made of Turnbull being a failed opposition leader, the same once applied to John Howard, who went on to be a very long-serving Prime Minister.

Alternative Explanations

I believe that those who saw the Turnbull comments as representing real leadership instability missed a number of relevant points.  Turnbull's popularity and history mean that he is not subject to the usual requirements of being a team player.  A sufficiently provoked Turnbull could do the Liberal Party immense damage as a backbench critic or if he left the party altogether.  Because of this, Turnbull is free to speak out and do as he will so long as he avoids doing anything utterly disloyal.  He may be simply letting off steam by expressing his sincere view of Bolt and Jones.  But even if the way in which he did it is a deliberate exercise in strengthening his own brand as a different, more authentic, politician, any leadership thoughts involved in it would more likely be long-term ambitions, premised on the hope of Abbott falling under the proverbial bus, rather than designed to trigger it.

Much of the analysis of the Turnbull situation coming from the left online seems driven more by a desperate hope to be rid of Abbott than by any rational evidence.  Yes, even mediocre polling sometimes drove the ALP into a leadership lather during its previous terms, but it doesn't automatically follow that the other side will be as foolish.  The intriguing thing is that the appeal of Turnbull to many in the left is so strong when there is so little by way of policy differentiation to show for it - differences on climate change and symbolic crumbs on the republic and same-sex marriage are enough to maintain a political cult that portrays Turnbull as much more left-wing than he actually is.

Updates on this article will follow below over coming weeks (not sure how many weeks yet) as more polls arise.
======================================================================

23 June Nielsen: What is believed to be the second-last ever Nielsen comes up with a very middle of the road 53:47, and with this further suggestion that post-budget anger could be fading a little, my aggregate has come down to 53.6.  But there is not much improvement for Tony Abbott who is on a -25 netsat (+3), still a very bad rating by Nielsen standards, while there is no significant shift in views on the fairness of the Budget either (a two-point drop in those finding it unfair from 63 to 61).  Nielsen, unlike Newspoll, still has Bill Shorten just in positive territory (+1).  Nielsen does also show a narrowing in Better Prime Minister, but from a very high base and to a still impressive 7-point lead for Shorten.

Nielsen also gets into the territory of this article's major theme with the following largely predictable results:

* Turnbull 40 Abbott 21 Hockey 11 Bishop 11
* Turnbull 62 Abbott 30
* Among Coalition voters, Abbott 59 Turnbull 39

The last one is the most significant.  The Essential and Morgan polls only polled the Coalition voter question in smorgasbord style and not as a direct Abbott-v-Turnbull question.  Essential and Morgan found Abbott's lead over Turnbull among Coalition supporters at 16 and 6 points respectively, but Nielsen shows that with the minor candidates removed, the gap if anything stretches.  Incidentally, the various leader polls have also showed support for Shorten as ALP leader is pretty soft - Nielsen has him leading Albanese and Plibersek 25-19-17, while Morgan had him leading Plibersek and Albanese 22-16-15.

More on Nielsen: An aspect of the Nielsen attracting some attention is the large-state breakdowns, which have Labor leading 65:35 in Victoria and the Coalition up 54:46 in NSW.  I would not take either of these remotely seriously given that the BludgerTrack state aggregates have been showing Labor up 57:43 in Victoria and 53.8:46.2 in NSW.  Of course the Nielsen samples are evidence that the gap between the states is actually somewhat wider than 3.2 points, but it would simply not suddenly blow out to 19 points with the national 2PP so little changed.  Both the Nielsen breakdowns are extreme results that can't be relied on; luckily for the overall result, they run in opposite directions and hence cancel each other out.

24 June Essential: Well this is a teensy bit surprising; Essential, which is notorious for lagging trends, has returned the Coalition's best result in four weeks with a 52:48 lead to Labor, bringing my aggregate down to 53.3, the Coalition's best result for six weeks.  This is the third two-point move in Essential in this term, the first two coming at the start and end of an excursion to the Coalition side back in March.  In general Essential has had a mild (about half a point) average lean to the Coalition in this term, but I think it is more that Essential has leant to the Coalition because it has lagged trends, rather than the other way around.  There's not much doubt now that we are past the peak of budget-blowout sentiment, though something else that is perhaps distinct going on here is that the popping of the recent Palmer bubble has seen votes return to the Coalition.  A nervous question for Labor could be: if there is a move back to the Coalition, where does it end up?

Essential also has some illuminating findings on perceptions of the holding of the balance of power by the Greens and PUP and the microparties (I do wish the question hadn't implied PUP is itself a micro-party).  Basically Labor voters think anyone holding the balance of power is good (partly because they hold the Coalition to account), Coalition voters don't get the idea of non-majority government (nothing new there), and Green and PUP voters like it if their lot has the balance of power but don't trust anybody else with it.

The poll also features a fun leadership attribute cage match between Christine Milne and Clive Palmer, with Milne notching up a 13-2 win; Palmer coming out ahead only on "visionary" and (by one point) "understands the problems facing Australia".  Milne's scores on some of the more negative attributes are much lower than might have been expected from some of the common criticisms of her style.

June 30: Newspoll Dampens Recovery: Just as it was starting to look like the Coalition could be recovering rapidly, a 55:45 Newspoll to Labor has put paid to all that.  While Labor's aggregate position now (53.5, the reverse of the last election) remains slightly lower than the 54 after the budget, that's only half a point recovered over six weeks.  There was also a new Morgan, which was 57.5 to Labor by respondent-allocated preferences but only 54.5 by last-election preferences, which comes out to around 53 based on evidence on Morgan's house effect.  The aggregate is no longer affected by the roguish one-week Morgan of three weeks ago.

Abbott has polled another -31 Newspoll netsat (31-62).  As noted above, 62 is the highest dissatisfaction score that is still moderately common; one more point and he will equal his worst ever as Opposition Leader and reach the top few percent of historical disdain.  Shorten's better-PM lead is back out to 10 points (the same as four and six weeks ago) and Shorten's own netsat, of -7, is slightly negative for the seventh time in the last eight Newspolls.

Here's the smoothed tracking graph; at this rate barring some disaster for Labor or unexpected event, the Coalition doesn't look like getting back in front for a while.

It is also worth noting that while Palmer United have been going gangbusters in recent Queensland polling, a recent ReachTEL poll has shown their leader to be in a losing position in his own seat of Fairfax.  Too much noise on the national stage and not enough electorate work, perhaps?

1 June: ReachTEL and Essential: I missed the ReachTEL taken last Thursday and released yesterday but have now added it.  The 2PP was released as 53:47; I make it 52.7 to Labor off the decimal primaries.  After a run of Labor-friendly readings relative to the aggregate from ReachTEL this one is actually slightly on the Coalition side of the aggregate line, and for the reason of that recent form it has pushed the aggregate down to 53.1.  The poll does not show that much movement in Tony Abbott's leadership ratings (27.6% good or very good, 15.5% satisfactory, 56.8% poor or very poor).  On my ever reliable ReachTEL-to-Newspoll converter it comes out to a -21.5 netsat equivalent, while Bill Shorten is at -0.5. A question on the influence of Clive Palmer showed a 35-35 split on whether his was a good influence, with Coalition fans annoyed they don't have a majority and Labor fans thinking that the enemy of their enemy is their friend.

Today's Essential again came in at 52:48.  Essential released illuminating findings on the approval of various Government ministers, with Turnbull leading on a +13 netsat followed by Bishop +3, Brandis +2, Morrison -1, Hunt -11, Hockey -12 and Pyne -18.  A shame Kevin Andrews wasn't canvassed.  The full report has a breakdown showing that Hockey rates very well among his own party's voters despite his poor overall result.  The polarisation shown with Hockey is likely even more extreme with Abbott, whose most recent Essential approval rating was higher among Liberal voters than that of any of these ministers, despite his overall netsat being -23.   Naturally the higher the profile of a controversial party figure, the more likely such a polarised reaction is.