Poll Roundup: Coalition Polling Improves, Abbott's Doesn't

2PP Aggregate: 54.6 to Labor (-1.1 since last week)
Labor would win election "held now" with a massive majority

Quite a lot has happened in federal politics in the last fortnight.  Shortly after the last round of polls (most of them disastrous for the Coalition) were released, there was a leadership spill attempt which Tony Abbott survived by the unconvincing margin of 61-39, even without any declared challenger.  This was followed by various declarations from the PM that he would be changing his spots, and then a steady flow of medium-level gaffes and leaks that mostly suggested otherwise.  None of this shows any sign of stopping, with new revelations about internal problems in the party (often arising from the Loughnane/Credlin conflict of interest) now arriving more or less daily.

Abbott was even the subject of a major article in The Australian, alleging that he had informally proposed sending 3500 Australian troops into ground combat against the so-self-labelled "Islamic State" in Iraq - an allegation that at this stage hasn't been verified by any named source.  Abbott's denials of the claim seemed somewhat coded, but maybe Abbott just wanted to be sure he didn't deny saying something he might have said in jest or as a brief thought-bubble. The polling fortnight ended with Abbott again focused on terrorism issues ahead of a major statement yesterday.

This week's polling

This week has seen a substantial improvement in the Coalition's 2PP polling.  When the Newspoll result (53-47 to Labor compared with 57-43) came out many observers, including me, expected that the change was almost entirely random poll-to-poll noise, but while the size of the shift back to the government wasn't replicated by other polls, its existence was.  Essential moved one point from 54:46 to 53:47 and Morgan moved two from 57:43 to 55:45 (by last-election preferences).  Considering that Morgan leans to Labor by about 1.5 points (my own estimate of the average gap has now come down to 1.3), that makes its result quite similar to the other two.

So we have a bunch of 53-ish results a fortnight after a bunch of results in the 54 to 57 range, which in turn came after a period of sparse polling.  This is a rather volatile polling picture and my aggregate still gives a lot of weight to the nasty stuff from 2-3 weeks ago compared to the three milder results this week.  But the combined weight of a 4 point 2PP swing in Newspoll, a 2 point swing in Morgan and a 1 point swing in Essential is enough to knock 1.1 points off Labor's lead.   There's very likely a large random-noise element in the Newspoll shift (with last Newspoll being a bit high for Labor and this one likely to be lower than reality) but to assume it is all random noise is not supported by the behaviour of other polls.  Here's the smoothed tracking graph:

Before I move on to some issues specific to Newspoll, let's assume this is strong enough evidence of at least a probable shift in voting intention back to the Coalition from two weeks ago and consider why it is occurring.


A shift around the time of a spill situation was also seen when the Labor leadership was spilled in early 2012.  The thinking at the time was that voters were expecting Julia Gillard to be replaced with Kevin Rudd and were pleased to see Labor moving in that direction.  When this didn't happen immediately, so the story goes, Labor's polling fell off a cliff.  It's possible there's something of this in the current polling - that Coalition voters displeased with Tony Abbott are thinking he will be gone soon enough and on that basis happier with their party.

It's also possible in such cases that the temporary factors (which can include bouncy polling results) that lead to a spill situation fade away, so that it's not so much a case of the government doing something right, but more of a case of it being further away from something that it did wrong.  The Prince Philip fiasco is a candidate here.

It is being argued that the current focus on national security has helped the Coalition improve, but past Coalition poll bounces linked to national security have generally been for defined events and not just for government waffle and tough-talk.  They have also tended to spill over into the PM's personal rating, which this particular wave of terror-talk hasn't.

More on Newspoll

It's likely the real 2PP movement in Newspoll's samples has been stretched by rounding, and my aggregate only counts it as a 3.5 point shift.  Last fortnight's primaries implied a 2PP most likely on the upside of 43 for the Coalition, but this fortnight's (38 each for the majors and 12 each for Greens and Others) imply something probably rounded up to 47.  The primary vote swing was only three points.

This is the first time the Coalition has recorded a four-point 2PP Newspoll move in its favour in this term of government, but in the long-term history of Newspoll, it's not that rare an event at all.  For instance the second Gillard government (2010-3) experienced five such moves in its favour (four of three points, two of five) even before Gillard's replacement with Kevin Rudd.  In those cases, more than half the points gained were, on average, lost in the subsequent poll.

A rather strong suggestion of Newspoll bouncing is the 16-point netsat drop for Opposition Leader Bill Shorten, who has gone from a +2 netsat last time (42-40) to -14 this time (35-49).  This is Shorten's worst net reading and his highest disapproval score, each by three points, thus far.   A 16 point netsat drop for an Opposition Leader in just one fortnight is very rare, with the last such case five years ago. Prior to this fortnight, Shorten's ratings had set all-time records for their relentless mediocrity, and there's nothing he's done in the last fortnight that should have caused him to be viewed any differently.

Some commentators have tried to seize on examples of Shorten standing for nothing or giving bland answers to questions, but that's just the same impression they could have had in any other period; there's no reason for it to suddenly make a big difference now when no-one is paying Shorten much attention. It's possible there could be statistical noise affecting Shorten's rating without affecting the party primaries, but the more likely explanation is that this sample contained more rusted-on Coalition voters (who tend to dislike Shorten) or fewer rusted-on Labor voters (who tend to like him) than the last.

Abbott's own personal ratings indeed showed no meaningful improvement, with his dissatisfaction rating at 68% for the second fortnight in a row, and his satisfaction rating up just one point to 25.  The only previous PM to record two or more consecutive netsats of -40 or worse was Keating, who recorded six in a row in 1993 during fallout from that year's "horror budget".  My suspicion is that Abbott's public standing might really be even lower now than last fortnight, but that the figures don't reflect that because of random movement in the partisan leanings of Newspoll respondents from sample to sample.  There was a decline in Bill Shorten's lead as "better Prime Minister", from last poll's massive 18 points to eight (still a very large lead for an Opposition Leader), but again it is unlikely that most of that shift is real.  The truth about voter perceptions of Shorten is likely somewhere between the last two polls.

With what is known about the strong relationship between personal ratings for most PMs and the 2PP picture, it may seem surprising that Abbott is still polling such terrible personal ratings in a poll that has a fairly modest Opposition 2PP.  Indeed, this is unusual, especially by Abbott standards - until this he had one of the tightest relationships between his own netsat and his party's standings of all PMs.  Here's a long-term scatter-plot of the PM's net satisfaction rating against the Opposition's 2PP:

The yellow dot on the left is Abbott's last Newspoll and the red dot is this one.  Both are among the ten worst netsats recorded in Newspoll history, but last week's is at least somewhere near the best-fit line.  This week's is a more serious outlier, but it does have some company: in two mid-1992 polls, Paul Keating polled similar netsats while the Coalition's 2PP lead under Hewson was only 52:48.

Abbott's Strange Response

Tony Abbott's own response  (link probably paywalled) to being "alive" in the polls was, incredibly, "Well look, Karl, you know what it's like to be young and vigorous and at the height of your powers and that's exactly how I feel".  For a PM to be self-boosting on the back of one of the ten worst personal ratings in Newspoll history, a still sickly primary vote situation and a pretty horrible list of personal attribute polls (see below) is all rather strange and would normally just seem desperate or delusional. (Even more so if the Turnbull-anticipation theory is true.)  Previous PMs might have either refused to talk about the polls at all, or else indicated calmly that while they welcomed signs of improvement, they realised there was still a long way to go.  What the PM seems to be conceding here is that, for instance, a 59-41 to Labor with a -55 netsat could have been the end of him immediately, and he considers his own position so fragile that he truly believes that his tenure is at the mercy of single bouncy opinion poll results.

That's not so surprising when Fairfax papers are reporting that numerous unnamed ministers consider the PM to be on a short string with removal likely around June unless there has been a major and sustained polling recovery - or unless there is a major stuff-up sooner.  The Credlin/Loughnane situation is clearly fuelling internal dissent with the PM and may be doing so more than just the bad polls, but I suspect that its impact on how the public views the party is quite minor, and that the real problem is perceptions of Abbott himself that will be very hard to shake no matter how his staffing changes.

Newspoll Personal Attribute Scores

This week, Newspoll also released a big range of personal attribute scores for Tony Abbott and Bill Shorten. It is some kind of Murphy's Law of polling that whenever a pollster does one of these rare special efforts they do so in a week in which their voting intention sample happens to look a bit odd.   Here is a full list of the results with comparisons with past leaders.  In general Abbott's results are pretty bad (in some cases terrible) while Shorten's are middling.

* More capable of handling the economy? Abbott 45 leads Shorten 37.  This question is a traditional strength area for incumbent Prime Ministers, especially Coalition ones, so this should be seen as a poor result and not a good one.  John Howard had an average lead of 26 points on this question during his career, and only led by less than that during some periods of bad polling in 1998 and 2001.

* More capable of handling national security?  Abbott 51 leads Shorten 31.  This question started being polled in 2001 and its history is dominated by Howard government ratings (Howard always led, by an average of 26 points.)  The question seems like normally Coalition territory - Kevin Rudd did have large leads over Nelson and Turnbull, but Abbott had large leads as Opposition Leader against Julia Gillard.

* More capable of handling asylum seekers?  Abbott leads 51-32.  He led Gillard 48-32 in 2010 and 48-26 in 2012 so not much to see here in a question on a Coalition strength area.

* More capable on health and Medicare?  Shorten leads 56-30.  Since this question was added in 2005 ALP leaders have always led Liberal leaders, by an average of 17.4 points.  Nothing to see here given the obvious influence of the Liberals' unpopular attempts to impose co-payments.

* More capable on education? Shorten leads 53-33.  This one, also added in 2005, seems to show both a bias to the government of the day and a stronger bias to the ALP.  Kevin Rudd had larger leads over Howard in 2007, but Howard led Kim Beazley not long before that.  Again, not much to see here.

* More capable on climate change?  Shorten leads 55-24.  The lead for Shorten is wider than Gillard's leads over Abbott the only two other times the question was asked (19 points in 2010 and 9 points in 2012).  This could be seen as Abbott being increasingly out of touch on the issue, but could also represent some moderate voters thinking Gillard was doing too much on the issue during the "carbon tax" days.

* Do they understand the major issues?  52% yes for Abbott, 68% for Shorten.  In the history of this question being asked (back to 1999) Abbott's rating is the worst by five points.  John Howard scored 57% in March 2001 and Simon Crean the same in July 2002.  Abbott's ratings when he was Opposition Leader ranged from 66 to 73.  Shorten's 16 point lead on this question is the highest for an Opposition Leader.

* Are they arrogant?  77% for Abbott, 46% for Shorten.  77% might sound high, but Paul Keating beat that every one of 11 times the question was asked, with a career average of 83% and a high of 87.  Since then only Kevin Rudd (71 just before the 2013 election) has broken 70.

* Do they care for people? Abbott 55, Shorten 77.  Again, Keating used to always poll worse on this one than Abbott has just done (his scores ranged from 37 to 53).  Since then, results as poor as Abbott's were only polled by Howard during his March 2001 slump, and by newish Opposition Leaders.

* Are they in touch with voters? Abbott 33 Shorten 63. This question has only been asked since 1999, in which time the previous worst rating was also 33, again polled by Howard in March 2001.  Shorten's result is quite good given that he has not been leader for all that long, but Kim Beazley in his first  go as Opposition Leader consistently polled better, and other Labor leaders like Mark Latham (peak 75%) and Kevin Rudd (peak 80%) also polled much higher scores than Shorten.

* Are they experienced? A rather odd one this, first asked in 2007.  Tony Abbott is somehow eight points less experienced (72%) now than when he was Opposition Leader in 2010.  Julia Gillard also became less experienced as she became less popular. Bill Shorten scores 66%.

* Are they decisive and strong?  Abbott 57 Shorten 56.  A surprising one this: John Howard was often seen as quite indecisive during his first term, polling as low as 45, but not thereafter, ending his career in the low 80s.  Both Rudd and Gillard as PM had better ratings on this question than Abbott the few times it was asked for them, as did Keating the many times it was asked re him, while Shorten as with many other questions rates above Crean and Nelson but generally below other Opposition Leaders since 1997.  Opposition Leaders tend to poll poorly on this question when their time is up.

* Are they likeable?  Abbott 40 Shorten 64.  Here Abbott's score is slightly above the Keating average of 38.7 (Keating ranged from 32 to 44) but way behind all other leaders since (Howard's worst was 49, equalled by Crean and Nelson who were not long established leaders when the question was polled for them.)

* Are they trustworthy?  Abbott 43 Shorten 59.  Abbott again shades only Keating, who ranged from 26 to 37 on this score, while trailing all other leaders bar a 43 for Howard in March 2001 and a 42 for an early Simon Crean.  Gillard in April 2012 polled 44. Shorten's is the best for an Opposition Leader since Rudd, but that's against few ratings for a pretty unpopular field.  It's still not bad - eg, John Howard was below Shorten's current level throughout his last four years.

* Do they have vision for Australia? 61 apiece.  Both generally trail all other leaders since 1999, except for Simon Crean and Brendan Nelson (the latter should be cut slack for not being in the job very long.)

* Are they ridiculous?  This is a question pollsters don't ask, but they should!  The point at which a leader becomes laughed at or not listened to is widely thought to be a surer indicator of their impending demise than the point at which they are merely hated or feared.  If this is true, it would be very useful to have data on it ... however disrespectful this may seem!

Some Essential issue questions
Essential had issue questions showing among other things that its respondents:

*  are generally less concerned about economic issues (especially the price of petrol, electricity and food) than last October but slightly more concerned about the national debt

* if with jobs, tend to expect them to become less secure

* interestingly, see Australia's relations with any named foreign nation as less important than they did in the last few years

* distrust "Tony Abbott and the Liberal Government" on international relations (to a remarkably strong degree by Essential standards - 22% have "a little trust" and 40% "no trust"with only 5% undecided)

* rate the government's handling of Indonesian relations fairly poorly, but not enough to suggest the apparently impending executions of two Australians are a major factor in this.

Last week, late-breaking results on Malcolm Turnbull not surprisingly showed he was much better viewed now than during his unhappy time as Opposition Leader.  Results on the Liberal leadership spill had voters a little cool on replacing Tony Abbott immediately (with 39% preferring this to 50% thinking he should at least get another six months) but this most likely plays into the idea that since Abbott has survived the spill motion, it would be silly to immediately boot him without giving him any chance to turn matters around.

I may have another instalment of this series next week following Ipsos, if I manage to find enough spare time while fighting the curse of the "AGM Season" that afflicts all those who are on too many voluntary club committees!