Poll Roundup: Bad Start For The Second Turnbull Government

It's early days, but here we go again ...

2PP Aggregate: 51.5 to Labor (+0.6)
Labor has led since aggregate resumed

Welcome back to Poll Roundup, a series which looks at the aggregated state of the federal polls, leadership ratings and polling snippets of interest.  Nearly three months after a narrow victory in the 2016 federal election, the Turnbull government's progress has now been measured in three Newspolls and eleven weekly Essential Research readings.  It's not a great start for the returned regime.  It hasn't been ahead in even one of those, and is now clearly behind.  Indeed, when the primaries from the early polls are converted using 2016 preferences, a case can be made that Labor has been leading in the lot.

If we look at the two-party-preferred votes for the newly returned government, the Newspoll sequence of 50-50-48 is exactly the same as what the Gillard government received in 2010, in its first three polls after an even narrower escape, on its way to three years of generally wretched polling and, eventually, a heavy defeat.



Where We're At

My aggregate, restarted last week, at present has the government trailing with 48.5% of the two-party-preferred vote.  The polls still in the aggregate at the moment are the current Newspoll (48:52, aggregated at 47.7 after considering the primaries), the previous Newspoll (50:50 by 2013 preferences, aggregated at 49.6), this week's Essential (48, aggregated at 48.2) and Essential from two weeks ago (48 by 2013 preferences, aggregated at 47).

In the last parliament both Newspoll and Essential often appeared to lean to Labor compared to other polls.  The new Newspoll consistently ran about 0.7 points below the aggregate for the government, while Essential ran with the aggregate over the term as a whole, but as much as 1.2 points below during the time after Malcolm Turnbull took the leadership.  Neither of these polls showed any skew to Labor in their final pre-election poll, however.  I'm adding a very conservative 0.4% to the Coalition's total while the aggregate contains only Newspoll and Essential, but even if I added (say) a point, the headline picture wouldn't change.  The government is still behind.

There is a Not-A-Poll in the sidebar where you can vote on when in this term the Turnbull government will take the 2PP lead on this site.  At present "they won't" has a hefty majority, but this was also the case when I started the same poll in December 2014.  In that case the Coalition did retake the lead, but only after getting rid of their Prime Minister.

A History Of Honeymoons: Honeymoons Are History

Once upon a time it was normal for governments - whether new or re-elected - to record polling better than their election results for at least the first couple of months after their re-election.  In the case of new governments, this honeymoon period would sometimes last for years and often involve sky-high ratings that made Opposition Leaders miserable.  It was quite an eye-opener (on this site and elsewhere) when the Abbott Government elected in 2013 not only got no lead to speak of but was behind within a few months of election.  But even for re-elected governments, it's usual for them to get a bit of a pat on the back from the voters for winning.  Now, for the third time in a row, it hasn't happened.  This is, indeed, the first time ever that three governments in a row have failed to get the bounce.

In all, the previous 26 governments (new or returned) can be divided as follows as concerns their 2PP polling in the few months after the election:

* Sixteen got a honeymoon effect and recorded polling above their election result, without then quickly falling behind.  Only two of these (1969, 2004) went on to lose the next election.

* The Hawke and Keating governments following the 1990 and 1993 elections both obtained shortlived honeymoons but fell behind within 3-4 months.

* The Menzies government in 1958 and the Fraser governments of 1977 and 1980 did not get a voting-intention honeymoon, but the first two still kept the polling lead while the latter hovered around 50-50 on average for the first four months.

* The Abbott government in 2013 took a polling lead but at levels generally below its election result, and soon fell behind.

* That leaves just four previous governments that have never taken the lead in the months after the election or obtained a honeymoon effect, and have fallen behind soon after the election.  The Menzies governments of 1951 and 1961, the second Whitlam government of 1974, and the Gillard government in 2010 all fit in this category.  However in deference to the first listed, polling was very sparse in those days, so there really isn't enough evidence to clearly include it.

So historically, Turnbull's government lines up alongside two of the more infamous and turbulent second-term duds.  (The Whitlam government in 1974 had by far the worst polling for a government three months after an election win - it was already behind about 57:43.) However it also keeps company with the 1961-3 Menzies government, which had also scraped in with a majority of one (but no crossbench in those days) and which polled pretty poorly for its first fourteen months before it got an enormous helping hand from 36 faceless men. (In fact, 35 men and one woman.)

Once again, we cannot say predictively that a government is going to lose because it is polling indifferently already.  There is a widespread narrative sense that this is so - that the Turnbull government is already running dead, its leadership strategy in tatters and the Senate certain to be extremely obstructive. But the historic evidence does not confirm it.  The historic evidence says only that governments that poll badly early are at elevated risk of losing.

As Adrian Beaumont notes, some government decisions have been well received in sympathetic circles, but can be spun in many ways.

Leaderships: Turnbull Less Popular After Winning!

The news from Newspoll for Prime Minister Turnbull is bad.  I mentioned in the Abbott piece that winning Prime Ministers always receive a netsat boost - well, not any more.  Following the election, Turnbull has recorded his three worst net personal ratings ever: -18, -19 and now -23.  In the process he has exceeded Paul Keating's career peak-to-trough netsat loss (55 points), so his loss of 61 points in ten months is now the fastest Newspoll-era loss of that much popularity.  It is not the fastest ever; Billy McMahon lost 68 points in old Morgan Gallups in about the same time period (+47 to -21).

Here are the immediate pre- and post- election netsats for all the election winners in Newspoll history:

It is true that in the past, post-election ratings were often taken more quickly than they have been this time around, but even so, none of the previous PMs dropped below their pre-election netsat at all in the first two months.  If we go back to pre-Newspoll ratings, the one PM who might have gone backwards as Turnbull did is Gorton 1969, though I don't have Gorton's final pre-poll rating.  (Like Turnbull's, Gorton's result was viewed as a big disappointment - perhaps even more so as polling suggested he would win rather easily).  What has happened with Turnbull's ratings is even more unusual than what has happened to his government's.

Essential has a similar story with post-election net ratings of -11, -5 and -8 for the PM, compared to a pre-election net rating of zero and a pre-election low of -2.

Essential's leader attribute ratings this week show only seemingly minor changes of up to five points in various attributes since May; however, most of the changes are negative.  Also, it's likely that (for instance) if the proportion finding Turnbull arrogant increases by five points, then the proportion finding who would find him not arrogant decreases by the same amount.  So these changes are fairly consistent with the changes in his net ratings, for which a five-point swing from one side to the other makes a difference in the net result of ten.  Elsewhere, Essential finds Turnbull mostly ahead of Shorten on both positive and negative attributes, but this probably only tells us that voters are more likely to have an opinion of the PM than the Opposition Leader at the best of times, and that Shorten still elicits quite a lot of "meh!" responses.

Turnbull's ratings are probably being squeezed from two sides at once.  Moderate voters who believed Turnbull would be something genuinely different from Abbott are increasingly deciding that his leadership is just a makeover, and that while Turnbull might not embarrass himself as frequently as Abbott, his party is the same and he cannot control it.  Right-wing voters who were happy with Turnbull if he would get the party re-elected comfortably are displeased their side only retained its majority by a single seat and with a rather difficult-looking Senate.

As for Bill Shorten, the patterns for continuing Opposition Leaders after a defeat are more variable than for PMs. Opposition Leaders can face heavy polling backlashes if they lose unexpectedly or more heavily than expected (Hewson 1993 lost 19 points and Latham 2004 lost 20 in their first polls after losing).  However they can also be rewarded if they are more competitive than expected: Kim Beazley gained nine net ratings points in 1998 and Tony Abbott gained 18 in 2010.  Shorten's Newspoll ratings show basically no change as a result of the election, with pre-election Newspoll net ratings of -16, -15, -16 being followed by post election scores of -14, -17, -15.  It's a similar story at Essential where he's followed up -6 and -2 pre-election with -2, -4, -5.

Newspoll's "better Prime Minister" question shows Turnbull falling from a pre-election lead of 17 points (48-31) to post-election leads of 11 (43-32), 12 (43-31) and now 11 again (44-33).  The pattern so far is that although Turnbull's lead has fallen, most of the losses are going to uncommitted.  Before the election, Turnbull's smallest lead over Shorten had been 15 points.   Historically in Newspoll, a 16-point lead is about the break-even point for a sitting Prime Minister.  Essential, however, after initially having Turnbull with leads of only eight and 10 points has now blown out to a 15-point lead (41-26) which seems surprisingly large.

Essential shows Shorten leading as most trusted on nearly all of a list of issues, but I would not take much notice of this.  The issues listed are generally issues which a Labor leader should do well on.

Hanson Etc

Despite polling only 1.3% in the House of Reps at the election, Pauline Hanson's One Nation is booming, polling 5-6% in recent Essential polls, and doubtless contributing to a surge in Others at Newspoll (which had Others at 15% this week).  In fact, One Nation only contested 16 House of Reps seats, and probably succeeded beyond their wildest dreams in those, averaging 12.2% and getting their deposit back and public funding in the lot.  Considering the 1998 results in the same seats (which were slightly stronger, on the whole), and also the party's 4% in the Senate, it's quite likely that One Nation would have polled about their current 5-6% in the House of Reps had they contested every seat.  It's also realistic to assume they would contest a lot more seats right now in the unlikely case of a fresh election.

Last week's Essential received much attention for its finding of 49-40 support for a ban on Muslim immigration to Australia.  There was also some other Hanson-related polling that showed that, basically, Hanson may as well be a mainstream politician.  Indeed, many major figures would not get near 42% agreement for "I agree with a lot of what <insert name here> says and it's good to see them in the parliament."

Concerning the Muslim immigration ban result, I've noticed Essential respondents being generally grumpy about migration before, and I'd really like to see this result retested by other pollsters, especially those of the live-polling variety where the respondent isn't hiding behind a computer screen.  Even so the result is a striking one, and the 60-31 support among Coalition supporters doesn't say a lot for the moderation of the Coalition's base under a new PM.  What's also interesting is that the reasons cited for supporting the ban are mostly reworked late-90s Hansonism, applied to Muslims instead of Asians, just as pre-Hanson strands were applied to other groups: "They do not integrate into Australian society" scored 41% and "They do not share our values" clocked in at 22%. The modern bugbear of "terrorist threat" managed just 27%.

Same-Sex Marriage Polling

The other issue attracting a lot of coverage at the moment is same-sex marriage. Support for same-sex marriage remains consistently strong (62-32 in Newspoll and 60-30 in Essential; results being even stronger in live polling) but support for the plebiscite itself has changed a lot as the plebiscite debate proceeds.

Support for the government's proposed plebiscite is falling apart as the idea is debated more.  Possible factors here include:

* own goals by the anti-SSM camp through inflammatory statements that play into the hands of those arguing that a plebiscite would cause hate.
* disagreement with the government's proposal to provide public funding to both sides (whether because it would fuel hatred, because it would be a waste of money or because some views should just never be funded).  Public funding is opposed 22:68 with supporters of all parties heavily against.
* issue fatigue.  The timetable for the plebiscite would currently involve a vote in February, and some voters are probably just sick of the issue dominating politics and want it all to go away and finally be over.

Newspoll has now found 48% support for a parliamentary vote with 39% supporting a plebiscite in February and 13% undecided (the latter probably including whatever rump opposes any movement on the issue of any kind).  This goes even beyond the findings of a commissioned monster ReachTEL just a month ago, and shows a rapid shift since the plebiscite had 69:24 support just before the election.

Same-sex marriage itself is perennially a low-salience issue, that a lot of voters support but few will change their vote over.  To the extent that the election shows any government mandate for its position, it is only that by promising a plebiscite it prevented a loss of votes from voters who expected it to provide a pathway to reform.  Had the Government gone to the election with the same policy on the issue as Labor, it would have lost a share of the primary vote to Christian right parties, but regained nearly all of it in preferences and taken a few primary votes from the left as well.  Same-sex marriage was an election issue in the last weeks of the campaign, but far from boasting of its position as a key claim for a mandate, the government seemed most to want to make the issue go away.

In this term of government I will not be putting quite the same effort level as in the last term into chasing down almost every morsel of issues polling released every fortnight, but feel free to mention issues poll findings of interest in comments to the most recent roundup piece at any time.  (Note that, as usual, commissioned issues polls that do not display pristine polling design are treated with a lot of caution here.)

A note on scheduling for these pieces.  During the previous parliament, roundups were normally issued every week in which a Newspoll or other major poll had been released.  I don't expect to get back to that schedule right away for work reasons, so it may well be another month or so before the next instalment.