Poll Roundup: Sea Of Red Polling Ink At Start Of Election Year

After being in a merely historically dicey polling position for much of 2021, the Morrison Government starts this year in more serious trouble.  

Last night's was one of those Newspolls that deserves a quick post because of the dramatic nature of the figures.  The Morrison Government starts the year down 44-56 on primary votes of Coalition 34 Labor 41 Green 11 One Nation 3 Others 11.  Scott Morrison is on a net rating of -19 (38-57), Anthony Albanese is on net zero (by 2022!) (43-43) and Morrison leads on the skewed "Better Prime Minister" indicator by 43-41.

This all represents:

* The Coalition's worst Newspoll 2PP since it polled a pair of 44s in August-September 2018 immediately after Malcolm Turnbull was removed as Prime Minister.

* The Coalition's worst primary vote in the same period, and equal second-worst of all time.

* Labor's best primary vote since December 2018.

* Morrison's worst net satisfaction rating since the early 2020 bushfires, marginally above the -22 he sank to during that time.

* Albanese's best net satisfaction since March 2021.  Albanese is now 19 points ahead on comparative netsat after only taking the lead one poll ago.  

* Morrison's lowest lead on the skewed "Better Prime Minister" question since the bushfire period (in which time Albanese briefly led).

* However, one of the very few cases of a Prime Minister leading on Better PM while this far behind on voting intention.  Not a record in that regard, eg Morrison led by 6 in the second of the 44-56 Newspolls after the removal of Turnbull.

Other Polls

This follows on from some other bad polls for the government.  Morgan also had a 56-44 to Labor (following on from the back-released 55.5 in late December), but more usefully it released numbers for Independents for the first time, which provided some possible insight into why Morgan's respondent 2PPs have been skewing to Labor.  The last two Morgan polls have had very high readings for Independent (7.5% and 8.5% respectively), suggesting that Morgan may be picking up a lot of "climate independent" type voters whose preferences flow about 70-30 to Labor.  I converted the most recent Morgan as 55.1 to Labor, but that is not comparable to my previous last-election conversions because I now have access to the Independent figure.  I'd expect that what we're seeing with Morgan is about a point of skew relative to Newspoll caused by its use of respondent preferences, and another point caused by its sampling being relatively strong for Greens and independents.  

The Resolve series, which has tended to be more friendly to the Coalition than others, also returned a poor reading for the government, with a 35-34 primary vote lead to Labor and a very high "independent" vote - overall I convert this poll to 54.2% to Labor by last-election preferences, but this is a very rubbery figure because of uncertainties about "independents" - does the voter really mean independent or are they picking that as a glorified "dunno", and if they do mean independent, of what sort.

On those independents, while there has been a lot of talk about a scenario where the Coalition loses votes, Labor makes only modest gains and independents pick up the slack with a growing crossbench the result, the sort of polling blowout we are seeing now suggests we could see something else.  If there is a reasonably decisive mood for throwing out the government then independents could become a sideshow and might struggle for campaign attention.

At this stage there is no fresh voting intention polling from Essential, but some will be released in February.

Newspoll also asked questions about which leader would better handle certain issues.  Albanese led Morrison 33-32 on "leading Australia's recovery out of the COVID-19 pandemic" and 39-21 on "leading Australia's response to the global climate change crisis".  Morrison led 33-31 on "creating jobs and growing the economy" (a question on which a Coalition PM would normally expect a large lead) and 31-26 on "dealing with the threat of China in the Indo-Pacific".

How Bad Is It?

The election is probably three and a half months away, with May 14 the most widely tipped date.  No government has ever polled a Newspoll 2PP this bad this close to an election and won.  The nearest approaches were a March 2001 Newspoll for the Howard government that equates to 43-57, and the above-mentioned 44-56 polls when Turnbull was replaced by Morrison.  These were, respectively, about six and eight months out.  45-55s have been recovered from from four and a half months out (1993) and five months out (2019).  

Such things having never happened doesn't make them impossible, especially when the bad polling has obvious contributing factors that might go away.  But unlike in 2018-9 when the bad polling was caused by internal ructions that the Prime Minister appeared to be blameless for (and at least at the end by polling error, though it's not clear at what point that kicked in), in this point he is seen as central to what is going on.  In particular, after the government brought Australia through the early days of the pandemic in a way that looked after both people and the economy, there is now some sense of abandonment amid the chaos of Omicron, reopening and related service delivery failures (including the shortage and costs of rapid antigen tests).  This also showed up in Essential last week, with approval of the Government's handling of COVID falling into net negatives for the first time.  

It's important to note that while federal governments have not polled Newspolls this bad close to an election and won, governments have recovered enough points from individual Newspolls that would put this government in a competitive position on polling day if it could do the same (see two examples mentioned by Peter Brent here.) Those polls were, however, from the bouncy days of the old Newspoll and were both pretty obvious outliers.  The poll these days has a larger sample size.

In late 2021 the government's polling position kept deteriorating, but Labor's lead remained modest.  In December Labor's aggregated 2PP lead may have reached 54-46, but this wasn't immediately obvious because of the delayed publication of the late December Morgan and the lack of other polling data.  In any case, these were the sorts of numbers governments often recover from.  

This is no longer the case.  The opposition is ticking boxes - they now have the sort of lead that doesn't guarantee a win but that winning oppositions will generally build at some stage of the cycle, and their leader is now clearly more popular than the PM.  Assuming the current version of Newspoll has no house effects I estimate the government's aggregated position at around 44.7% 2PP. That is historically very likely to increase before election day (so there's no point getting excited about Labor winning 95 seats since that probably won't happen), but the three governments that were about this badly placed around this time in the Newspoll era all lost.  The governments that survived while trailing were at around 47.  Indeed, polling at around this time in the electoral cycle has been reasonably predictive in recent decades with the exception of 2019 (when the final polls were wrong and it is not clear at what point they started being wrong).  At this stage, Labor was ahead in the 1993 cycle, the Coalition was gaining under Howard in 2001, Kevin Rudd's huge leads in the 2010 cycle had evaporated, Rudd had not yet been restored in 2013 and the Turnbull bounce had disappeared in 2016.

Turbulence still to come in past cycles included Labor falling well behind in the early 1993 cycle and the very large temporary bounces to the Coalition in 2001 following the September 11 attacks, and to Labor in 2013 after Rudd was restored as Prime Minister.  So while this quick graph I did for elections in the Newspoll era has the Government on track for a 2PP result of about 48% with a margin of error of about 3% (and something like a 65-70% chance of losing depending on how 2PP converts to seats) I suspect the real margin of error is greater.  

The current situation is also potentially turbulent, for a few reasons.  Omicron seems to have peaked (though there is always time for another variant to come along and vex this government) and perhaps as the election approaches there will be a strong switch back to focus on the economy. However, I am not convinced the Coalition holds all the good cards there either.  There's also the possibility of the issue mix going khaki, for instance with the Russia/Ukraine situation (The Australian also thought China was worth polling about).  It's important to keep in mind though that khaki elections are not always good for governments - they have to have the right issue responses at the right time.  Even early in world wars the Cook government was kicked out when it went hunting for a bigger majority in 1914 and the first Menzies government lost its majority in 1940.

Could the polls still be wrong by a similar amount to last time; might the current 44.3 be really a far more recoverable 47.3 or so?  It's possible.  Resolve is publicly untried.  Morgan has predicted Labor to win six of the last nine elections but Labor only won two of them, and shows no sign of having responded to the polling failure in any meaningful way.  (It has however ditched face-to-face polling, but only because of the pandemic).  Only Newspoll has taken obvious well publicised and constructive steps that seem usefully targeted at the cause of the 2019 failure, but that doesn't mean it will necessarily be right either.  Overall though, overseas experience is that global house effects usually don't repeat from election to election, and the 2019 failure may also have been about the kind of campaign as much as errors by pollsters. 

When governments win elections from seemingly hopeless positions, it is easy to overrate their chances of doing so again.  This was the case with Keating in 1996 and to a degree Howard in 2007.  Coalition governments are often good at getting out of polling trouble, but that doesn't make them bulletproof.