Tuesday, September 22, 2020

Newspoll: The Sound Of Narratives Dying

 This week's Newspoll results have included fresh approval ratings for Labor Premiers Daniel Andrews (Vic) and Annastacia Palaszczuk (Qld), along with a number of issues questions regarding handling of COVID-19.  These are well worthy of comment - Victoria especially - along with some brief comments on less striking results (but strikingly if typically bad reporting) in federal Newspoll land.


A widespread narrative in Victoria has been that Premier Daniel Andrews is severely on the nose and his Premiership is in crisis, either because of the second wave of COVID-19 in the state that resulted from avoidable quarantine failures on his watch, or from the severity of lockdowns deployed in response to (so far successfully) bring new case numbers down.  

Victoria has recently seen public voting intention polling, finally, via a 51.5-48.5 lead in Roy Morgan Research's SMS polling.  I am not a fan of SMS polls as a method (I think they are too prone to motivated response), and Morgan's recent state polls have a history of volatility and being rather inaccurate, but at least this is something.  Also, ALP-linked campaigning firm Red Bridge has issued a 53.5-46.5 result for Labor.  These results followed Liberal Party MediaReach polling that had the Liberals picking up massive swings in a group of Labor-held seats (which if applied statewide would see Labor losing), but this is the same firm that, for instance, had the Territory Alliance on course to be the largest party in the NT parliament (it won a single seat very narrowly).  The internal poll also showed a rather heavy fall in the Greens vote, which seemed unlikely in the absence of anything that would cause it and given the general resilience of the Greens vote in recent elections (OK, except for Eden-Monaro.)  

Critics have continued to suggest that Andrews has cooked his goose and is now doomed, perhaps even at risk of a leadership challenge before the term is out.  But Newspoll says otherwise, for now at least: Andrews' net rating is up seven points from +20 to +27 (62-35).  Perhaps it is helping his cause that case numbers in the second wave are rapidly dropping (and have dropped further since the new poll was taken), whereas at the time of the previous poll they were growing out of control.  

It should also be noted that this result can't be explained away by modest sample size.  The theoretical margin of error is around 4%, so maybe Andrews' approval is really 58 or 66 not 62, but it doesn't really matter.  

Nationally, 61% think Victoria's restrictions are just right, 25% think too strict and 10% say too lenient.  But in Victoria, the lead for just right falls to 54-37.  

Much gnashing and wailing has been seen on the pages of the Australian already in response to this inconvenient result and much more will be seen elsewhere, but there's a lot of overthinking going on.  At the moment, most people really want to feel safe, and movements and attacks based on any other presumption have been falling flat.  

It is worth noting that while Newspoll finds satisfaction with Daniel Andrews to be on a par with some of the other Premiers, Essential finds satisfaction with his government's performance in managing COVID-19 to be lagging, at 47% compared with values from 67-84% for other states.  


The Queensland results are also quite significant because Annastacia Palaszczuk's government is just six weeks out from an election that at this point is hard to predict.  There has been no useful statewide voting intention polling data for Queensland since way back at the end of July, so it's hard to even estimate what's going on there, but so far the heat of battle has done little to deflate the Premier's COVID-19 approval bounce.  Annastacia Palaszczuk is down insignificantly from net +35 in late July to net +30 (63-33) now.  

There is actually no precedent for a Premier with such a high net satisfaction rating in a sample taken entirely this close to an election losing.  The nearest approaches are:

* Rob Kerin (SA) +28 on the eve of an election he very narrowly lost (early Feb 2002) and +39 in Nov-Dec sample of the year before.  Kerin had only been Premier for a few months.

* Wayne Goss (Qld) +45 in May-June and +28 two weeks out from the mid-July 1995 election that Labor initially won by one seat, but ended up handing over government mid-term after one of their wins was annulled and they lost the subsequent by-election.

* Jeff Kennett (Vic) +22 in July-August and +12 on election eve for the mid-September 1999 election that he narrowly lost.

We shouldn't place too much stock in Palaszczuk's current high netsat as a predictor because the relationship between COVID-19-fuelled approval ratings and actual voting intention seems to be very patchy - super-popular Premiers like Gutwein and probably McGowan are getting large lifts but the Morrison federal government has gained relatively little, and Queensland Labor was trailing in voting intention even while the Premier polled strong personal ratings in June-July.  Nonetheless I don't recommend ignoring it either.  

Newspoll finds 53% of voters nationwide think Queensland's restrictions are about right, 37% think they are too strict and 7% think they are too lenient.  However, in Queensland, "about right" holds a 58-32 lead.  It's NSW, Victoria and the ACT where the margin is closer - and they won't be voting on October 31. (Essential has made a very similar finding this week.)

South Australia

It is also worth briefly mentioning the YouGov poll of SA released last week.  Despite a rough run with expenses scandals that have cost it a few Ministers and triggered chaotic speakership election chaos in both houses, Steven Marshall's Liberal government has polled a 53-47 lead.  This may sound underwhelming since it is only a 1% swing before the election, but it is a six-point turnaround from the same poll in early March.  Premier Marshall enjoys a very strong +52 net personal rating (68-16) and a 54-26 lead on the Better Premier metric (which skews to incumbents).  The poll is far from terrible for Labor, with leader Peter Malinauskas continuing to poll better than other Opposition Leaders for whom data exists (he's at net +22 (44-22).  However the 46-35 primary vote lead to the government would be of some concern, as would the possibility that without the ministerial and parliamentary chaos, the government would be even further ahead.


The federal Newspoll came in at 51-49 2PP, a one-point lift for the government.  However this one-point 2PP lift came from a poll with the government and Opposition up 2% and down 2% respectively on the primary vote.  I get the typical 2PP off the published primaries at 51.3%, so the government may well have been close to getting rounded up to 52-48.  

The poll mainly attracted commentary because the previous Newspoll had seen a tightening to 50-50.  This may have been explained by heat on the government over aged care failures in the COVID-19 response during the previous poll cycle, but it may also be viewed as a bit of an outlier.

As well as the large degree of attention heaped (as usual) on a statistically insignificant partial correction to the shift in the previous poll, the new poll generated an error-ridden article from Simon Benson, which I decline to link to:

* "[..] Labor leader Anthony Albanese’s approval ratings tumble into negative territory for the first time." - False.  Albanese had recorded six previous negative net satisfaction ratings, five of them last year and one in late February.

* "For the first time this year, the Labor leader’s net approval ratings went negative, with a four-point fall to 39 per cent in satisfaction levels compared with 40 per cent dissatisfied." - False.  See above; net -1 is not even Albanese's worst rating of the year.

* "The improvement maintains Mr Morrison’s long-running popularity since the pandemic but is still short of the record highs in support in June and July when he reached 68 per cent approval." - Misleading.  68% was not any kind of record (the record is 71% held by Kevin Rudd).  Morrison did set a record of sorts for keeping his satisfaction rating so high over an extended period (nobody else has been at +66 or above over a three and a half month period) but this is meaningless because the change in YouGov methods late last year has artificially increased Morrison's satisfaction rating.  Net satisfaction is the better measure for historical comparisons, and on that front Morrison still has several weeks to go to break Kevin Rudd's record for the longest streak above net +30 (just over six months).

* "a corresponding fall for Labor, which has returned to near-historical lows of 34 per cent." - Misleading.  34% is near to the historic low for the party in an election, but it is not near the historic low for the party in polling.  The Gillard government at one point sank to a primary vote of 26%.

I have had a long-running Not-A-Poll in the sidebar regarding what sort of net rating Scott Morrison would be polling come September.  The PM easily outperformed the expectation of the voters on this site.  The most popular pick was +20, with 28% of the votes, and the average pick was +16.2, but closest to the pin was +30, with 22.6%.  Morrison's actual netsat was +34 (65-31).  

Comparative Summary

I thought it would be worthwhile writing a comparative summary of where the different governments and leaders stand in terms of leader approval and whether they have got anything out of COVID-19 in voting intention terms:

Federal (Morrison, Lib): leader approval high, slight boost to voting intention

NSW (Berejiklian, Lib): leader approval high, no evidence on voting intention

Vic (Andrews, ALP): leader approval high, ahead on voting intention (weak evidence) but with swing against from previous election

Qld (Palaszczuk, ALP): leader approval high, no recent evidence on voting intention but was slightly behind in late July

WA (McGowan, ALP): leader approval stratospheric, sketchy evidence on voting intention suggests massive lead

SA (Marshall, Lib): leader approval very high, large boost on voting intention off low polling base

Tas (Gutwein, Lib): leader approval stratospheric, massive boost in voting intention

We have no reliable approval data for the NT or ACT Chief Ministers but NT Labor was comfortably returned despite seat losses, while there is not much to see on voting intention in a single recent commissioned ACT poll.

1 comment:

  1. You're right that ACT aren't voting October 31, they're voting October 17. Seems like even less interest than last time outside Canberra (and even within).