Tuesday, April 28, 2020

Newspoll: Record Premier Ratings And A Very Strange Federal Poll

This week Newspoll polled state Premier approval ratings, but not voting intentions (perhaps because samples by state would have been too small for voting intention sampling).  It was to be expected that several state Premiers would have very high approval ratings given their handling of the coronavirus crisis, but perhaps not that the figures would be quite so spectacular:

Prev = previous poll.  *= As opposition leader.  #=YouGov poll not branded as Newspoll.
As high as Scott Morrison's current net rating of +40 is (more on that later), all the Premiers except Palaszczuk have beaten it.  None of them were coming off a particularly high base, though the most recent polling for Victoria and WA is ancient.  For Tasmania this is the first Newspoll of Premier satisfaction since the 2014 state election.

To place these results in historic context, the highest approval for a Premier of any state in the 35 years of Newspoll was 75%, recorded by Steve Bracks in 2000 and Wayne Goss in 1992.  Others to make it above 70 were Brian Burke, John Bannon, Mike Rann and Peter Beattie.  One Prime Minister (Kevin Rudd) also did so, peaking at 71%. The highest net rating was Bracks' +62.  Mark McGowan (WA) and Peter Gutwein (Tas) have therefore recorded the highest two Premier satisfaction ratings and net satisfaction ratings in Newspoll history, while McGowan has also recorded the lowest dissatisfaction.  The approximately 31 poll respondents who were dissatisfied with McGowan's performance would be a very select group indeed.

Considering polls before Newspoll, John Curtin as Prime Minister polled approval ratings of 81% and 82% in two old Gallup polls taken during World War 2.  Bob Hawke reached 75% in Nielsen before Newspoll started.  I can find references with incomplete documentation to three state Premiers - Neville Wran, Don Dunstan and Rupert Hamer - who appear to have polled 80% or more.  Wran's first approval rating was 80% and accounts vary as to whether or not he may have later exceeded this.  Dunstan is variously said to have reached 82% or 83% (the 82% coming six years into his reign).

Any confirmed sources for old approval ratings above 80% would be welcome but I am provisionally regarding McGowan's 89% satisfaction and 83% net satisfaction as the highest approval ratings for any Australian Premier or PM in a scientific opinion poll ever by some margin.

(I did find a claim by Bruce Hawker that Mike Rann had had an approval rating of 94% in an Advertiser poll in 2002 but I am assuming that that, if correct, was an opt-in. Other sources refer to an unstated non-Newspoll poll with Rann over 80%, and there are similar claims for non-Newspoll polls in the Newspoll era for other leaders including Burke.)

McGowan's current popularity would rest in part on his resistance to cruise ship docking, also on his relatable "kebab moment" where he collapsed into laughter while answering questions about a NSW case of a man being fined for eating a kebab, and also on the ease with which the COVID-19 situation reinforces WA's distinct state identity.  The latter is also the case with Tasmania, but Tasmania has had an outbreak of the disease within its health system, while WA has been very successful in containing the disease so far.

Peter Gutwein has won support for moving early on quarantining new arrivals to the state and for a firm response to the health system outbreak once it started.  He also has the benefit that he is very new to the job, having only taken over three months ago.  Approval ratings have a "friends come and go but enemies accumulate" aspect to them so it is somewhat easier for a new Premier to impress voters of the opposite political persuasion.  However, Daniel Andrews as a long-serving Premier has also scored a very high rating.  Andrews' dominance of the Victorian scene is such that he may well have been rating highly before the crisis, but there were no polls so no-one knows.  There has been some resistance to Victoria's harsher lockdown measures, which in some cases made the government look slightly silly (eg the "bonk ban" episode) but the desperate attempts of some Victorian Liberals to weaponise this appear to have failed.

All of McGowan, Gutwein and Andrews are straight-talking no-nonsense types.  I haven't followed SA's response as closely as some other states but SA seems to have been quite successful in slowing the outbreak despite adopting less severe restrictions than other states.  The interesting thing is that although net approval for Palaszczuk's response is ten points behind Berejiklian's (see full charts here), Palaszczuk's own net rating is only +16, thirty points behind Berejiklian.  That's even though the NSW response has been clouded by the Ruby Princess situation.  Possibly Palaszczuk is being dragged down by her previous unpopularity.  It may also be that right-wing minor-party supporters (One Nation, KAP etc), of whom Queensland has a lot, are not buying into the current bipartisan mood, or that pre-election politicking has made LNP supporters less willing to approve of the current Premier who they wish to get rid of.

I suspect the NT and ACT Chief Ministers would also have very high approval at the moment (Michael Gunner in the NT especially) but there has been no polling.

Federal: Morrison Plus 40 Government Plus Nothing?

This week's federal Newspoll was quite an eye-opener because of its strange combination of results.  Scott Morrison gained another 14 points on net satisfaction to +40 (68-28).  Only Kevin Rudd (early and late 2008 and now and then in 2009) and John Howard (twice in 1996 when he was newly elected) have exceeded this figure in the Newspoll era, peaking at +57 and +53 respectively.  Morrison's approval of 68% is exceeded only by Rudd's peak of 71%, beating Howard's peak of 67%, but this is not a like-for-like comparison because Newspoll now drives down the "don't know" response by prodding initial "don't know" respondents so that they have to say "don't know" twice to be counted as ambivalent.

Prime Minister ratings have a close relationship to voting intention so one might expect that a popular PM would have a large lead.  Normally in the past a PM who is rating net +40 would be leading a party with a 2PP lead in the range of 54-60 (in Rudd and Howard's cases, such ratings included honeymoon-period wins after taking government from opposition).  Therefore it is quite a shock to see a 2PP of only 50-50 (and given the primaries, the government was slightly lucky to even get that; it was probably rounded up to 50 rather than down.)  This result is a major outlier in the historic relationship between these scores:

So although voter focus on the coronavirus issue is causing them to give Scott Morrison's performance an emphatic tick, it's not causing any real change in voting intention if Newspoll is correct.  This has important ramifications especially if there is a by-election in Eden-Monaro - by-elections normally see swings against incumbent governments, but the more popular the government is, the more the swing tends to reduce.  Given Eden-Monaro's very marginal status and (albeit lately tarnished) bellwether history that would raise the prospects of only the second ever case of an Opposition loss to a Government at a federal by-election.  However at this stage there isn't any sign of the government being able to take a COVID-19 bonus of any size into the by-election if one is held soon.  It would be nice to see confirmation that there is no such bonus from other pollsters, but no-one else is reporting voting intention at the moment.

The Better PM indicator (Morrison leading 56-28) continues to excite commentary from people who have not noticed it is skewed and a poor predictor (see my recent technical post about Better PM and its problems).  In fact, Morrison's Better PM lead is low given his personal popularity, but high given his party's voting intention, so overall it just reflects the other indicators plus the usual skew to the incumbent.  Meanwhile Anthony Albanese has polled a personal best net satisfaction of +11 (45-34) but there is so little focus on him at present it is hard to know what to make of that figure.  As noted last time the idea that the current polling is bad for Albanese is simply baseless - in a situation that obviously benefits an incumbent, the indicators for him are a lot less bad than they could have been.

Finally, a minor correction to something I may or may not have said on here before, but have certainly said on Twitter.  I have mentioned that the current Newspoll only shows the major parties and the Greens with all others lumped together as a voting intention response on the first screen (for those picking the others option there is then a more detailed range of choices on a second screen).  However I am now advised that One Nation is also included on the first screen in those electorates it contests.  Hopefully the upcoming formation of the Australian Polling Council will see such methods details being openly published by all pollsters.

Addendum:  Morgan OMGWTFBBQ

A Morgan poll has been released but it displays more of the company's typically abysmal public data handling practices that cause me to be unable to take Morgan results seriously.

The poll is said to show the Coalition up 4.5 points on 2PP to 51.5 from a previously unreleased result.  The previous result is described in text as "mid March" but the table below it shows the most recent previous poll dates as "Feb 29/Mar 1 & 7/8, 2020".  The chart however shows the most recent prior reading as earlier in February.  The poll is said to be a multi-mode "telephone and online" poll but it is unclear whether "telephone" means SMS.

The chart shows a total of 20 (!!) 2PP data points after the 2019 election, alleging that the government got a brief honeymoon before trading 2PP leads with the Opposition through late 2019, until a supposed crash caused by the 2019-2020 bushfires, which was supposed to have bottomed out with Labor leading 55-45.  However the x-axis only includes nine of these data points.  Moreover, none of these data points are currently shown on Morgan's archive of 2PP results, and the graph omits Morgan's final 2019 election poll which "showed" Labor with a 52-48 lead.

This continues a recent unfortunate trend of Morgan not releasing polls publicly at the time they were taken, and the data concerning past results is inadequate.


  1. Kevin,
    If we take newspol seriously and there are good reason not to could it be the punters are saying yes Morrison is doing a good job at present but the next election will be fought on the aftermath of Coronavirus and we aint sure of how to vote on that as yet.

  2. Anecdotally, even I as a confirmed Morrison hater, am forced to admit that after a shaky start, he's done more or less that right thing and if asked in a poll if he'd handled the crisis well, I wouldn't mark him down, and would probably even throw him a bone with a more mild "disatisfied" rating where I would normally go for the "very disatisfied" option, were there a choice.

    But you can bet your arse I won't be voting for those child torturing, climate denying, big business pandering, sports rorts/au pair/list-goes-on corruptering arseholes at any future election.