Monday, August 27, 2018

Rolling Poll Roundup: Early Morrison Polls

2PP Aggregate: 56.1 to Labor (+3.3 since last week) by last-election preferences
55.5 with One Nation adjusted preferences
Labor would win an election "held now" in a landslide with a seat count around the high nineties.

This post will follow polls released with Scott Morrison as PM for at least two weeks.  I would have called this post "Early Morrison Era Polls" but I'm not sure he'll be there long enough (this time anyway) to qualify as an "era". My commentary during the spill can be found in the amazingly titled Spill! thread, and my post-spill comments can be found in Australia - So What The Bloody Hell Was That?

Newspoll 24-6 Aug

Newspoll is a shocker for the reassembling government.  It loses five 2PP points and we get Labor credited with a lead of 56-44 off primaries of Coalition 33 Labor 41 Green 10 PHON 7 Others 9.  All previous mid-term PM changes have resulted in a Newspoll 2PP gain to the incumbent party:

Hawke to Keating +2 (est)
Rudd to Gillard +1
Gillard to Rudd +6
Abbott to Turnbull +5
Turnbull to Morrison -5

This is the worst Coalition Newspoll 2PP since the 57-43 to Labor in February 2015, taken at the time when Tony Abbott knighted Prince Philip, helped Campbell Newman lose the Queensland election and faced a damaging "empty chair" challenge.  If last-election preferences are used the current poll comes out fully as bad as that one (I've aggregated it at 57.1 to ALP).

This is also the equal second-worst Coalition primary ever, and their worst ever primary in government.  The Coalition had a 31% primary in Feb-Mar 2008 and a 33% in June 2008, both while in opposition to Kevin Rudd.

In terms of recovery from such bad polling, the Howard government won the 2001 election six months after recording a poll that I calculate as 44-56 (Newspoll didn't publish 2PPs away from election times at that stage).  The Keating government in 1993 and the Howard government in 1998 were each returned four months after recording polls that I have as 45-55.  So re-election from this position wouldn't be unprecedented, and the current poll is very much capturing the immediate blowback from the spill rather than any kind of settled view, but still, it is better not to poll such results.

Bill Shorten also leads Scott Morrison as better PM, 39-33, but this is rather meaningless as Morrison is new, and with time he may capture the lead.  However that ends Shorten's record streak of 58 consecutive Better Prime Minister losses.  It seems personal ratings are on holidays because of the leadership change.  Morrison is preferred to Shorten 44-34 as most capable to handle the economy, a question that tends to favour the Coalition.

The Newspoll, taken from Friday to Sunday, also finds that by that stage Julie Bishop (29%) was still preferred Liberal leader, although support had started to transfer from Turnbull (now 14%) to Morrison (25%).  Over time I expect Coalition supporters to lock in behind the new leader.

The Newspoll was much worse than Not-A-Poll voters in the sidebar thought it would be!  The average response was 53.1% to Labor, with only 8/11 having 56 or more, including two who had exactly 56.

Fairfax Seat ReachTEL

Fairfax have done seat ReachTELs of an interesting choice of seats: Deakin (Vic, held by now-demoted Dutton backer Michael Sukkar), Reid (NSW, held by Turnbull backer Craig Laundy) and Dutton's own Dickson (Qld).  Seat polling is highly unreliable so I would not read too much into figures that show a 9.3% swing against Sukkar as adjusted for redistribution (incorrectly rendered on the graphic as a 53-47 win), 2.7% against Laundy and 2.0% to Dutton, for an average of -3.3%.  Also, it is not yet clear whether ReachTEL's endlessly problematic "undecided" vote was redistributed in the quoted Coalition primary votes.

The issues questions are more useful because the margins are too large to be affected by the issues with seat poll voting intentions.  The poll finds voters in all seats giving Morrison a thumbs-up for his job as Treasurer (net results approaching +20), think Tony Abbott should remain on the backbench, think Australia should not withdraw from the Paris agreement, and think Turnbull shouldn't have been rolled.  There was also a question on calling an election right now, which had from 59-64% of voters (varying by seat) saying the government should go full term.  However, much as I dislike reflexive calls for immediate elections, I also have an issue with the wording of this one:

"Should Prime Minister Scott Morrison allow the government to run full term and hold an election by May next year, or call an election right now?"

The use of the term "allow" presents running full term as a positive.  A better design would be simply to ask when the election should be called - right now or next year by May?


Essential is 55-45 to Labor. Full results on site.

Notably the Essential contradicts Newspoll by having Morrison as Better PM 39-29 (little different from Turnbull vs Shorten).  It finds 35% approving and 40% disapproving, with the result that Coalition voters (those who are left) have already begun locking in behind the change (56-28).  In a typical example of Essential respondents contradicting themselves, respondents both agreed that Morrison should call an early election (52-30) and that he should be given a chance to show he can do a better job (56-29).  The best Liberal leader poll is of limited use because of the high Someone Else and Don't Know votes, but reminds us of how obscure Morrison was as a possible leader until now.


Morgan have released the following via their newsletter:

The latest Morgan Poll shows Federal support for the L-NP down 3.5% to 46% following a week of leadership upheaval for the Federal Liberal Party boosting ALP support by 3.5% to 54% on a two-party preferred basis according to interviewing conducted on the weekend of August 25/26, 2018 with 821 electors.

The two Morgan Polls taken over the past week show the former Malcolm Turnbull-led L-NP Government was within striking distance of the ALP on the weekend before Turnbull was ousted as Prime Minister and Liberal Party leader. The L-NP was on 49.5% cf. 50.5% ALP on a two-party preferred basis on the weekend of August 18/19, 2018 according to interviewing conducted with 886 electors.

This is a face-to-face poll, which also briefly resurfaced at the time of Turnbull's 30th  Newspoll loss. Primaries showing the Greens too high and One Nation too low are on Morgan's website. It turns out that a lot of the 2PP change between the two readings is a result of preferencing volatility (presumably respondent-allocated) as the primary vote change is quite low.

Aggregate Restarted 28 August

The Newspoll, Essential and Morgan polls are enough for me to restart my aggregate already, though the Morgan has a very low weighting.  I have also revised the weightings of different polls slightly for reasons stated here. The aggregate restarts at 56.1 to Labor by 2016 election preferences, 55.5 with One Nation adjustment.

Wentworth ReachTEL

The Daily Telegraph is publishing a ReachTEL of unknown source which gives a 50-50 2PP off primaries of Liberal 39 Labor 29 Phelps (IND) 11 Greenwich (IND) 11 Greens 9.  I'm not sure if "undecided" were redistributed or not (looks like they may have been for once , but depends on rounding).  One problem is that Alex Greenwich has declared he's a non-starter, while the Liberal candidate is not yet known.  [UPDATE: It turns out the source was the Australia Institute, the reporting was garbled and the actual Liberal primary for the question including Phelps and Greenwich was even worse - see William Bowe.]

I'll have more to say about Wentworth closer to the expected by-election date (Oct 6 is expected). Seat polling is unreliable generally and this electorate has a history of very bad seat polls.  However, given the loss of Malcolm Turnbull's huge personal vote and the current state of polling nationally it's not implausible to me that a pollster would get this result this early.  Whether it lasts until polling day in an electorate the left has never won in 118 years is another question.

Greens Canberra Senate Poll

A rather weird poll-shaped offering has been a Greens-commissioned ReachTEL of Senate voting intention in the seat of Canberra.  This has resulted in the faithful speculating that they're on course to rip a Senate seat from Zed Seselja, but not so fast just yet.  The poll shows the Greens 1 point ahead of the Liberals, but the key point is that it covers only the seat of Canberra (which is the greenest bit) and not the ACT as a whole - extrapolating to the ACT as a whole turns it into about a five point deficit.  This would close up on preferences (because ACT has low exhaust), but it still seems to require Labor to poll a very high vote (say, 47%) for the Greens to win.  The poll also has a range of issue questions of the usual useless more likely/less likely format (as does the one above), and its full results have not yet been published to my knowledge.  Also, it's a seat poll.  Also, it's a seat poll commissioned by the Greens.  


  1. I thought Newspoll would have waited another week. If they polled Friday, some respondents would not even have known that the leadership had been resolved. It would be expected that polls for the government would be much, much worse during / very close to the "in-limbo" period of the actual challenge. In making the comparisons between the polling effects, it could therefore be important to know the time that elapsed from the leadership change and the polling in the earlier leadership changes.

    I find it interesting that those selecting Turnbull as their choice of Liberal leader also fell away very sharply once he was removed in a couple of the Polls. 14% nominating him as preferred leader within a couple days of removal would appear to belie his supposed appeal to non-government voters, who I would have thought would have rallied to his cause given the nature of his removal. There was also suprisingly high support for the leadership change.. 35% in Essential and similar in the Reachtel seat polls... especially given the very messy nature of his removal. Opposition of only 40% in Essential and mid-50's in Reachtel is hardly overwhelming and adds credence to the need for caution with these very early (too early!) polls.

  2. Kevin, I am doing regular poll aggregations in the first days of the Morrison Government. The latest can be seen here:

  3. key interest of Wentworth is what happens to the 12% add on to Turnbulls vote from 2010 onwards...… this seems to have stayed...… a 60% vote in the marginal areas seems much inflated in a good time this could vote 55% alp....this seat is a liberal seat but not with 18% margin