Wednesday, July 9, 2014

Queensland Polling: Careful With That Pendulum, Eugene

ReachTEL (Queensland): 51:49 to LNP (LNP 38.7 ALP 34.4 Greens 6.1 PUP 15.4 Other 5.4)
Interpretation based on this poll "if election held now": LNP would probably win; majority status touch and go.
Widespread claim that poll implies c. 40 seat losses and defeat for LNP is wrong.

=====================================================================

I'm currently conducting a poll on the sidebar on whether readers want more shorter federal polling articles instead of longer ones with updates tacked onto the end. The results will also be used to inform how I cover state polls for states other than Tasmania.  This piece (not that it's that short) is a bit of a trial run for the more-but-shorter option and will not be endlessly updated.  One of my main concerns about the shorter option has been how to keep the titles snappy, rather than having an endless procession of pieces called "Poll Roundup Week 37" or "Essential: 51:49 For The Sixth Week In A Row While The Universe Burns". Anyway, I hope that the title for this one passes muster, with apologies to early Pink Floyd and to all readers whose name is actually Eugene.

For an extreme example of how I've approached state polls up til now, see my previous Queensland article Is Campbell Newman Actually In Trouble?, which started in April and ran through to the end of June.  The comments on the recent quarterly Newspoll aggregate (51-49 to Labor) bear repeating: while the result was more generous to Labor than everything else about in recent months, it did put a dampener on weak signs that the LNP were recovering from the low-50s 2PPs they were getting in previous months.  The poll also showed Campbell Newman at a new personal low: a -24 Newspoll netsat, giving him the worst result for any Queensland Premier bar Bjelke-Petersen and Bligh.  I noted that if Newman can win the next election he will equal Bob Carr's record for the worst netsat from which any Australian state premier in the last few decades has been re-elected.  That said, given that the poll was worse for the LNP than others around the same time, it may be exaggerating Newman's plight.



The new ReachTEL (51:49 to LNP, for what a 2PP estimate can be worth with the PUP vote surging) is more evidence that the LNP are not in liftoff mode and indeed are shedding primary votes to PUP.  They remain with the same sort of unimpressive lead that would probably be enough for a narrow win, but that gives them serious problems with what to do with Campbell Newman in his own seat.  While the PUP bubble may well have burst in the rest of the nation, in Queensland it shows no signs of doing so and the current poll has the party on a record (for a public poll) 15.4% primary. 

But if you look at the media reporting, you might think this poll was a lot worse for the LNP than it was.   The reporting of this poll has been sensationalist and more or less universally poor.  Sources as disparate as the Courier Mail and the Guardian have claimed that the poll points to a loss of around 40 seats, while AAP is among those claiming that the recent polling shows Newman as likely to lose.

I'll take a big wild guess about what happened here and someone can tell me if I'm wrong.   My hunch is that what we have here is a big bad case of Pendulum Abuse and some lazy journo read down the Wikipedia pendulum and counted off 39 seats on 11.8% or less, rounded off to 40, and then some other lazy journos copied them. 

But firstly two of those 39 are Yeerongpilly and Redcliffe, which have already left the building (to be joined almost certainly by Stafford next weekend). So they can't be counted against the LNP's existing swag of 73.

Secondly of the remaining 37, only 31 were LNP vs Labor at the last election.  Two were LNP vs IND and four were LNP vs KAP.  Antony Green's site presents a pendulum in estimated 2PP terms that gives a fairer LNP-vs-ALP picture, and only Thuringowa is under the 2PP swing implied.  The other five are all conservative seats.  What will happen in those seats with the KAP vote collapsing is hard to say.  If it transfers mostly to PUP and PUP's vote increases by the levels implied then PUP could well get some of these seats.  But we know from the ReachTEL that only 15% of PUP voters self-report voting for KAP last time. While polls of self-reported voting intention are always a little bit troublesome (because voters don't always tell the truth) it does seem from that that only a minority of KAP's vote is currently heading PUP's way and that converting strong KAP seats to PUP seats does get messy.  If there was a statewide swing of 11.8%, I'd expect the LNP to lose some of these six non-classic seats, but not all.

Thirdly as done to death in the previous article, the likely impact of sophomore effect on the Queensland election can't be stressed enough, and for a 51-49 to LNP result, it would be likely to save about four seats.

Fourthly it's probable that the 2PP off those primaries is much more on the 51.5 side of 51 than the 50.5 side.

Adding these factors together I don't think a 51:49 2PP to the LNP would see it drop anywhere near 40 seats to Labor; indeed, my modelled estimate is 25 seat losses to Labor (assuming they have already gained Stafford). However I'd expect PUP and indies to gain more seats than KAP would lose, and to mainly do so from the LNP, making it touch and go whether the LNP could maintain its majority.  A messy crossbench of a dozen or so (about half PUP and the rest a mix of indies and KAP) would give the government - and its new Premier, ho ho -  plenty still to work with if it did fall a couple of seats short. 

The Courier Mail's claim that the poll would allow Labor to lose the 2PP and yet triumph in 40 LNP held seats, thus winning with 48 seats to the LNP's 33, is patently absurd.  And even if the poll really did imply 40 seat losses if an election was held right now, the argument that a single relatively close poll taken about nine months old shows a government is headed for defeat is even more so. It's odd that a polling result that only stands out because of the high (and clearly genuinely growing) PUP primary and is otherwise very close to others recently has attracted such a frenzy of incompetent reporting.

Finally ReachTEL approval figures tend to make for stark reading but on my trusty ReachTEL-to-Newspoll conversion this one has Newman on a net -16.3 and Palaszczuk on a net +2.8.  And the bizarre result that 2.4% don't know who Campbell Newman is seems to have no source other than random noise in the 18-34 age group.

2 comments:

  1. 15.4% is the record result for PUP for a public poll? Kevin, I do hope you are not having a sly did at Palmer's own internal polling that showed they were on track to win the Tassie election?

    ReplyDelete
  2. I wasn't, much as they deserved it! I wasn't sure offhand if PUP had polled higher in Queensland in any of the large volume of union-commissioned polling or other internals. (One recent union poll had them on 15.1 after redistributing "undecided".)

    ReplyDelete

The comment system is unreliable. If you cannot submit comments you can email me a comment (via email link in profile) - email must be entitled: Comment for publication, followed by the name of the article you wish to comment on. Comments are accepted in full or not at all. Comments will be published under the name the email is sent from unless an alias is clearly requested and stated. If you submit a comment which is not accepted within a few days you can also email me and I will check if it has been received.