Wednesday, December 16, 2015

Poll Roundup: 2015 Year In Review

2PP Aggregate: 53.5 to Coalition (+0.1)
Coalition would win election held now with unchanged to slightly increased majority

It's just about the end of another year in federal polling; should any unexpected late polls appear I will edit this article to add them in.  After an update for this week's polling I'll launch into an annual review along similar lines to last year's.  From here on in the pollsters tend to go into summer recess with Morgan and Essential returning in mid-January and the heavy hitters coming back in late January and early February.

This week's polls

This week we have had readings from Morgan and Essential, which continue to sit at opposite ends of the Turnbull-era spectrum, this week returning 56-44 and 52-48 respectively.  The former was Morgan's highest reading for the Coalition this term, and the respondent-preferences reading was even higher (57.5%).  Essential has had the Labor primary at 35-36 in the last four weeks while Morgan has had it at 28.5 then 27.  Either both are wrong or one is very, very wrong.

Although both pollsters showed an uptick to the Coalition, this was tempered by the Ipsos from a few weeks ago falling out of sample, so the net result is just a 0.1 point gain, for the Coalition, after everything, to finish the year in exactly their 2013 election result position.  At least, that's my take; as usual recently, others may well be higher.  (Edit: Yep; Bludgertrack 54.1 Mark the Ballot 55 and Phantom Trend 55.2.  MtB assumes zero-sum and includes Morgan but not Essential, and Phantom Trend treats Morgan as having the same sorts of house effects it's had for decades, so those points explain why the latter two are so high.)



Here's the smoothed tracking graph:


Essential gave us some late-breaking personal ratings with a generic "performance" question, which showed Clive Palmer recording a similarly dire -52 net rating (cf last year's -50), and Julie Bishop again polling well (+22 compared to last year's +28) - note that the 2014 question was specific to that year.  Richard di Natale polled -12 (16-28) compared to Christine Milne's -23 last year (16-39).  The main difference there was a higher neutral rating rather than (as I might have expected) a higher don't know score.  Malcolm Turnbull had the same net rating (+33) on the performance approve/disapprove question as last week, but Bill Shorten copped a 10-point net hit for a net -30.

Strikingly, last week 27% of Essential respondents approved of Shorten's performance, but this week only 14% rated Shorten's performance as good.  Some of this will be down to different question designs, but it's hard to avoid suspecting that the incident in which Shorten was filmed using a phone while driving erratically has done some damage to the Opposition Leader here.

There are also Essential's usual end-of-year questions which show Essential respondents less the pessimists they usually are and more positive about 2015 and 2016 prospects for the economy, small business, the average Australian and Australian politics especially.  I'll just link to their download rather than discuss these in detail.

There is also a slightly dubious one-two in which Essential asks respondents whether they think government spending has increased then asks them if they are worried that it in fact has.  Had they also mentioned that revenue has also increased (albeit inadequately) then there may have been a different response.  It's interesting that 61% of Coalition supporters took the bait, but it would be more interesting if there was any party with the economic credibility to make some use of that.

Annual review

This year we've seen 123 federal 2PP polls released - 49 Essentials, 26 Morgan multi-modes, 12 ReachTELs, 11 new Galaxy-run Newspolls, 9 old Newspolls, 9 Ipsoses, four Galaxies, and three Morgan Phones.  The year saw one major polling-methods change with Newspoll becoming a Galaxy-administered brand that now polls an online panel/landline-robodial mix, in contrast to the old Newspoll, which used live landline interviewing.

The year splits of course in two sections.  In the Abbott section, the Coalition scored one 50-50 2PP tie (May Ipsos) and 88 2PP losses, the worst being 43s from Morgan, Newspoll and Galaxy twice (all in early February).  Two of the losses might have been considered draws after taking house effects into account.  In the Turnbull section, the Coalition scored 31 2PP wins and three 50-50 ties, with the biggest wins being a 56 apiece from Morgan and Ipsos (the former of which has had a strong pro-Coalition house effect under Turnbull to date).  Noting that something in the high 48s would probably be enough for the government to win, the government has in fact won every poll under Turnbull, and was competitive at times under Abbott.

The average reading for my aggregate for the year was 51.5:48.5 to Labor, but Labor averaged 53% against Abbott then only 47.4% against Turnbull.

Tony Abbott recorded 14 consecutive negative Newspoll netsats with a high of -13 and a low of -44, and was removed at -33.  Malcolm Turnbull then recorded six consecutive positives with a low of +18 and a high of +38.  Historical comparisons involving Turnbull's ratings and subsequent election results are not very useful because they are honeymoon-period ratings and we are yet to see what level they settle at.

Bill Shorten polled a +2 Newspoll netsat in February followed by nineteen consecutive negatives, ending the year on his worst ever, -38.  Opposition Leaders to have rated negatively in every poll while surviving a whole calendar year in the job were John Howard (1986, 1987, 1988), John Hewson (1993) and Tony Abbott (2011, 2012).  Among opposition leaders in the role for at least nine months of a calendar year, Shorten's -21.5 average sits above only Simon Crean in 2003 (-27.2), John Howard in 1988 (-26.4) and Abbott in 2012 (-25.8). Crean was removed with one Newspoll to go.

Abbott's extreme unpopularity, combined with Shorten's benign ratings early in the year, meant that Shorten still managed to be rated better PM seven times to Abbott's four with three ties.  On average Shorten led Abbott by three points (even with the house advantage to incumbent PMs) but Turnbull has led Shorten by an average of 42.7.

Betting

Again I note some current betting figures.    Implied winning chances for the Coalition were at not much over 55% towards the end of the Abbott regime and have since jumped to about 85%, with bookies currently offering $1.10-$1.15 on the Coalition and $5-$6.50 on Labor.  The former does not mean punters expected Abbott to win; many would have been pricing in their perceived chance of Abbott being removed.

Seat-betting at Sportsbet currently has the Coalition gaining McEwen (Vic), Fairfax (Qld from PUP) and the new seat of Burt, and Labor picking up Solomon (NT), Hindmarsh (SA) and Barton (NSW) while losing Hunter (redistributed) - a virtually zero-change outcome if the favourite wins every seat.  Fairfax is the only contested seat here the markets are all that confident about, and they are also not that confident about the Coalition seats of Cowan, Dobell, Paterson, Petrie, Capricornia and New England (vs Ind), and the ALP seat of Parramatta.  There would not be that much money in these markets; they would still mainly reflect the bookies' modelling.   Consistent with all this, a seat total market at William Hill has 86-90 Coalition seats (they have 90 now) favourite with 81-85 considered slightly more likely than 91-95.

Bill Shorten is at $1.40 to survive til the next election as Labor leader with Tanya Plibersek at $3.75 and Anthony Albanese at $5.  A name that wasn't even in the frame when I started my ALP leader poll (but is rated very highly by a few readers) has been added with Mark Dreyfus at $11.

The overall picture

This year featured four major polling events.  In February, the knighting of Prince Philip contributed to the LNP losing the Queensland state election, and these issues plus consistent poor polling and internal tensions with the PM's office caused a spill motion against Tony Abbott.  The blowout caused by these items was severe but surprisingly brief.

A benign Budget saw the Coalition in an almost election-saving position by mid-year, but the entitlements scandal in July that led to the downfall of the Speaker, Bronwyn Bishop, triggered a moderate but sustained drift away from the government.

A perception that the rot had now set in for good was among many factors leading to a successful leadership challenge and change of Prime Minister, leading to one of the biggest polling shifts in favour of an incumbent government in Australian history.

Where does it go from here?

We don't know when the next election is yet, but let's assume it's about nine months away.  Historically, the form guide for governments at least this far ahead in polling that far out is quite a simple one: they win.  But there have only been six of them, and two of them won ugly, with only around 50-50 2PPs.

At this stage, polling normally explains about a third of variation in the final election result, and it doesn't get much more predictive than that until the last few weeks of a campaign.  All else being equal, a simple model based on outcomes back to 1949 projects a Coalition 2PP at the election of 51.8 +/- 2.6%.  One that takes into account the Labor Fail Factor (Labor's tendency to perform worse on the day relative to lead-up polling) says 52.3 +/- 2.2%.  Those sound competitive, but Labor's serious problem is that the scars left from its last-election drubbing mean that the Coalition probably only needs high 48s to win.  (Once all the redistributions are done I'll be posting a target figure in the sidebar and updating it for new retirement announcements).  Once that is taken into account, Labor's chances become very slim indeed.

However there are still two reservations about this.  One is that we still don't know if the Coalition's current lead is real, or if it still represents an artificial honeymoon bounce that will go down.  The other is that leadership changes create instability, and while voters will probably ignore further strange and damaged outbursts from the Abbott camp, tensions between the new PM and the Nationals will need to be managed carefully.

Are these enough to shift the Coalition from firm favouritism to win (whether by a little or a lot), now that they have listened to the people by getting rid of Tony Abbott?  I don't think so.

Christmas Present!

Recently I had a great success with the use of the Wayback Machine to retrieve old Morgan Gallup and Morgan polling.  At one stage Morgan had a large volume of polling back to 1960 on its website and this was later removed.  Until now my attempts to recover it had only had partial success, but somehow this time I managed to recover the lot.  I have packaged this together with some hand-transcribed results from 1943-1959, as well as some results missing even from Morgan's site but present in old paper archives.  (I thank the staff of the University of Tasmania Morris Miller Library for their help over the years with these items.)

Morgan is not a very well regarded pollster these days, but the classic Morgan Gallups were a really good polling series given the challenges of polling at the time.  So if readers would like an extensive spreadsheet of Morgan federal voting intention primaries (1943-2000, with Morgan's 2PPs for 1993-2000 thrown in) for non-profit interest or research purposes only, just email me at k_bonham@tassie.net.au .  Over time I hope to convert these primaries to 2PPs and do some historic graphs for terms of government in the pre-Newspoll era.

Update 19 Dec: There's a ReachTEL commissioned by The Australia Institute (leftwing thinktank) of Tony Abbott's electorate of Warringah.  It is reported as showing 50.9% believing Abbott should quit the seat at the next election, and 36.7% saying they would be more likely to vote Liberal if he did.  As usual with these polls the latter result should be ignored as such results always exaggerate the impact of an event massively.  Further results of the poll are here and judgement should be suspended until the full question wordings have been seen.

Update 22 Dec: The full results of the Warringah ReachTEL are available for download from TAI's site. The poll includes voting intention results suggesting that the eviction of the sitting member from the Prime Ministership has done nothing to harm the Liberal Party's standing; in fact the Liberals are on 62.2% (+1.2) after redistributing leaning voters, though the 4.7% swing against Labor on that basis seems a bit much (and the usual caveats about seat-poll unreliability apply).

39.4% in this poll supported raising the GST to 15%, with 46.5% opposed.  However the increase has net support among Coalition voters (52.2:32.4) with strong opposition from the minority left vote in the electorate.  The question about Tony Abbott retiring does unnecessarily stress his status as "former prime minister", which might affect the results slightly - the key point here being that the voting intention shows that if he recontests voters will still vote for him.

Lastly, TAI have depicted Abbott's "war on renewables" as a major part of his downfall, and claim strong support for "gradually transitioning to 100 per cent renewable energy" as evidence for this.  I think it is very likely true that Abbott's idiosyncratic attitude to windfarms fuelled the view that he was out of touch in a rather strange way, but I wouldn't read too much into this poll result as further evidence.  It's a bit of a pony poll - the sort of thing that sounds great on paper, but where people's commitment might waver when confronted with concrete costs and employment outcomes.

No comments:

Post a Comment

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. 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.