Scott Morrison will at least continue to lead Australia in minority government and may yet achieve a narrow majority government after a stunning election rebuff to Labor that has finished Bill Shorten’s leadership career. If the Coalition fails to achieve 76 seats, this will be the third election of the last four that has resulted in minority government in Australia, but for Morrison it will be a victory unique in recent decades and a remarkable political feat: virtually single-handedly, he has turned around a massive poll deficit and inflicted a searing defeat on a Labor team that believed itself heading for a narrow but relatively comfortable win.
Instead, unaided by senior members of his team, who had either quit in disgust at the dispatch of Malcolm Turnbull or who were tied down defending their own seats, Morrison was able to hold off Labor with the help of a far-right surge in Queensland where one-in-eight voters backed One Nation or Clive Palmer.
The result is now an electorate more divided than at any time in recent elections. Wealthy metropolitan electorates in New South Wales and Victoria swung hard against the Liberals: Tony Abbott was ejected, Dave Sharma’s expected easy victory in Wentworth is in doubt, and Melbourne voters swung against senior Liberals, while voters in Queensland turned on Labor, shifting to Hanson (8.7%) and to a far lesser extent Palmer (3.4%) — over 12% of the state-wide vote compared to just over 4% in NSW and Victoria, and enough to put One Nation’s Malcolm Roberts back in the Senate. With little movement elsewhere, and Hanson and Palmer preferences flowing strongly to the Coalition, it enabled Morrison to halt the swing to Labor needed to deliver 76 seats and retain a real chance of majority government in a result that could be weeks away.
It is a singular achievement for a leader whose party has not led in a single opinion poll for several years. Morrison now takes his place in the pantheon of Liberal heroes, delivering a victory many even on his own side thought impossible with an aggressive scare campaign that targeted Labor’s range of high-profile policies. Refusing to accept the defeat pollsters foretold over and over, Morrison kept hammering at Labor, a virtual one-man band without a treasurer or home affairs minister to back him on the hustings.
Labor’s failure is stunning and the autopsy on Bill Shorten’s campaign and leadership will be prolonged. Even the most pessimistic Labor hands thought they’d get between 76 and 80 seats. Instead, Shorten’s lack of popularity and a deliberate strategy of high-profile tax policies that created vocal losers appears to have prevented Labor from kicking clear of the government, in a campaign where Shorten failed to produce the same effective on-the-ground campaigning as in 2016 — and in which he was up against a much better campaigner than Malcolm Turnbull. Late last night, Shorten announced his resignation as leader.
In Queensland, Labor may have also suffered for its perceived opposition to the Adani Carmichael coal project, but that fails to explain why even Brisbane seats swung hard against the opposition — something that the Queensland ALP will have to account for to interstate colleagues given the state was the burial ground of Labor hopes.
The other losers are opinion pollsters and bookies, who both managed to get it wrong. The polls were uniform around 52-48 and 51-49 to Labor, even up to exit polls late on Saturday — not a single poll during the entire election had the Coalition ahead, a remarkable result given the ordinary vagaries of statistics. But they, like the Queensland state polls that showed a swing to Labor, proved badly off-beam. The bookies’ predictions of a comfortable Labor win were even more wrong. This was Morrison’s night, and Liberals will celebrate his remarkable victory for a long time to come.
A hell of a lot of thinking to do now.
A minor point, just on the uniformly wrong polls, exit polls and betting market – Anthony Green says the ubiquity of mobiles means pollsters don’t know who they are talking to and so can’t sample properly. But surely that would lead to a randomised wide spray of results. Instead, from Trump, to Brexit to this result, in the last few years polls have been in error in one direction by up to 2%. It must be that those intending to vote right (or who have, in exit polls) are now consistently more likely to disguise their intentions if asked. Two recent comments from Abbott suggest the reason: “Hawke’s heart was Labour but his head was Liberal”, and “when climate change is a moral issue we do poorly, when it is an economic issue we do well”. People don’t like to admit to intentions which they feel others will perceive as lacking compassion or less than moral. But they feel vindicated (usually with a slightly aggressive fuck-you edge) if they are proved to be in the majority.
Anyway, something is creating this new reality, across the board, in polling. How can strategists factor it in, or deal with it?
Cass Sunstein has written extensively on what he calls “preference falsification”: “Under the pressure of social norms, people sometimes falsify their preferences. They do not feel free to say or do as they wish.”
Like it or not, some elements of left wing politics may have had this effect to a very profound extent.
Even I, as a rusted-on left winger on economics and the environment, find myself mute on a number of issues related to gender identity and cultural change. I wonder if that is in play.
And yet the same issue doesn’t show up in state polling. It didn’t show up in 2016. It didn’t show up at the time of the Super Saturday byelections. I doubt it is suddenly true now.
“Across the board” is obviously wrong. However, it may be that a small but significant section of swinging voters will always say the are going to vote for “compassion and fairness” regardless of their actual intentions. In the case of the Andrew’s victory in Victoria, say, they happened to be telling the truth, because they also saw his government as representing their self-interest.
People like their self-interest represented, and to feel good – in that order.
I don’t think this phenomenon (if it is a thing) is new or suddenly true – my memory is that polls and expectations in 1972 were somewhat ahead of the final margin of victory and that polls often overstated the Labour vote in those times. But the sharper divisions of contemporary politics – the move to the populist right – appear to be providing more occasions for it to re-surface.
> Anthony Green says the ubiquity of mobiles means pollsters don’t know who they
> are talking to and so can’t sample properly. But surely that would lead to a
> randomised wide spray of results.
One of the golden rules in regard to sampling and estimation (distinct but related topics in statistics) is not to make assumptions about the assumptions. But let’s take last things (per your post) first.
> But they feel vindicated (usually with a slightly aggressive fuck-you edge) if they are
> proved to be in the majority. Anyway, something is creating this new reality,
> across the board, in polling. How can strategists factor it in, or deal with it
Modern politics, from circa the 1830s (per se) has itself to blame here. We have ALL encountered the statement from a politician : “the solution to the problem is xyz and the majority of (e.g.) the country – or whomever – agree”.
For a hitherto unencountered problem (in business or society or whatever) the minority are correct to the extent that the majority are never (strong word – but history supports the assertion – from Evolution,
to the properties of gravity to the solution to stagflation in the late 70s etc.) correct.
Random samples are obtained at GREAT expense. The HSBC Bank (one of many) is a collective expert in Big Data and Machine Learning and knows when YOU are about to make a purchase prior to YOU doing so! That bank has impressive stats and data frames on god-knows what and everything but the stuff is (obviously) confidential (but available on occasion).
The pollsters that exist in Oz do what snake-oil merchants have always done; they provide a minimal but plausible product to a grossly ignorant public. Anyone with the least clue could have obtained a random sample for any major town (or region) north of Brisbane but, like anything, the project would have cost money and it would have had to have had a range from mid Feb to mid May at least. There would have had to have been significant quantities of trial data and test data. It would have been a large but entirely feasible project but well out of the range for the average bookie or newsroom.
Now, lets take a look at some specifics that you mention. Consider Abbott’s remark : “Hawke’s heart was Labour but his head was Liberal” is ENTIRELY ACCURATE. Fraser did NOT touch one of Whitlam’s reforms; keeping in mind that Whitlam changed the political face of Australia by wrenching it from the late 1940s to the 70s. As an aside Whitlam was not popular within his own party and was almost expelled twice.
Hawke, by comparison trashed a good many reforms from education and, (to be fair – against his own wishes) constructed the path to negative gearing and thus neo-Libed the entire tax structure. Clinton (both on reflection) are much more Republican than Democrat. A host of welfare provisions disappeared under the
WJ Clinton administration. As for Fraser there is an argument that he was a frustrated Labor politician. His “heart”, on egalitarian issues,was much closer to Labor (particularly in the 60s) than to that of his own party.
The entire “Abbot” remark illustrated (1) the poverty of Australian politics – to say noting of the collective ignorance of the topic – and (2) the knee-jerking that is inimical to Australian politics. To this end Australian politics will never be a mature product.
Thanks for your comments Kyle, and good to see you back in these pages.
My point about Green’s mobile phone theory is that if the pollsters now do not know who they are speaking to they can’t lazily fall back on their usual patterns (I think I’m paraphrasing something you wrote in another post). The play-safe herding effect would weaken and the results from the different polls would spray around (I withdraw my careless use of “randomised”) – that is, we wouldn’t have the extraordinary degree of conformity witnessed this time.
While it is no doubt true that Australian pollsters provide a cheap product – the least that will satisfy an undemanding market – nonetheless they have not historically done such a uniformly bad job as they did this time. In the past, some of the polling conducted by parties (when leaked) proved superior to the media products, and the fabled bookies (whatever their sources) have had their day in the sun too. This time, my son tells me, there was a point on Friday when Betfair was offering $10 about the LNP and $1.10 about Labour. It is the uniformity of failure that seems puzzling, to me.
It may be, when the dust settles, that you will be proved quite correct and the problem was simply a failure to poll regional Qld. I wouldn’t know what efforts were or were not put in there and by whom. But it also seems there was an underestimate (less dramatic) of the LNP preferred vote throughout Australia.
Anyway, sooner or later there will be an analysis that puts my little puzzles to rest. In the meantime, I wish I had taken the $10 and I wish Australia were large enough to have our own version of Nate Silver’s 538.
On Hawke – I am an admirer of Hawke (no doubt a sentimental admirer from your perspective), but/and I quite agree with you about Abbott’s comment on him. I found the outrage at Abbott over that both confected and puzzling. Of Abbott’s two remarks, though, I think the second one – about climate change and morality versus economic interest – is more supportive of my speculations about the hiding of self-interested, non-“compassionate” voting intentions from pollsters. In any case, the use of the word “moral” was quite an interesting and significant admission from a Catholic Liberal politician whose party trades on religious allegiances.
It is a very big topic Keith and given the sophistication of the market (nowadays) it is not going to get smaller.
The major point is that “in the old days” the answers to simple questions by pollsters were (1) more or less forthright and (2) uncorrelated. Constructing inferences on uncorrelated data is relatively straight forward but rather complicated for data that has even a trace of correlation.
The Scottish and Brexit polls possess data that is highly correlated which (over-illustrating the point) prevents the forest being seen for the trees. If the pollsters do not know (even in broad terms) who they are polling then the sample is quite literally
useless.
Polling x thousand mobile numbers and asking questions is just plain stupid. Polling x thousand mobile numbers with questions that could provide a stratified sample (a technical term) is entirely feasible. I envisage that the respondents would have to be induced to answer questions and a payment of (e.g. $350 – $500) would have to be made per person; I would not participate for less.
Then (2) the replies would have to be tested for accuracy and the samples (i.e. the participants) placed into “strata”. Then they would have to be offered email accounts (with a life of e.g. 96 hours) with more inducement to answer questions along with with a means of statistical control.
Having got the “training data” their responses on “test” data would have to match within a pre-determined error range; otherwise the sample would have to be jettisoned and the process begun anew. As you may appreciate the former methods no longer work and the “new” methods (if you prefer) cost quite a bit to implement.
Its not for nothing that FB pulls the revenues that it does. A good deal of analysis has been applied to their “customers” or members in exchange for free FB products.
If the same analytics are to be adopted in Oz for elections, or anything else, to quote Hawking, we would come to know “the mind of God” (but not otherwise)!
On the point of “the sophistication of the market” our civilisation is on the threshold of being able to quantify the information of the market. Within a matter of a few years the tools AND computing power will exist to literally model the market in real time.
In many respects we live in a fascinating age. A major question remains : what (morally) do we do with the information?
Easy answer, we use the information to make the wealthy wealthier.
Kyle, I see the LNP claim their internal polling had it right. They may be lying, but assuming they are not, any thoughts on how they might have managed it?
https://www.smh.com.au/federal-election-2019/very-tight-very-deliberate-internal-liberal-polling-showed-coalition-was-on-track-to-win-20190521-p51pi0.html
“Fraser did NOT touch one of Whitlam’s reforms; ”
Fraser got rid of Goughs Medibank. Looking forward to your explanation of how an obvious fact of history somehow didn’t happen.
You go Kyle.
> Fraser got rid of Goughs Medibank. Looking forward to your explanation of how an obvious fact of history somehow didn’t happen.
You tend to “specialise” on the surface and not in the depths DB. We have two options DB and I don’t mind what you select. I could construct a table of Whitlams reforms AND an analysis of legalisation under Fraser (75-Feb 83) OR I could point out that the Medibank was not at all popular and not considered as a final product. Two links will prove interesting. The first describes the general dissatisfaction of the initial plan, namely, https://sydney.edu.au/news-opinion/news/2014/11/05/whitlam–medibank-and-health-system-reform.html and this link johnmenadue.com/john-menadue-the-ama-did-its-best-to-scuttle-medicare-over-40-years-ago
gave Fraser little option but to come to heel with Australia’s largest and most influential trade union.
As an aside the election occurred early in 1983 because at a lunch of business gurus in Melbourne Fraser was “directed” to “go early”. A very grave mistake. Fraser was vulnerable to this kind of influence. The economy was, in fact, picking up and an election gave Hawke (who had just knifed Hayden – a lot of it about in Labor politics would you not say DB?) something to do.
As to “pollsters” I’m NOT suggesting that the publlic pays for their product OR at least not directly. That the product has been amateurish for quite some
number of years (a decade or more) is all too apparent. The polling is, as I have gone to some trouble to explain, (1) amateurish, (2) uninformative
and (3) not at all comprehensive. I have ALSO pointed out that “real” organisations (e.g. HSBC and JP) do the job correctly and at great expense.
“The pollsters that exist in Oz do what snake-oil merchants have always done; they provide a minimal but plausible product to a grossly ignorant public.”
Actually Kyle, they are provided to two groups almost entirely, news media and the major political parties. The public doesn’t ask for them or pay for them, and largely just gets the confected interpretations from mostly innumerate journalists.
Cheers.
The ALP spent too much time explaining and defending the funding mechanism of their policies, franking credits and neg gearing and not enough time explaining the policy benefits.
Right on.
Yes, that was something obvious even during the campaign. I kept crying out for the ads pushing the cancer payments policy; renewable energy bringing power prices down (something Barnaby Joyce referenced in his victory rant on the ABC – why was the ALP not talking about power costs, which have been demonstrably rising under the Coalition and were a huge sore point with voters); wage increases; and so on. Talk about the winners. When they bring up the costs, connect it to the benefits, pivot to talking about your policy.
The ALP got snowed into talking almost endlessly about the parts of their policies that people didn’t like. They didn’t even bother explaining the point of the negative gearing changes being to make housing more affordable, apparently assuming people would remember from 2016 when they prosecuted that case far more effectively. They just kinda assumed everyone agreed that climate change was an important issue that they didn’t need to win anyone over on. Bill’s “never you mind about the costs, the cost of inaction is more important” strategy might be true in one sense, but left all these economically insecure voters worried that the costs were big and would be paid by them, not paid by the wealthy voters in Kooyong and Warringah who Labor was courting. This reinforced the Adani thing- southern educated voters whose jobs don’t depend on mining telling Queenslanders to sacrifice, while not offering to make any such sacrifice themselves. At least I can say I always thought the myopic focus on Adani, rather than broader climate change issues, was a dumb strategy by environmental groups- and it was (and if they double down on it out of defiance, even dumber).
I could not believe Labor did that “not talking about the costs” thing and half expected a 5 point poll drop to follow it. It turned out that may have happened, sort of.
The Liberal advertising hammered their main points – TAXES! SHORTEN! YOUR MONEY! relentlessly, and Labor didn’t answer those points. The top end of town advertising might make sense against Turnbull and his big business tax cut; it made no sense when your weak spot is retirees and small investors and regional voters who feel they are the losers of your policies and do not define themselves as the top end of town, nor do others see them that way. Labor had 6 years of government bungles to use in advertising and didn’t. They had policies people genuinely liked, like the cancer policy, to use in advertising, and didn’t. Their advertising people should be cleaning out their desks already.
I agree with all of that. I also think that the franking fiasco was a major focus, one that I could never get my head around, until I finally realised that the critics were right. It was disastrous in itself and a lightning rod – people could see how it affected them, or their parents or others, and they thought, “If they’ve got that wrong, why should I trust any of their complicated policies.”
Basically, as I understand it, retirees on low or no tax incomes (some artificially low, but not most) dependent on franked shares were to lose, typically, $1,000 to $4,000 a year. Meanwhile those with higher, taxed incomes were to retain the full value. And those of us on government super (I’m one) were not being asked to make any sacrifices to support Labour’s programs.
No wonder people were screaming.
I’ve just read a very powerful letter on this in the Age
https://www.theage.com.au/national/the-result-chris-bowens-advice-sealed-it-for-this-voter-20190519-h1ekli.html
let’s not give ScoMo too much credit – it was Rupert Wot Won It – if there was any doubt, we now know we live under the control of an unelected, foreign national
Sadly this will probably be the last time that an opposition campaigns on a big proposal and especially the important topic of tax reform. It seems that this is a task to be left to the first year of a new government…..and to be not spoken of before entering government….just ask John Hewson, who may well have made a far better PM than the devious Howard.
As Jack Lang said “in the race of life, always back self interest, at least you know it’s trying”
In the race of life, far too many people do not actually know where their own best interests actually lie.
The lies and prevarications again won the contest; wooing the weak minded and gullible has succeeded once more.
Actually , in the race of adapt and die mortal life people generally do understand (after about the age of five –some slow developers may take until they’re ten ) that allusions ,illusions & lies help get you by with all the tragedy and grief ..A self -reflexive talking mammal of physical consciousness tends to know that something known as ‘a truth ‘ is only one of the many perplexing elements of being alive…
Peter Brent has made this point ad nauseam in Inside Story: you get into government with a modest agenda (“stable, competent government” might have had a nice ring to it), you establish some credibility (especially economic) and then you do your “reforms”.
None of the incoming post-war Labor governments had gone full big target. Gough didn’t have any revenue-raising plans for his program, he was relying on economic growth (didn’t work); Hawke campaigned as Hawke and Rudd as a “fiscal conservative”.
If I had a vote in the ALP leadership (if this were the UK, I could fork out ten bucks and get one, Labor here not that silly), I’d want to know what the candidates think about a strategy for gaining office (with special attention to Qld and WA), not where they stand in the identity politics fashion stakes.
Labor promised change and the Libs promised nothing.
Apparently Australians prefer nothing.
“Relaxed and comfortable”?