Sunday, May 28, 2017

Queensland: One Nation Have Peaked, So Now What?

I last looked at Queensland state polling back in February, when a rampant One Nation were polling 23% in state voting intention and threatening to win something like 15 seats at the next state election.  A lot has happened in the few months since.  In the Western Australian state election, both the party itself and the Liberals' preference deal with it got found out in the heat of the campaign. What benefit on preferences accrued from the deal was drowned out by its distraction factor and damage to primary vote support for all involved.

As One Nation 2.0 comes under further scrutiny, it is under attack on two major fronts.  Firstly there's the suggestion that the party is simply corrupt, as seen in the current scandals surrounding the undeclared provision of a plane by a property developer and the recording of a meeting in which James Ashby seems to have seriously suggested a blatant electoral expenses rort.  Secondly, the party sends mixed messages about whether it is a purist party of revolt against existing politics, or an active inside player that wants to supplant the Nationals on the Coalition side.  This is making it harder for it to take votes away from Labor.

Compared to the February polls which showed One Nation at 23% (Galaxy) and 22.3% (ReachTEL) there is good evidence that the One Nation primary has declined (as it has done to some degree in polling nationwide), but it's not as if the floor has fallen out of it just yet.  The following table shows the April 30 Galaxy, plus two more recent commissioned ReachTELs.

The full source document for the Together Union ReachTEL has not been seen, meaning some of my numbers (in italics) are estimates, while the source for the Australia Institute ReachTEL is available.  As usual I have calculated out the ReachTEL primary votes to distribute the so-called "undecided" and make results comparable to how pollsters should report primaries, and I've also derived a 2PP to one decimal from the data for the TAI poll. Changes from the last election are shown.  Both 2PPs for the ReachTELs are by respondent preferences, a method I generally don't rate, but there is not any easy way to calculate "last-election preferences" as they're normally understood.

I am usually very reluctant to use data from commissioned polls, but in both these cases the primary purpose of the poll seems to have been the issue questions, so I think these results are worth noting (but I treat them with more caution than the Galaxy).

The differences between these polls are insignificant.  All are saying that One Nation has lost five or six points from its February peak, that about half the One Nation gain is coming off the LNP with most of the rest off Other (mainly because Palmer United are gone), and all have a two-party preferred vote that is little different to the 51.1% that Labor recorded in winning in 2015.

The Galaxy poll was closer to a major event which could have generated some bounce for the government (the government's very well received response to Cyclone Debbie  - a similar situation caused a large poll bounce for the Bligh government, but didn't stop it being thrashed eventually).  On this basis the conclusion suggested by the trend tracker (that the return of One Nation has greatly harmed the LNP primary while doing nothing to Labor's) may be a little premature.  However polling just hasn't been as volatile lately as in the Bligh years, and I suspect any bounce would have been much smaller for coming from a much higher base.  As for ReachTEL, the Coalition has had a good run from them in primary votes in many states recently, so this is another reason to treat those results with some caution.

Will there be another preference shift?

Normally in Australian elections if Labor beats the conservative major parties by even a small amount on primary votes (as in the recent polling) then they win the two-party preferred vote very easily.  The idea that they might not do so here rests on an assumption that preferences will flow very much more weakly to Labor this time than in 2015.  The ReachTELs give Labor a stingy proportion of third-party respondent preferences (47% and about 49%) and Galaxy's 53% share (determined by methods unknown to me) is not much higher.    The ReachTEL figures work if one assumes that the Greens are breaking about 75-25 to Labor and that One Nation and others are breaking at about 60-40.

In any case, that would amount to an enormous shift from the 2015 election, where Labor got something like 83% of all Greens preferences that actually flowed (75% if exhausting votes are split evenly) and 67% (60% if splitting exhausting votes) of all Others preferences.  Obviously, not everyone can be campaigning against Campbell Newman and Tony Abbott both at once.  But the suggestion is that with the switch of Others votes to overwhelmingly One Nation (rather than mostly a mix of Palmer and Katter), Labor could perform worse on third-party preferences than it did even in its abysmal 2012 Queensland wipeout.

I can credit, just, that this could happen, but I think the share of respondent preferences for Labor we're seeing in the current polls is about as bad as it could reasonably get, and there's a fair chance it will end up being better.  Perhaps Queensland is different, but the Liberal National Party will presumably be gun-shy about a fully fledged preference deal with One Nation based on the WA experience.  There is a widespread view that Labor's decision to reintroduce compulsory preferencing will come back to bite them, but that assumes that the LNP can actually deal with One Nation to secure a 60+% preference flow without destroying their own campaign in the process.  It also assumes that the One Nation primary will stay somewhere near where it is now.

What would the current polling mean for One Nation?

Modelling how the One Nation vote could play out in different parts of the state is very difficult because of the length of time since the party was this competitive, and the extensive redistributions, as well as the change in the preferencing system.  Booth-by-booth modelling off the 2016 Senate vote might be the way to go here, but I suspect it's above my computer skills grade.

 After winning 11 state seats off 22.7% of the vote in 1998, the party won three seats with 8.7% in 2001. Two of these were seats (Lockyer and Tablelands) that it had won in 1998 with different candidates.  The third 2001 win (Gympie) was because the election overall was so lopsided that the Nationals fell into third place in Gympie, allowing One Nation to beat Labor. It might seem from that history that a 17% vote would be good for, say, 7-8 seats, but it's actually not that simple.

In 1998, One Nation won two types of seats: safe National seats (4) and marginals (7).  Most of its wins were pretty narrow.  One Nation didn't win safe Labor seats because these were typically in major cities, where One Nation has little appeal.  Of the marginals it won, six of the seven were Labor's, but this wasn't because Labor was unusually prone for any reason to losing to One Nation. It was simply because there were so many more Labor marginals than Liberal and National ones.  (Labor did extremely well in the marginals in 1995 and held sixteen of them by under 3% going into 1998).

Applying swings from the 1998 result to the current polling and ignoring the preferencing changes and redistributions since, I find that had the primaries in 2001 been similar to current polling, One Nation would actually have done rather poorly in seat terms.  In most of the marginals, taking five points off One Nation and improving the Coalition vote slightly would have put One Nation in third place, eliminating them from the seat.  By this model, off 17% in 2001 but had there been a closer statewide context, One Nation would have only won three or four seats.  A lot has changed since, but having lots of close contests between the majors should not help One Nation convert vote share to seat share, unless their vote share is well into the 20s.  So on current numbers we're probably looking at a few One Nation seats rather than a swag of them, reducing the change of a hung parliament.

One obvious One Nation target is Lockyer, a seat the party has frequently won or nearly won in the past, and where the sitting member is retiring.

In spruiking the Together poll, Together secretary Alex Scott said that the result would "clearly" see a "significant number of rural seats probably fall" to One Nation.  I expect Together have finer detail local polling but the public polling doesn't clearly support such comments, which might be seen as intended to spook the LNP.

What sort of 2PP do the major parties need for a majority?

 Using Antony Green's estimated margins and taking into account new personal votes for sitting members, I get that both Labor and the LNP need about 50.6% to win the two-party-preferred vote in 47 of the 91 seats excluding KAP-held Traeger and independent-held Nicklin.  Nicklin might revert to the LNP, but it's also possible that if Peter Wellington endorses a successor, that successor might retain it.  On current numbers I'm throwing about four seats to One Nation (but that may change should their polling decline further) and on that basis I get about 51.2% as the target for both Labor and the LNP for a 50-50 chance of an outright majority - a swing of 0.1% for Labor or 2.3% for the LNP.   If I give the LNP a 50% chance of winning Nicklin, then I get their target score at 50.9%, a 2% swing.  These are all all-else-being-equal conclusions that can easily be undone by locally uneven swings.

The concern for Labor would be that, unlike in the current parliament, most hung parliament results possible in the new one will result in an LNP government.  A very close election on a two-party basis might still result in an outright Labor win, because of their personal vote advantages from the last election, but on balance the zone between 49% and 51% doesn't look a very safe place for Labor to be.

For the time being I have treated the three seats where incumbents have quit their parties as returning to their party fold.  I'm not convinced any of them will be re-elected.


The Adani Carmichael coal mine issue is one of the biggest environmental issues to impact on an Australian election campaign for some time.  Predictable responses, usually drawn from questions that tell only half the story or blatantly lead the respondent, have been seen in many commissioned polls (such as the TAI one, or another reported just before).  Nonetheless the mine is a hard sell, especially for a Labor government relying on Green preferences, on the basis of both its impact on global climate change and its potential to pollute the Great Barrier Reef.  Perceived local impacts and the whole question of governments subsidising overseas developments in search of local jobs have also come into play.  This has finally spooked the Palaszczuk Government, with the announcements in the last 48 hours that the company must pay full royalties and the government won't facilitate loans.

It will be interesting to see if this improves the government's share of preferences in future respondent-allocated polls.  It will also be interesting to see how well the LNP can walk the line between attacking the government for putting jobs at risk, and being seen as itself unconcerned about the environment.


The only new leader rating polling I've seen lately came from Galaxy, which had Annastacia Palaszczuk on a personal net rating of +12 (47-35) and Tim Nicholls on a pretty lousy -18 (27-45). Palaszczuk had a large lead as preferred Premier (48-28), but bear in mind that preferred Premier polling is generally skewed to incumbents (federally, by about 16 points, and not much less at state level).

Electoral history

As Queensland governments go, the Palaszczuk government has been inoffensive, providing a political and polling "Mogadon state" (Graeme Orr on Twitter) after the chaotic adventures of the Bligh and Newman years. It also has the good fortune to be going to an election with the opposite side in power federally.  Despite me repeatedly stressing what a massive factor this is in state elections, awareness as yet to filter through to the mainstream commentariat.  Only six of the last 33 Australian state governments that went to elections with the opposite party in power federally have lost.  Only one of those was in decent political health at the time - the remaining governments had collapsed in chaos, lost their majorities at the previous election, or made enormous blunders.

Also, state governments more often lose with age.  The Palaszczuk government is a first-term government, and in the last 50 years first-term state governments have won 72% of the time.

Considering state governments with both advantages (first-term and opposing party in power federally) their electoral track record is awesome.  The last one to actually lose was the Walsh/Dunstan Labor government in South Australia in 1968, and it was diddled by the Playmander (Labor got 53.2% of the two-party vote).  All eleven state governments to seek a second term with the opposite party in power federally since then have been returned.  Eight of those wins were landslides with at least 55% of the two-party vote (or estimated 2PP vote for Tasmania).  Two of the remaining three scored over 52% 2PP, and the only one remaining (Ray Groom's Liberals losing their majority in Tasmania 1996) doesn't really count as it was fought during a federal election campaign where the incumbent Labor government was clearly on the way out.  Leaders like Peter Beattie, Bob Carr and Steve Bracks became acclaimed as political geniuses for their landslide re-elections when all they were doing was what leaders in their position always do.

By historic standards then, Palaszczuk's government should smash it out of the park, or even if it doesn't win easily, it should still win.

What is the argument for an upset?  Firstly, the LNP lost the last election with an assist from a very unpopular Abbott federal government (this was just after Abbott's farcical decision to confer a knighthood on Prince Philip).  The last-election result may have included a protest-vote component from voters who didn't really expect Labor to win and were sending a message rather than being actual Labor supporters, so it's possible Labor's support base in this term in office has always been weak.

Secondly, the government started life as a minority government (which creates an increased chance of defeat).  This would not have been an issue had it kept its original 44 seats and relied only on the support of the highly respected independent Peter Wellington.  However, the losses of Billy Gordon and Rob Pyne to the crossbenches have frequently created uncertainty, and reflect on the government as it preselected them.

Thirdly there is the crossbench factor - the possibility that Labor could win the 2PP fairly easily and yet still face unlucky local losses to One Nation or others, as a result of which a conservative coalition forms.  If anything, this scenario is fading as the One Nation vote declines.

Fourthly, there are all the usual arguments that might be made about whether the government is actually any good.

In my view, the second of these is the government's biggest problem - had it not been rocked by these defections, I suspect it would be romping in the polls and there would be nothing to talk about concerning the next Queensland election.  Even with these problems, anyone who wants to argue that the LNP are favourites is up against some serious electoral history.

The election doesn't have to be held until May 2018, but is generally expected to be held sometime this year.

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. 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.