Friday, March 22, 2019

NSW 2019: Final Day Roundup

SUMMARY: Polls imply likely Coalition victory, but if so, majority status is still touch and go.

This thread will cover anything I think is of interest re Saturday's election, starting with a pre-Newspoll look at where things might stand (but there is some weird stuff going on, so who knows) and followed up with anything I want to add once Newspoll (or any other late polls) come out.  I'll be working in the afternoon, so don't expect updates between about 11 am - 6 pm.

There will be live coverage here on election night from 6:00 pm and going through til very late, and the live thread may be started earlier in the day if there is anything of special interest going on.


Statewide polling and modelling

Since the previous post there's been just one more statewide poll, a YouGov-Galaxy with a 50-50 two-party preferred result (barely enough to move the dial).  The primaries were Coalition 41 Labor 38 Greens 9 Shooters, Fishers and Farmers 3, One Nation 1 and others 8.  The SF+F and One Nation figures clearly reflect that the pollster has adjusted for these parties not running in many seats, but I have not seen any confirmation on whether they polled these parties only in the seats they were contesting, or used some other adjustment method.  The poll was a glass-half-full, glass-half-empty exercise for Labor.  To one degree it showed at least that the party was getting a primary vote swing, which it desperately needs to win seats in the Sydney metro area and avoid a scenario of the 2PP swing being wasted on rural seats where Labor finishes third.  On the other hand, the 2PP swing just didn't look to be quite enough.  (There was some good news for Labor at the time in a close 38-36 result in Gladys Berejiklian's favour in the preferred Premier question.)

In an aggregate, this poll could only move the dial by tenths of a percent.  Pending any further polling, I'm using 50.3 to Labor, based on which the projected results in my model are about the same as the previous update - so close that I don't see any point in posting fresh conditional seat probabilities.  Two seat polls (Goulburn 50-50 and Penrith 51-49 to Liberal) also didn't add a lot of action to my model.  If all this was all we knew, I'd say that the Coalition deserved to be very slight favourites, but were likely to fall a seat or two shy of majority.

However, it's been a strange week.  Labor's final campaign week has been regarded as a shocker, partly off the release of Michael Daley's September 2018 comments about "young people from typically Asia with PhDs" causing "Our young children" to "flee" from areas with rising house prices.  Daley has apologised for these comments.  However Daley also had a very poor appearance in the People's Forum debate (though the Premier's performance was hardly spotless either), mixing up figures from areas of Labor policy strength and generally being attacked as not up to it after the election.  The "vibe" is suddenly that the Coalition will not only win but could cut its losses enough to do so with a small majority.

The Asians-with-PhDs preheated gaffe has resulted in bizarre scenes in the final days as Labor is forced to sandbag Kogarah (which it ought to be winning 62-38) on account of the electorate's very high Chinese population.  What is not getting so much attention in all of this is that there are a pack of rural and regional seats where nobody cares about inner-city housing prices or whether or not Michael Daley can remember what he's spending on TAFE.  As Peter Brent has observed, within weeks Labor's final week of shockers will be either "forgotten" or the obvious reason for the result.

Both sides have done enough to lose this election and the post-election narrative about causes and issues will probably be very different to the one we're hearing now.  William Bowe is amusing about how this played out in Victoria (about nine minutes from the end of this Tally Room podcast.)

The Newspoll will be very interesting in terms of whether there has been any late movement to the government or not.  

Betting

Betting odds are not reliably predictive but they give some good examples of how expectations are shifting.  The implied chance of a majority government on one of the major exchanges has grown from 24% (which was always too low anyway) to 42% over the last few days.  The markets still have the Coalition favoures to lose in East Hills, Coogee, Tweed, Lismore and Upper Hunter and have now added Barwon to the mix (the Shooters are narrowly favoured to grab it).  However, there are only four LNP occupied seats where Labor are now at $3 or less, while on the Labor side, Kogarah (can such things be true?) has been added to the close-seat list.  Ballina is the only seat where there is nobody under $2 because of the three-cornered nature of the contest.  The Coalition are also rapidly firming statewide, with their implied chance of winning up from 54% a few days ago to 64% now [edit Friday morning: now 68%].

Minority scenarios

One of the odd aspects of this election is that if it does come to a minority government situation, the independents won't necessarily support Labor (despite two of the three current independents being recognisably left-wing) while the Shooters, who one would have thought of as a conservative party, won't necessarily support the Coalition.  Indeed the Coalition has said it will not accept Shooters support.  This is all among the oddities of political life in NSW where the Liberal Party at state level is very moderate, and is also up against a rather moderate Labor Opposition.

There has been some canvassing of a possible Labor-Greens-Shooters alliance.  

Upper House

Unfortunately I have not found the time I was hoping to to have a good look at the Upper House so I refer readers to Ben Raue (whose overall in-depth preview coverage of NSW has been terrific), William Bowe and Antony Green.   The Liberal Democrats' David Leyonhjelm faces a challenge because he has drawn to the right of the Liberals on the ballot paper, which normally results in LDP candidates polling less than 2% of the vote.  I may have time to post some further comments tonight.

Update: Newspoll

The Newspoll has come out and it is 51-49 to Coalition, off a 41-35 primary vote lead.  The primary vote for Labor is a problem if this poll is accurate, and Michael Daley's netsat (never all that high) has been damaged by the events of this week, knocking him down to -15.  When I feed the 2PP into my seat model, it gives the Coalition 47 or 48 seats on 2PP, but that includes seats (most seriously Barwon but potentially a handful of others) that it could drop to indies or Shooters.  However the weak Labor primary suggests this could be conservative.  The model only wants to give Labor Lismore, East Hills, Monaro (which I'd be inclined to override) and Coogee, with Upper Hunter and Tweed tossups and also Goulburn close to that status.

There is a breakdown into Sydney seats (52-48 to Coalition) and the rest of NSW (50-50).  Those are swings of about 2.1% and 4.5% respectively.  This fits with the view (it's the sort of election where I have to stop myself from saying "vibe" all the time) that the government and its campaign have been Sydney-centric.  If I throw in a Sydney vs non-Sydney factor into my model using this poll then the government wins the 2PP in 47 seats, which all else being equal puts it one loss to the Shooters or an indie away from minority government.  With the Sydney/non-Sydney factor in play, the model now gives Labor Lismore, East Hills, Monaro (caution), Upper Hunter, Tweed and by a whisker Goulburn, but not Coogee.  But it gives the Coalition 47 not 46 despite the six projected losses because of the small chance a Labor seat drops, and because outside of the seats being mentioned here, it doesn't see all that much else as in play.

Even a 51-49 Newspoll is at best marginal for the Coalition in terms of majority status.  Factoring in previous polls there is still not an objective data basis for the view that the Coalition will win a majority, but following this poll a lot of people are bound to expect that they will.

Using the weights of my federal polling model, an aggregate of 50.3 to Coalition gives a seat reading of 46 Coalition and 40 Labor, with the seven crossbench seats not included in the model and assumed to be held by their owners absent of any indication otherwise.  

No comments:

Post a Comment