Sunday, March 29, 2015

New South Wales: Decisive Win For Coalition

Seats apparently won: Coalition 52 Labor 32 Green 3 Ind 2
(Includes some seats not absolutely certain)
Seats in doubt and not included in list above:

East Hills: likely Liberal retain
Gosford: Labor leading
The Entrance: likely Labor win
Lismore: National vs Green, Nationals currently slightly favoured

Expected outcome Coalition 53-54 Labor 34 Green 3-4 Ind 2

Normal Programming Resumes

The Liberal-National coalition under Premier Mike Baird has won the NSW state election comfortably and decisively despite a large swing back from the massive margin in the 2011 poll.  On present counting, both the primary votes and the seat distribution will finish up about as predicted before the election.  The main surprise is a strong seat performance by the Greens, who without much increasing their statewide vote have won three or perhaps four seats.


After the predictable Victorian state election, the Queensland election threw up a surprise when Green and minor party preferencing patterns changed enormously, and this wasn't even muted by polling errors in the opposite direction.  Instead of a 52% 2PP win, the LNP government ended up with 48.9% and lost. 

The NSW election has been hard to predict exactly because of this same issue but in the end it looks like the preference flow change has been much weaker than in Queensland and indeed that even the most conservative respondent-allocated preferences overestimated it slightly.  The gap between 2011 preferences and 2015 preferences looks like being nothing like the 3-4 point gaps implied by Newspoll and Ipsos and according to William Bowe, is around half that seen in Queensland.  

William has the 2PP swing at 9.4%, and I earlier got 9.9% using the ABC's projections for 88 of 93 seats (excluding Tamworth).  It will take a while for the dust to settle on these estimates but let's say for now that we're looking at a 2PP somewhere around the high 54s and a swing of nine point something.

On that basis the Mackerras pendulum prediction would have been for the Coalition to lose 17 seats to Labor.  On current counting they appear to have lost 16, though they might save some of those.  The Coalition appears to have retained six seats on below 9% margins (East Hills*, Monaro, Oatley, Coogee, Kiama and Seven Hills) and to have lost five seats on margins above 10% (Gosford*, The Entrance*, Port Stephens, Ballina and Lismore*).  Seats marked * are still in significant doubt.  

What is striking is the extreme variation in swings by seat, ranging from Oatley where the Coalition received a nearly 4% swing, to Ballina where the swing against is currently running at 32. I have the standard deviation of the seat swings at an enormous 6.7%. 

Here's the usual graph of swings required for Labor to win seats against swings achieved:


A number of seats are labelled: EH East Hills, MO Monaro, OA Oatley, ST Strathfield, G/TE Gosford/The Entrance (these two fall very close together on the graph and look like a single dot), PS Port Stephens, CT Charlestown, BA Ballina, LI Lismore (these two shown in green).  In the case of Charlestown, the swing required is from the previous election, and the disruption of the seat being won by Labor in a by-election mid-term has blown out Labor's margin.

There is enormous variation all over the pendulum but especially among the Coalition seats within the normally winnable range.  This could have been a really underwhelming seat tally for the left but for the North Coast outliers and what looks like a couple of narrow wins in Gosford and The Entrance (though both of those are still in doubt).  

The Coalition has had relatively few close seat wins and on the pendulum for the next election Labor's task may look extremely difficult, since it is likely to need a uniform swing of a further 7% to win next time around.  However, as we have seen in this election, swings are not uniform, and in fact it looks like there will be a large number of Coalition seats on 7-8% margins, meaning that a swing of five or six percent would probably capture enough of those.  This is especially the case as none of these seats will have the same level of personal vote protection that the Coalition had in twelve seats this time around.

For those wondering about such things: the swing against the Coalition in double sophomore seats (those its incumbent won from the Labor incumbent in 2011) was a mere 5.0%, over four points below the state average.  However, I don't think this election is a reliable test of personal vote theories, because of the massive regional swing variations.  All we can say is that this is yet another election that could have debunked the sophomore effect theory but didn't.

The NSW election confirmed the impression from Victoria that the Greens' ability to win targeted seats has improved.  Once upon a time they would perennially poll well in inner city seats in election leadups but then consistently underperform on the day.  This is the second state election in which we've seen them able to win target seats and do so from any party.  

The Scale of The Win

Some may seek to diminish the scale of the Coalition's thrashing of Labor in this election by pointing out that the Coalition suffered a swing against it of nearly 10 points and lost numerous seats.  However, in an electorate with diminished party loyalties, it is no longer the case that a party thrashed at one election is condemned to spend the next few rebuilding gradually.  (Perhaps it never was - federally, there is almost no link between the 2PP result at one election and the next, and this has been the case for a long time). Queensland has showed that: a government that had won the previous election massively was evicted in a single term.  

It follows that we should be more inclined to view each result in isolation, accepting that the personal-vote advantages of new governments help them slightly in  the seat tally but that otherwise any party can win, whatever its previous result, if it deserves to.  

In this regard the result shows both strong support for Mike Baird as Premier and a general lack of enthusiasm for the NSW ALP.  The latter also showed up in Labor's poor results in its battles with the Greens, who beat Labor everywhere they reasonably could. (In the case of Lismore and Ballina, I suspect voters concerned about coal seam gas eventually decided Labor was at least as guilty as the Coalition.)

The result is especially impressive given the influence of federal drag in state elections.  It is challenging for any government to win re-election (or win an election from opposition) while the same party is in power federally, and more so if the federal government is unpopular.  The Coalition won no state elections outright after the first year of John Howard's 11-year reign, and none at all after its second. 

The last performance this strong by an incumbent state government that was also in power federally was the Coalition's under Richard Court in the 1996 WA state election (55.2% 2PP) and at that time the Howard government had been in power only nine months and was polling similarly strongly.

For all their public comments about being pleased to be back in the game, this result shows that NSW Labor still has a lot of work to do to recover the trust of voters after its previous term in government.

Poll Accuracy - Provisional

It is too soon to judge which was the best final poll of this election; this sort of task should always wait until all the numbers are in, and in this case below 60% has been counted.  (The total eventually reaches around 90% as some people do not vote).  However as of election night the final state polls by ReachTEL and Galaxy (the former especially) were very close to the current primaries, which will move slightly in post-counting.  Ipsos, Lonergan and especially the final Morgan seem to have the primary gap too high and Newspoll and especially Essential seem to have it too low.  It remains to be seen who had the most accurate 2PP, with several pollsters on 54 and 55.  

The two pollsters who put out what seem to be the most accurate final polls were also the only pollsters to release seat polling.  The seat polls however have not gone so well - on present figures each has the right winner in only one seat out of three! In Galaxy's case the primary votes for Campbelltown were way out while in the case of The Entrance they were only two points out, but those two points appear to have made the difference.  In ReachTEL's there was vast underestimation of the Green vote in both Newtown and Ballina - in the former case perhaps connected to difficulties in polling inner-city electorates with high proportions of young voters.  I believe there was some late swing from Labor to the Greens in Newtown at least but not enough to account for the huge scale of the Greens' win there, with their highest primary vote (47%!) in a single state or federal seat anywhere in the country ever.

I will soon put up a postcount thread for the seats in serious doubt and the Legislative Council.

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.