Friday, March 4, 2016

EMRS: Libs Could Have Copped A Whacking, But ...

EMRS Liberal 46 Labor 27 Green 18 Independent 8 Others 1
Interpretation (provisional) Liberal 47 Labor 31 Green 15 Others 7
Result of poll if election "held now": Probable Liberal majority government (13-9-3)
Aggregate of all state polling: Liberal majority government (13-8-4)

The new EMRS poll of Tasmanian voting intentions is out (PDF link) and the results are surprisingly harmless for the Hodgman Liberal government.  Given that the gloss has come off the federal Turnbull bounce which probably helped the Liberals to such a good result last time, there were many reasons to suspect this poll could be a shocker.

In the last few months the state has been beseiged by bushfires, drought, floods, traffic congestion in Hobart, energy issues and talk of a crisis as a result of a still-unfixed fault in the Basslink power cable, and even an outbreak of mass oyster death.  While most of these are just "natural disasters" that shouldn't hurt governments (and might even help them if their response was good), responses to both the energy and traffic issues have raised the question of preparedness.  While the preparedness question applies to previous governments at least as much as this one, and goes to the question of whether government generally is all that good these days, one might expect the current government to cop some blowback from it all.  Perhaps in the energy case we'll need to wait and see whether it actually affects consumers first.



This isn't all; the government has also lost Resources Minister Paul Harriss for "family reasons" (the reason that always excites conspiracy theories immediately, especially in the case of someone who appeared to enjoy socking it to the Greens to the point of addiction) and Energy Minister Matthew Groom has had a troubled time (example here).  So I would not have been surprised in the least to see a more substantial loss of government support than a statistically inconclusive two points, which, again, could be explained by federal factors.

EMRS has a long-established habit of skewing to the Greens compared to actual election results, and its estimates of the "Independent" vote are far too high to take seriously in the absence of any known independent candidate.  Pollsters that include "Independent" as an option alone in the readout generally have this problem but I also think that with declining landline response rates, EMRS is capturing a disproportionate sample of highly politically engaged voters.  The pollster's trend tracker shows a continuing rise for "independent" lately and I think this is a method issue and not real.  Who these people will really vote for when they have to choose between Labor, Liberal, Green and very minor indies they have never heard of is an important question.

After considering what is known of EMRS's house effect, this poll implies a familiar picture.  If the trends based on my "interpretation" scores were repeated evenly statewide, the Liberals would lose their fluked fourth seat in Braddon and also their third seat in Franklin.  The key seat is always Lyons where the projection would have the Liberals on about 2.75 quotas, Labor on 2.14, Green 0.83.  Based on minor party preferences in this electorate in 2014, on the Greens' lack of an incumbent with profile, and on the potential for the Liberals to split votes between their candidates, I don't think the Greens would recover their former seat here.  Note also that the Shooters and Fishers could be a presence in this seat for the first time in 2018, and are well capable of polling say 4% in Lyons, though that is not enough to win a seat.

The assumptions of uniform swing could well be wrong, but while that could cost the Liberals their Lyons seat, it might (for instance) save them their Franklin one. That said, Franklin is going to be even more challenging for the Liberals to hold three in now with new MHA Nic Street having just two years to build his profile sufficiently to hold it should the party's general support level be high enough.

There is also a history of voters flocking to whichever party can win majority government, so if the government was to go into a campaign with polling like this it would be very likely to retain office.

Leaderships

There is not a lot to see in the preferred-premier ratings with Will Hodgman down four points to 52 and Bryan Green recovering two points to 21.   Incumbents normally have large leads in this figure, especially over relatively new opposition leaders.  Green has made some attacks on the government that appeared to be effective at the time but the poll does not suggest much attention is being paid.  There could be some relief that Green is not stuck in a sub-20% rut after his poor result last time created much discussion.

Aggregate

Because individual poll results can be misleading I am keeping a running aggregate of Tasmanian state polling.  As well as the current EMRS poll there was another Morgan state poll but I regard the Morgan Tasmanian state poll, conducted exclusively by SMS, as a trashy poll with massive skew (in more recent polls, all to the Greens and against the Liberals) that is very unreliable even after accounting for its sample size.  Currently I am correcting the Morgan state results by taking 5 points from the Greens and adding 4 points to the Liberals and (perhaps unnecessarily) 1 to Others.

I have weighted the current EMRS poll at 40% and the February Morgan at 3% after adjusting each for apparent house effects.  After this there are slight changes in the percentages but no change in the aggregate seat projection which remains at 13-8-4, although 13-9-3 is not much less likely.


3 comments:

  1. Hi Kevin,

    Under proposed change in case I vote above the line will I be limited to ticking up to 6 parties or can I number all of them?

    Regards,
    Igor

    ReplyDelete
  2. You can number as many parties above the line as you like. If you number all parties, that will have the same effect as putting the last-numbered party last, with the exception of any ungrouped candidates.

    ReplyDelete