Thursday, September 1, 2016

EMRS: Liberals Rebound, But Still Shaky

EMRS August 2016: Liberal 41 Labor 31 Green 15 "Independent" 11 Others 2
Intepretation: Liberal 42.5 Labor 33.5 Green 12 Others 12
Seat projection on this poll: Liberals would be slightly favoured to just retain majority (approx 13-10-2, with 12-10-3 next most likely)
No seats are projected to fourth parties/independents as no prominent fourth parties/independents are yet known to be running

Aggregate of all recent polling 12-10-3 (no majority)

Another EMRS poll of state voting intentions is out.  Also see the useful trend tracker.  After dipping to its lowest level for many years in the July poll, the Hodgman Liberal government has rebounded to a position similar to that in the May poll.  This is consistent with the July poll result having been contaminated by the Liberals' woeful performance in Tasmania at the federal election, and not too much should be read into the four-point shift in this poll for that reason.

While I previously commented that based on the May poll, the government would have lost its majority, in the present poll they have a helping hand from a crash in the Green vote.  EMRS, a pollster with a very long history of having the Greens a few points high at election time, has a reading of 15% for them, the equal lowest reading in this term.  The pollster last had them below this level in May 2013.  

Adjusting the Green vote for EMRS' history of having it on the high side means that this sample is pointing to the Greens probably being cut back to just two seats in Denison and Franklin, allowing the government to probably just save its crucial third seat in Lyons.  Based on preferences in Lyons tending to favour the major parties, and based on the Greens' lack of a sitting member in the electorate, my reading of the most likely result based on this sample in an election "held now" is the Liberals just clinging to a majority with 13 seats to 10 for Labor and 2 for the Greens.

That said, I am sceptical that the Greens would actually do quite as badly as my "interpretation" figure for this sample of 12%, given that they polled 11.2% in the Senate and 10.2% in the Reps despite significant vote losses to Lisa Singh and Andrew Wilkie (respectively).  Nonetheless the recent polling is not pointing to much if any recovery from their 2013 state vote of 13.8%.  The traffic since the state election is mainly from the Liberals to Labor and the Independent/Others column.

I should also note that I have seen a presently unpublished commissioned poll that is somewhat worse for the Liberals than this one.  So there is still plenty of room for doubt about whether the Liberals are in a majority position or not.  In an election campaign that doubt might evaporate if the party can make the case that only it can win majority government, but with the drag factor of the party also holding government federally I wouldn't count on it.


There is not much to see on the leadership front with Will Hodgman leading Bryan Green as Preferred Premier, 50-25.  Preferred Premier scores tend to skew to incumbent leaders so this is about the sort of margin we'd expect if the government is slightly more popular than the Opposition (as appears to be the case).  About the only vaguely interesting aspect here is that Green has retained his improved result from the previous poll rather than returning to the low 20s or worse.

Polling aggregate

The following is an updated reading from my aggregate of all state polling.  I have used the federal results to rescale the electorate breakdowns (without changing the total votes) but only at a weighting of 20% (because of issues such as Wilkie and Singh drawing votes from the Greens, and a strong anti-Liberal backlash in the Reps compared to the Senate in Bass).  Apart from that I have weighted this EMRS poll at 40%.   The aggregate is now nearly all EMRS data so I am hoping to add some data from another pollster (other than Morgan) soon.

At this stage, although the aggregate has a 10% estimate for "others", an independent winning an unrealistic 100% of the expected Other vote would only be likely to win in Braddon (and even then only if Greens voters liked them).  An independent getting 80% of the Others vote would not win anywhere. However, this is an artefact of the major parties happening to land very close to whole quotas in the projection.  If they did not do this, then high-profile independents would have a chance in some electorates.  The projection is also perhaps kind to the Government in Braddon, where major-league vote-winner Adam Brooks is on the backbench under a cloud and it is unclear if he will contest the next election.

At this stage no high-profile independents are known to be running and therefore no seats are projected to them in this aggregate.  Should any announce they are running, I will be modelling their chances accordingly.


  1. Any thoughts on the upcoming federal redistribution in Tasmania, seems like the current arrangement is pretty balanced, with Lyons and Franklin being 0.5-0.6% above an average seat margin and Bass and Denison being about 0.2-0.3% under. Braddon is the standout at 0.8% under.

    It appears that the easiest changes would be to transfer Richmond and surrounds from Franklin to Lyons. Denison stay pretty much as is (although Collinsville could be transferred into Lyons). Then move Port Sorell or Railton and surrounds from Lyons into Braddon and Exeter from Lyons to Bass.

    Be interested to see what you think.

    1. I don't have any detailed comments on it at the moment, especially not having really had time to look at it because of work and an overseas trip. There are quite a few people on other sites who get excited about possible redistribution boundaries but I'm not really one of them except when a draft proposal is released that is obviously bad, or when there are complications like the SA fairness clause. I'm happy for people to post comments about it though.

  2. Are the Morgan state polls worth looking at, or is the Tasmanian sample size too small (~350) shows better results for Labor but then Morgan usually does. Probably reinforces your 12-10-3 assessment with the next state election clearly hanging on Lyons at this stage (and whether the massive swing there federally is a once of or reaction to the federal state of play or whether there is growing distrust in the Liberals at a state level, i think it's the former).

    1. By the time the most recent Morgan sample is adjusted for the apparent house effects of the series, its results become very similar to my current aggregate. I am going to process it into the aggregate at some token weighting like 5% but haven't done so yet. Only just back from overseas.