Sunday, July 23, 2017

Reachtel: It's All About Lyons

Mercury ReachTEL Lib 43 ALP 32.9 Green 13.4 Other 10.7 (after redistributing "undecided")
Interpretation Lib 43 ALP 36.7 Green 10.7 Other 9.8
Most likely result right now based on this poll would be hung parliament (12-10-3) closely followed by narrow Liberal majority (13-10-2)
New aggregate of all polling: Liberal majority (13-10-2) with hung parliament (12-10-3) next most likely.

A Mercury ReachTEL of state voting intention is now out with a sample size of a whopping 2817 voters.  My initial comments on it will be very brief because I am playing in a chess tournament this weekend and also so that the Mercury get good commercial value for their polling data, which I expect can be found in full in the Sunday Tasmanian.  More detailed comments may be posted on Sunday night.  There was also a commissioned poll of Lyons this week - see Fishy Prospects In The Seat Of Lyons.

This new poll again presents a story that I have repeated so many times in state polling coverage over the last two years that presumably something entirely different will happen and it will all be wrong!  The overall picture of polling for some time has shown the Hodgman Government's majority hanging by a thread, given the virtually certain loss of a seat in Braddon and the likely loss of another in Franklin.  With the Greens struggling to hold their seat in Bass, the key question then is whether the Greens (or somebody) can knock off one of the three Liberal MPs in Lyons.  If that happens the majority goes, and it could be that the government goes with it.  There are a number of possible fourth-party/independent wildcards, but at this stage none of them are known to have their acts together.

Polling is consistently showing the Greens are ahead of the Liberals on raw quotas in the race for the last seat in Lyons, but that they are short of a quota themselves. On the raw numbers in this poll the Greens would win in Lyons anyway, but ReachTEL polling in Tasmania has typically had the Greens too high and Labor too low.  If this is the case again, the Green vote drops to the point where it is possible for three Liberals to beat them if the votes for the three main Liberal contenders are evenly spread.  As a typical example, suppose the Greens have 0.7 quotas and the Liberals have 2.4.  It seems the Greens would win easily, but if this Liberal vote ends up spread between three Liberals with, say, 0.85, 0.8 and 0.75 quotas each (after the preferences of their minor candidates), then the three Liberals win.  This is because Hare-Clark is about candidates not parties.

In this particular poll, the Liberals would really have to be lucky to get an even enough spread.  In theory given the house effect of past ReachTELs in the state, Labor might even outpoll the Liberals in Lyons based on this sample, but Labor is not so well placed to take advantage.  As leader and the only sitting Labor MP in Lyons, Rebecca White may well get a quota, meaning what's left over splits between two candidates rather than three.  If she has well over a quota, surplus votes will leak, and it's unlikely the rest of Labor's Lyons team will poll massive votes in their own right (meaning more leakage as they pull up towards quota.)

This poll has a high Ind/Other vote in Bass, which may be a result of small sample size, but much of the seat did recently have a Legislative Council election where independents ran first and second and the parties that bothered contesting were crushed.  Hence there are rumours that Neroli Ellis, who ran a close second for the LegCo seat, might want a go at this one too.  In general, any high-profile and well-funded fourth party run will be bad news for the government.

I'll post a new state aggregate soon but I expect little will have changed and it will still be hovering between 12-10-3 and 13-10-2.


This poll shows Rebecca White as preferred premier against Will Hodgman, 50.6-49.4, with big leads in Denison, Franklin and Lyons.  ReachTEL polling generally does not favour incumbents on this question in the way that other polls do, but even so this finding suggests the change to White remains rather well received.  There is a common beltway view in Tasmanian politics that Will Hodgman is super-popular, but I am not aware of any empirical evidence whatsoever that supports it.  The fact that he routinely belted Bryan Green as preferred premier in EMRS polling proves nothing in this regard.  There has been no actual approval rate polling.  Certainly Hodgman does not appear to have many enemies who are not just implacably anti-Liberal, but the question is to what extent dissatisfied views of his government may have dragged his personal standings down.  I get tired of saying it but these kinds of "beauty contest" preferred premier polls can only compare two leaders and do not tell us whether both are popular, both are unpopular or something else.

ReachTEL preferred premier figures should not be compared with those of EMRS because EMRS includes an undecided option, which can favour the incumbent.  The best comparison is with the statewide ReachTEL last November, where Will Hodgman led Bryan Green 59.8-40.2, so there has been a 10.4 point swing since that poll.  That comfortably exceeds the modest swing in voting intentions of 2.6 points away from the Liberals and 2 points to Labor.  I don't have the breakdowns from the November poll to hand, but from memory the main thing that is going on here is that Labor supporters are much more solidly behind White than Green.

I am a little surprised that the Premier has been referred to in this poll as "William Hodgman".  He is pretty much universally known as Will.

Detailed Breakdown Estimates

Breakdowns with "undecided" included were published in the Sunday Tasmanian.  The full poll report includes a statewide breakdown of the 6.4% of "undecided" votes (Liberal 35 Labor 26.1 Green 8.9 Ind/Other 30) but there is not an electorate by electorate breakdown, making scaling the specific samples a bit tricky.  I have done this in such a way as to make the "undecided" votes lean more strongly to the Liberals in areas where the Liberals are strongest (etc).  On this basis this is my estimate of how the poll looks on a seat by seat basis with the "undecided" passed to the party they are leaning to, and with no adjustments for house effects:

Not too much should be read into the seat-specific samples because they are not that large and seat-polling is a risky business at the best of times, but the result is probably fairly representative.  In this case the seats are all straightforward, save that perhaps if someone was able to scoop nearly all the Other vote in Bass they might dislodge the third Liberal there.

Now here is what it looks like with adjustments for apparent house effects from past elections:

In this case the final Lyons seat is a mess.  With no party anywhere near the fifth quota, much would depend on the candidate totals within each party and the distribution of preferences from minor candidates.  In practice, 11.4% worth of minor candidates may well not exist, making it pretty much guesswork as to where those votes would go.

New Aggregate

In adding this poll to my aggregate I've decided not to focus too heavily on the electorate breakdowns, because the sample sizes by electorate are not that large and there is obviously a lot of noise in the specific breakdowns.  I've decided to aggregate this poll at 30% for the total and 15% for the individual electorate breakdowns, with the previous aggregate carrying 55% of the total weight.

In this aggregate all the seats are straightforward except Lyons, though again this might change if a fourth party runs prominently for one of them.  In the case of Lyons I believe the three Liberals with 2.48 quotas would probably beat the lone Green with 0.68.  Therefore, I am altering my sidebar aggregate to give the Liberals the third seat back in Lyons.

It is possible that the assumptions I am making about house effects will prove to be wrong.  ReachTEL may have changed their weightings in Tasmania in response to past errors and not told us about this.  In general, Australian polling is very opaque in this regard.  Perhaps the Greens will actually poll the 13.4% ReachTEL are giving them, or more.  But the past history of Tasmanian polls overpolling the Green vote is a very strong one and one that no polls have previously made any effective attempt to fix.


There is also a question on handling of the economy, which finds the Liberals rated best by 50.3%, Labor by 38.7, the Greens by 11.0.  The Liberals lead in every electorate.  Given the history of polls of this sort favouring conservative parties, this is actually not a very strong result, though the inclusion of the Greens in the mix does mess up comparisons somewhat.

More issues questions to come; updates may be added through the week.\

Cable Car

Updates on the proposed kunanyi/Mt Wellington cable car poll have been added to a previous article on cable car related polling.

Issues Polling

A question about the most important issues in the state sees health (36.4%) narrowly leading jobs and the economy (32.8%) and education (14%), with energy security (6.9), forestry (3.7), same-sex marriage (3.3) and fish farms (2.9) barely troubling the scorers even in the more pro-Green electorates.  What is interesting here is that Bass and Braddon have jobs and the economy well ahead of health (38.6-33.1 and 40.3-31.2) but the others (Denison 27.2-40.7, Franklin 28.4-37.3 and Lyons 29.8-39.6) all have it trailing.  If the Lyons result holds up this could suggest the electorate is breaking away from the profile of its northern neighbours (they tended to move in lock step around 2013-4 when all were swinging to the Liberal Party) and that could be a big concern for the government.  But it could also be that this sample - which was unusually weak for the Liberals in Lyons - was just atypical.


  1. Just wanted to note, Labor holds two seats in Lyons, David Llewellyn and Bec White.

    1. True but Llewellyn is retiring so White will be the only sitting ALP MP in Lyons at the election.

  2. Hi Kevin.
    What effect would a Wilkie backed independent have in the seat of Denison?

    1. Well firstly it has to be a Wilkie-backed independent who people have actually heard of, so the choice is probably Kristie Johnston, Kristie Johnston or Kristie Johnston. Assuming such a candidate runs and has a more coherent rationale for people to vote for them than "Andrew Wilkie says so", they will probably pull votes from all parties. However the problem is that polling keeps showing the majors with their two quotas each and the Greens usually with their one, so it's challenging for someone else to knock any of the big three much below their quota and hence have any sort of chance.

  3. Hi again Kevin - fully acknowledge that I am clutching at straws to find a bit of excitement outside Lyons but if Neroli Ellis was to run in as an independent (6000 votes in recent Leg co election) and outpoll the Greens could she potentially snag the fifth seat on the back of green and Labour preferences?

    1. It's plausible Ellis could win as an independent but she would not be getting major party preferences to do so, on current numbers. She would probably need to get over the Greens and then beat the third Liberal. Preference flows in Hare-Clark are not all that strong because of exhaust and the lack of how-to-vote cards.


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.