Saturday, June 25, 2016

ReachTEL Says Lyons Going, North In Doubt

Note: National poll updates are continuing in the rolling poll roundup below.

ReachTEL: Bass and Braddon 50-50, Lyons 55-45 to Labor, Franklin 59-41 to Labor, Denison 65-35 Wilkie vs Labor
Interpretation: Bass 51-49 Liberal, Braddon 51-49 Labor, Lyons 54-46 Labor, Franklin 58-42, Denison see below
(Poll taken before Brexit and Launceston university funding announcement)

The Mercury has released ReachTEL polling of the five Tasmanian federal electorates.  For my general background to them see The Five Tasmanian House Of Representatives Seats and for a previous ReachTEL from mid-May see ReachTEL Points To Tasmanian Status Quo.  It isn't pointing that way any longer.  There is also some Senate-related polling coming that I will cover in an update to this piece.

The previous Mercury poll had all three Liberal incumbents (Andrew Nikolic in Bass, Brett Whiteley in Braddon and Eric Hutchinson in Lyons) in fairly comfortable positions.  Although their two-party preferred votes were only 51% in two cases and 53% in the third, these were based on a strong flow of respondent-allocated preferences and in reality the leads were greater.

Since then parts of Bass, Braddon and Lyons have been flooded (although this had no obvious political implications beyond a distraction from the campaign and an attempt by Nikolic allies to bestow credit on their man for Launceston escaping the worst of it).  The issues mix also turned to health via Medicare in the last week, which suited Labor fine with health services the biggest issue of the 2016 campaign for voters in the northern seats.  Hospital funding is a major issue in urban parts of Bass and Braddon and remote health services a perennial issue in Lyons.

The poll overall suggests a 3.7% swing to Labor in the four Liberal-vs-ALP electorates.  This is higher than the current national swing.  Until now the trend in Tasmanian-sample polling has mostly been one of no swing or a muted swing.  However this trend has often been based on small and unreliable samples.  This new sample is very large so we shouldn't place too much weight now on what we had before.

Poll design

The poll is another ReachTEL robopoll with about 550 voters per seat.  Voters were asked questions on primary voting intention, their ratings of the local member and important issues in their seat.

The previous poll muddied the waters by including the Lambie Network, who are not fielding lower house candidates.  This poll includes only those candidates who are standing, all of whom are named and presented in ballot order.

Preferences are as stated by the respondents.  The use of respondent versus last-election preferences has been a hot potato as usual in polling for this election campaign, and for the May poll I preferred to calculate the latter.  However there is now a very good reason not to use last-election preferences for Tasmania, and that is that we know the micro-party candidate mix, and know that it is more left-wing than last time.  The Recreational Fishers (a left-wing outfit led by union figure Kevin Harkins who was twice spectacularly stymied in his bids to run for Labor), the Arts Party and the Renewable Energy Party will all send preferences to Labor.  Together they account for half of Tasmania's micro-party Reps candidates, and the Recreational Fishers (who have TV ads) are polling well.  I have also checked the preference flows in this poll and none of them seem problematic.

Given that ReachTEL supply data on primary votes to one decimal place, I find it useful to calculate a non-rounded 2PP to one decimal place from the data.  After all, if the poll is 50-50 it is nice to know if that is really 50.4 one way or the other.  However, when I do this I often end up with a slightly different 2PP result even after rounding, and I am not sure why this is.  I include my own 2PP estimates from the poll below.

Primary votes

As usual I have converted the poll into figures that resemble those published by other pollsters, with undecided voters reallocated.  A little later on I will paste the full primary figures in for data hounds who may want them.

Some comments here:

Firstly I am sceptical of the size of the votes for some of the micro-parties, once undecided voters are included or even before.  The Recreational Fishers have been advertising heavily and might be confused for more right-wing user-group parties (like the Shooters, Fishers and Farmers) by unwary voters, but some of the other parties have very little profile in Tasmania and are running obscure candidates.  So whether voters will actually vote for them in the numbers shown remains to be seen.  This is especially so as the poll suggests the Green vote is rebuilding well (something we should also be a bit cautious about since the Greens are frequently too high in polling).

The size of the Recreational Fishers vote is interesting as while they probably would not do as well in the Senate (where there is competition from Lambie as well as many other micro-parties) it does suggest that if there is a seat left over for a micro-party they might poll the 3% or so needed to be in the mix for it.  That said I am not convinced there will be.

The undecided rate is high in Lyons (13.9%) and Franklin (12.2%) but in Lyons there were only about two undecided voters in the whole sample who said they were leaning to Eric Hutchinson.  That at least suggests room to pick up votes from undecided voters who preferred other parties.

The poll has published 2PPs of 50-50 for Bass and Braddon here, but I get Bass at 50.8 to Andrew Nikolic (LIB), and Braddon at 51.2 to Justine Keay (ALP).  So on my reading if this poll is correct, only Nikolic will hold.  However the margins in Bass and Braddon are so close that the Liberals could plausibly hold or lose both.  The university announcement may help them, or not.

The implied result in Franklin is not surprising as Liberal candidate Amanda-Sue Markham has had an even worse campaign than Julie Collins' previous opponent.

The Mercury have not had much luck with Denison.  In May they asked for a Wilkie vs Liberal 2PP but produced figures on which the 2PP would probably have been Wilkie vs Labor, and this time it's the other way around.  That said I doubt the Liberals are this far ahead of Labor on primaries (if at all) and we do at least have useful data to suggest the flow of preferences to Wilkie has been unchanged from the last election.  Both the Greens (who preferenced Labor last election) and Labor (who initially put Wilkie last, then raised him to above Rise Up Australia only) have seen the error of their ways, with the Greens issuing an open ticket and Labor preferencing Wilkie over the Liberals.

Candidate ratings are also supplied.  Here's a summary:

The negative ratings for Andrew Nikolic and Brett Whiteley are similar, but respondents are much less likely to rate Whiteley "very good" and more likely to rate him "average".  Julie Collins remains popular and Andrew Wilkie's figures would be matched by few sitting MPs in the nation.  All the Liberal incumbents have fallen since the May poll (Nikolic by 13.6 points, Hutchinson by 4.6, Whiteley by 2.9) and while this is probably partly caused by worsening voting intention results, there is a suggestion here that the personal campaigns against Nikolic have rubbed off to some degree.

There is much more I could include but time is short and I am not sure what the Mercury is covering today so I will leave it there for now.  Assessments for the Tasmanian seats will be updated with this polling later today.  ReachTEL polls correctly picked the winners of all Tasmanian seats in 2013, but the margins were a fair way out (in favour of the Coalition that time), so it shouldn't be assumed these samples are the final word.  Nonetheless the margin for Lyons isn't trivial and unless something else comes along I now expect the seat to fall, and my national seat model agrees, dropping the Coalition's probability of holding it to 25% (conditional on the national 2PP not changing), having not been significantly below 50% before now.

More to add in coming days and more detail in the print edition of the Mercury today.


There are some figures reported in the Sunday Tasmanian of Senate polling in which voters were asked whether they were more or less likely to vote for various Senators (Jacqui Lambie, Richard Colbeck, Lisa Singh.)

The table below shows the more likely - less likely breakdown for each Senator by electorate.  It also shows the average more likely score by party voting intention across the five electorates for each Senator.  Note that this average is not a state total, although it should be close to the state totals in most cases.

This is an extremely difficult polling format to draw useful conclusions from.  Many voters when asked if they are more likely or less likely to vote some way based on something will say they are more likely if they just like that thing, or less likely if they don't.  Many voters who say they are more likely to vote for Singh or Colbeck will end up simply voting 1 for Labor or Liberal in that party's box.  A more useful format would be to ask voters their Senate voting intention by party, then for those saying Labor or Liberal ask if they intended voting above the line or below the line, then for those saying below the line offering them a list of candidates for their party.

Some observations, however:

* The Lisa Singh push is mainly a Hobart thing and especially a Denison thing.  She was former state member for Denison and her left-wing views appeal to the Denison electorate.

* Jacqui Lambie has very high "more likely" ratings among Labor voters.  This suggests a significant, perhaps very significant, number of Labor voters in the House of Reps will vote for Lambie in the Senate.  This will make it really difficult for Labor to win five.

* Jacqui Lambie has high ratings in Franklin as a consequence of Franklin having lots of Labor voters.

* Richard Colbeck has nothing like the cross-party appeal of Singh and Lambie.  His fate is most likely to depend on how many Liberal voters decide to vote below the line for him.

* Lisa Singh has strong support from Greens voters.  They will probably vote below the line in large numbers and preference her strongly.  However the Greens are unlikely to have much influence on Singh's fate, because they will probably either fall short of two quotas or clear two quotas by not much.  In the first case those preferences will not reach Singh and in the second they will not be worth a lot.


  1. "I am not sure why this is" Love your work. Pat Caplice

    1. Ta. I just added the words "even after rounding" to make that bit slightly clearer.

  2. Be interesting to see if Libs do get ahead in Denison, not that it changes the outcome much, on those numbers Wilkie will win. Labor may get punished in Denison over the Lisa Signh issue, but then again I would expect most of the Pro Lisa Singh voters to be Wilkie backers based on ideology.