Sunday, July 28, 2013

ReachTEL - Tas Federal BaByLon Still Falling?

ReachTEL: Bass 54-46 to Coalition, Braddon 56.8-43.2, Lyons 54.4-45.6

This week's Sunday Examiner contained details of a fresh ReachTEL of the northern Tasmanian federal seats of Bass, Braddon and Lyons.  The sample sizes (Bass 626, Braddon 617, Lyons 659) are extremely large for Tasmanian seat polling and possibly the largest Tasmanian seat polls ever undertaken. A previous large statewide federal ReachTEL taken in early June shortly before the replacement of former Prime Minister Julia Gillard showed Labor being thrashed in all these seats, despite Lyons being held by Labor's Dick Adams on a margin of 12.3%.  In the case of Bass, this repeated a result from the start of the year.  The strength of the Lyons result was quite surprising, and in some circles disbelieved.

The return of Kevin Rudd to the Prime Ministership was expected to create a lift in Labor's fortunes, as it has everywhere else, but there were reasons to suspect this might be muted in Tasmania and that Bass and Braddon at least were still in trouble.  Primarily, the massive swing against Labor at federal level under Gillard (probably running at around 14% considering data from a range of polls) was being boosted by the delicate state of the Tasmanian economy and dissatisfaction with the state Labor-Green government.  The margins shown in the previous ReachTEL were very large, and even taking sample size and the apparently slightly Coalition-leaning nature of ReachTEL federal results into account, it seemed unlikely that a change at the top alone would fix the problems.

An early online taster article provided details of the new ReachTEL as follows:

* That the poll shows Labor "comfortably behind" in all three seats but having obtained "a surge in the polls".  I am not quite sure what is comfortable about being behind, but there you go.

* That the Green vote has dropped "in Bass (13.4 down to 9.4) and Lyons (13 per cent down to 10.2) but held steady in Braddon at relative low of 7.4 per cent."  (The previous Braddon result was 6.9).

* That Kevin Rudd leads Tony Abbott as Preferred Prime Minister "51.2 to 48.8 per cent (Braddon), 50.7 to 49.3 (Lyons) and 50.6 to 49.4 (Bass)". This is claimed to be much closer than the margin elsewhere, but it isn't actually significantly different, since the most recent national reading by that company was 52.6:47.4.  (ReachTEL preferred PM methods, unlike those for other pollsters, do not produce results that innately favour the incumbent.)

The average suggested fall in the Green vote since the last poll is 2.1 points, and it is possibly a bit higher than that if (as was the case in the previous poll) the poll includes "undecided" in its headline rates.  That is off a low base and would point to a state House of Reps vote of only about 9%, but the vote in the Senate would be substantially higher with the Green-friendly side of the Wilkie vote included.

Figures published in the paper edition now show the following:

* In Bass (ALP +6.7), Labor has experienced a 7-point post-Rudd swing to trail 54-46, largely on the back of a primary increase of 10.7% (partly at the expense of the Greens).  Geoff Lyons has a net favourability rating of -4.7, which is actually worse than his January rating of -1.9.  Andrew Nikolic has a net favourability rating of +18.7, also worse than the +24.6 he polled in January (with a 10-point increase in "unfavourable" to 24.6) but still pretty good.

* In Braddon (ALP +7.5), the post-Rudd swing is 5.4 points and Labor trails 56.8-43.2.  The Labor primary is up 8.6 points but this is mainly at the expense of Other and Undecided.  Sid Sidebottom's netsat is -5.7 while Brett Whiteley's is +17.4.  Very few voters (less than 2% each) have not heard of either candidate.

* In Lyons (ALP +12.3), the post-Rudd swing is only 4.6 points and Labor trails 54.4-45.6.  The Labor primary is up about 7 points but almost half of this comes from the Greens.  Dick Adams' netsat is -8.9 while Eric Hutchinson's is +18.6.  15.6% say they have never heard of Hutchinson, even though he was the candidate last time as well.  What rock the 3.2% who have never heard of the incumbent have been hiding under isn't recorded.


As expected the post-Rudd swing back in these seats is substantial.  Indeed on average it is about the same as the national swing back.  However, it has not overturned the huge leads the Liberal Party had in these seats in the previous poll.

On these figures, and assuming the poll has no unknown problems that might render its results invalid, the Liberal Party would be likely to win all these seats if an election was proverbially "held now".  The only one they could be very confident of winning, however, would be Braddon.  I say this because while the Liberal leads in Bass and Lyons are notionally just outside the poll's margin of error, national federal ReachTELs have produced results that have been slightly more favourable to the Coalition than other polls, and therefore it appears likely that their robopolling/scaling methods have a slight house effect in favour of the Coalition, at least for federal polling.  Mark the Ballot currently assesses the difference as being about 1.3 points in the Coalition's favour.  Until there is evidence otherwise, this should be mentally deducted from the Coalition's lead in these three seats, making Lyons and Bass relatively close.

Furthermore the use of relatively heavy scaling in robopolls increases the effective margin of error of the poll somewhat.  So it is plausible on these results that Labor would actually win either Bass or Lyons (I suggest more probably Lyons since it seems unlikely Labor is really quite as far behind there as Bass.)  All the same it is not a good position for Labor to be in, and it suggests that despite the return to Rudd, Tasmania is still a serious problem for federal Labor.   As the party patches up holes in other states, but without showing clear signs of winning a brace of seats anywhere, this is a problem that is only going to get bigger if it cannot be reversed. 

The Examiner's article claims "With Denison tipped to be retained by controversial [sic] independent Andrew Wilkie and Franklin too close to call, Labor could lose all four of its Tasmanian seats in an electoral disaster."  While it is still possible that Labor will be wiped off the map in Tasmania, the problem with this claim is that the assessment of Franklin as "too close to call" was based on a sample taken when Julia Gillard was PM.  It is virtually certain that the return of Kevin Rudd has boosted Labor in Franklin as it has in Bass, Braddon and Lyons, and highly likely on that basis that Labor would retain the seat if an election was held now.

The poll was taken on Thursday night and does not factor in the impact, if any, of the commitments made during the visits by both Kevin Rudd and Tony Abbott over the past few days.

I will consider the Senate picture and update my Senate assessment later today.

Note: State Swing and ReachTEL-Morgan differences:  This poll points to an average 14.4% swing in the three sampled electorates, or about 13% accounting for apparent ReachTEL house effects.  We know from the previous poll that the swing against Labor is probably much worse in the north of the state than in the south, so this ReachTEL might be taken as pointing to a state two-party preferred swing around 9.6 points to the Coalition, for a two-party preferred result around 50.6:49.4% to ALP.  Yes, they really could be ahead on 2PP but only win one of five seats.

In the Not-A-Poll article I mentioned that previous ReachTEL post-Rudd samples have shown a slightly better picture, but those samples were very small compared to this one.  Morgan, meanwhile, continues to show a strong two-party-preferred picture compared to this ReachTEL, with a post-Rudd average now running at about 57.6% 2PP off a sample now up to 676 voters. 

That difference justifies some caution about any Tasmanian findings, since two pollsters are producing radically different results and one of them at least must be getting it wrong.  Oddly enough, pending substantial evidence from a third pollster, it is ReachTEL that has the established position through having been tested at elections and performed passably on the whole and sometimes very well (for instance, see house effects discussion here), while Morgan Multi-Mode is a frustratingly opaque polling method that has not yet been tested at an election and that may be prone to regional glitches.


The weekly federal Poll Roundup and Seat Betting Watch will probably be published on Tuesday during the daytime, but the Galaxy just out was rather good news for Labor with a 50-50 2PP and a 40-38 lead on the inevitable asylum seeker question.  This may seem to be a big turnaround from Newspoll last week (Coalition leading 33-26 on the same question) but it probably isn't, since the questions had a different range of answer options.  Greens supporters preferring Labor's "solution" to the Coalition's probably chose "someone else" in Newspoll but they had no such option in Galaxy.


  1. Kevin what do you think about the low Green and undecided? In my recent experience, I would think that in Bass, for example, there is normally a rusted on ( if you can call a Green voter that) Green vote of about 14% plus...

    1. I don't think the undecided vote (average of 5% across the three seats) is all that low. It's quite similar to the rate Newspoll currently upfront-excludes. I also think that someone who has no view who gets a robopoll dealing solely with politics will most likely hang up. Observers of Tasmanian polls are used to the very inflated undecided rates of EMRS but they are not typical from other pollsters.

      The Greens do tend to poll more poorly in the House of Reps than at state level or in the Senate; the myth of a wasted vote may have something to do with that. However, the Green vote in this poll overall does seem terribly low, with an average swing away of 7% from the 2010 results, or a loss of 43% of their 2010 vote base.

      I've never been convinced that as much as three-quarters of the Green vote was really locked in. At least a third of their stronger results, in my experience, is an extremely soft neither-of-the-above response to the two major parties, and that third gets poached or becomes disillusioned with the party very easily. Even so the figures in this poll seem very low, but when I looked at what other state breakdowns post-Rudd I could find, I didn't find anything all that different. So maybe it just really is that bad, or somewhere near it.

      Through the 90s the Greens results at state level were very modest, at times only just in double figures. It's been tempting to think from the last few state elections that they've improved above that level for good, and some people push a theory that the Green vote is tied to a certain demographic that means that it must (over time and with some ups and downs) tend to increase. With a lot of the polling that is coming out now I am not sure about either of those things.

    2. Thanks for that, mind you it works both ways, there were those within the anti mill camp who were convinced in 2007 that the Green vote would be much higher...


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.