Saturday, March 8, 2014

ReachTEL: The Campaign Has Changed Nothing

ReachTEL: Liberal 47.4 Labor 23.6 Green 18.2 PUP 6.7 Other 4.1
Outcome if election was held now based on this poll: Liberal Majority (14-5-5-1 based on sum of individual breakdowns, though Labor would more likely get one more seat somewhere)
New aggregate of all state polls: Liberal 14 Labor 6 Green 4 PUP 1
My current forecast: Liberal 14 Labor 6 Green 4 PUP 1 

In the three weeks since the last public opinion polling most of the formal 2014 state election campaign has gone by.  People often expect events that happen during campaigns (such as campaign incidents or policy announcements) to affect the outcome, but the great majority don't have much impact.  With the release of the most recent ReachTEL, taken just nine days from polling day and with prepoll voting already open, it doesn't look like this campaign period has offered any respite for a Labor government on the verge of being put out of its misery.  Indeed, this poll is if anything a shade worse for the government than its predecessor (see ReachTEL: The PUP Surge Has Landed), which was also at the low end of Labor's recent form.  Not so much because of the loss of a point of support (an insignificant difference), but because the distribution of votes between seats is even nastier.

Here is the polled distribution:

 If these figures were repeated at the election then Bass and Lyons would be straight 3-1-1s. Franklin would also be 3-1-1, although the Liberals' lead over Labor (21.6 points) is not that far above what they need in that electorate (probably over 18 points to counter the impacts of greater leakage.)

In Braddon, on this sample, the Liberals easily get three, Labor gets one, and the last is between the Greens, PUP and Labor.  It might seem odd to say that Labor with 1.38 quotas is in the mix for two against the Greens with 0.63 quotas and PUP with 0.58 but Hare-Clark is about candidates not just parties.  If after the exclusion of minor Labor candidates, Brenton Best and Bryan Green were on very similar totals, they could both stay ahead of the lead Green and PUP candidates.  I add the usual note that the Greens are likely to slightly underperform this polling.  Also it may turn out that PUP have leakage problems as their lesser candidates are excluded.  On the poll sample numbers I prefer Morgan's (PUP) chances but there isn't much in it.

Denison is the most interesting one again, because while the high PUP vote in the previous poll has not been repeated in this one, the Greens are ahead of Labor.  Even accounting for the likely overestimation of the Green vote, and even assuming that that is more of a factor where the Green vote is high, a poll sample with the Greens 2.7 points up on Labor in Denison is interesting given that the previous sample also showed the Greens to be competitive.  On this breakdown Bill Harvey (who has known form as a good preference-sponger by Green standards) would be slightly more likely than not to win, but the prospect of two Greens in Denison has reared its head in electorate samples at past election without actually happening.

This poll if completely accurate - or even accounting for likely minor house effects in the Green and PUP votes - could thus produce an outcome like 14-5-5-1, although Labor is close to the support level for a second seat in three electorates and with minor variations would get at least one of those somewhere.  The first ReachTEL was the first poll in which the scenario of the Labor vote completely melting while the Greens escaped with moderate votes and no seat damage at all reared its head.  This now is another, only even more in that direction than the first.  I can't stress enough the Greens' history of underperforming their polling but it seems Labor's decision to throw them out of cabinet has only given them oxygen, especially in Denison. 

The distribution of support between seats in this poll is disastrous for Labor (not that 23.6% will give you much joy anywhere anyway) but ideal for the Liberals.  Running well behind their state average in Denison and spreading the rest quite evenly is exactly what they want.

Labor's hope here of escaping the most serious damage is that despite the significant methods change I mentioned in the previous article (removal of the undecided option for voting intention) there is still something systematically wrong with ReachTEL polling in Tasmania.     But given the points I made in the previous article about changes relative to EMRS, I would not be too optimistic on that score just yet.  The final EMRS and Newspoll will shed light on whether Labor has any hope of staying in office at all; the present poll provides none.

The Preferred Premier results are also worth a look:

This is the same statewide pattern as last time in that all the leaders predictably outscore their parties, but proportionally Giddings does so by the least.  The electorate pattern in this one is different to the previous ones though, with Giddings running slightly behind her party in Franklin, doing nothing spectacular in Denison, and performing best relative to it in Bass. 

I am hoping we will see some individual candidate polling soon because, vague as it is, it might at least provide evidence on whether the Premier is at risk in her own seat.  The Giddings-as-candidate campaign has been nowhere near as vigorous or as visible as that for David O' Byrne, and the argument for keeping O'Byrne is that he is the likely future leader.  No Tasmanian Premier has lost their own seat since the advent of Hare-Clark, and the Premiership is a huge profile asset in an election, but it would be interesting to get data to confirm that Giddings is safe.

Some might think Giddings' reported win of the People's Forum debate  might boost her and her party's chances.  However, Hodgman made no reported blunders, few people watched the full debate live, and it is not news that Giddings is a strong performer in debate formats.

There are five other poll questions:

* best party to reduce unemployment
* support for tourism development in national parks
* support for extending high schools to year 11 and 12
* best solution to budget
* best solution to hospital waiting lists

Here's the first; I'll save the rest in case they're not reported on today:

No surprises here. The Greens tend to get huge support as the best party to handle their strength issues and to lag their vote otherwise.  Labor also lags its vote a little (because of the Undecided option) but the Liberals do not.

On to the new aggregate:

 I've weighted the new poll at 40% but this will be reduced when the EMRS data are also available as the aggregate is currently extremely ReachTEL-heavy.  The two strong polls for the Greens in Denison now have them well and truly in the mix for two but it's possible I should be deducting more than the usual point there, and it still falls in the "extraordinary claims require extraordinary evidence" box for now.  The aggregate is now showing the Liberals with enough of a lead to probably get three in Franklin while the last Braddon seat remains a mess.  In my view it is still at 14-6-4-1 as the most likely outcome.  I have also switched my own forecast to this but Franklin remains very close.

Update: Double Trouble: Quite oddly The Examiner has also reported a new ReachTEL with a smaller sample size.  Figures are nearly identical: Lib 47.1 ALP 23.6 Green 18.2. The ALP primary in Franklin is reported slightly higher on 26.9, the Braddon PUP primary at 9.7, Hodgman's preferred premier score at 54.4.  The figures are so close to identical as to make me wonder whether the data sets for the two polls are completely distinct or not.  I will have more comments on the Examiner poll later today - have another commitment on and off this weekend.

The Examiner's question list clearly contains some items different to the Mercury's:

"The polling, conducted on Thursday night, also showed more than half of Tasmanians rated the current government as poor or very poor.However, almost a quarter were satisfied with its performance and 21 per cent thought it had done a very good or good job. More than 63 per cent of respondents wanted a majority government while almost one in five was undecided on the issue. Liberal leader Will Hodgman was the preferred premier of more than 54 per cent."

If you believe Crikey's peculiar fascination with making lists of things, Will Hodgman's chief of staff Bradley Stansfield is Tasmania's most powerful non-state-politician!  Whatever, the case, this tweet on the change in Bass samples between the mid-Feb poll and now (don't be fooled by the trendline, it's just two data points) is well worth a look.   There is no evidence Labor's games on the imaginary pulp mill have brought it any joy at all and they may have just shovelled votes to Kim Booth.

In other news, Nick McKim has declared the Liberals will win as the Greens ramp up the push for enough seats for opposition status.  The Greens have been increasingly brazen and confident in their campaigning lately, even trying to push for small-l liberal votes in Denison.  (Their argument is generally sneaky and extremely contestable but that might be a story for another article!)

If the current poll is to be believed, the vultures are descending on Labor from all directions, nothing the party is attempting to gain votes is working and the question now is whether this can still get even worse.

Update 9 March: I've now examined the full figures from the Examiner's voting preference and better premier results (as published in the Saturday print edition) and they're consistent with my previous view that the Examiner and Mercury samples probably aren't independent. Not one of the 40 individual electorate breakdowns differs from another by more than a point with 31 of 40 differing by 0.4 points or less. I've not modelled the odds against this occurring by chance with independent samples but I'm confident they're long.

The Examiner not surprisingly finds voters wanting majority government (without asking whose). The net margin is 63.3-17.9 and it's only close-ish in Denison (45.7-34.5). Opposition to majority government is about equal to the Green vote in Braddon and Lyons, about 4.5 points below it in Bass and Franklin and about seven points ahead of it in Denison.

The Examiner asks voters to rate the success of the ALP-Green government over the past four years with only 21.4% finding it good or very good and 55.6% finding it poor or very poor.  It's my habit with ReachTEL findings that include a "satisfactory" option to treat half of such responses as mildly positive and the other half as neutral but even this leaves the government with a net rating of -22.6.  In Denison this method raises the governments net rating to the lofty summit of exactly zero (ie neutral) but the government is viewed very dimly everywhere else: Franklin -19, Lyons -22.6, Bass -27 and Braddon -45 (or a whopping -55.6 without the correction)

The Examiner also publishes pulp mill results which I have updated on Polling and the Proposed Pulp Mill.

The Mercury has published results showing strong support for unspecified "tourism development in our national parks" (62.2-17.5 with little variation across electorates.)



  1. Its interesting to see a slight increase in the Green vote in recent polls. It may be because the major parties and PUP have made great efforts in appearing as pro development, anti Green as possible. I also believe that people who move to Tasmania for lifestyle reasons tend to be attracted to the Greens. It may be also that the Greens are now by themselves on the left and votes continue to bleed from the ALP. Kevin, where do you believe PUP voters have come from as far as their previous voting patterns are concerned?

    1. I don't think PUP have been consistently "anti-green" - although they've been gung-ho about "lockups" they've sent out extremely mixed messages on the pulp mill for example. My strong suspicion is that disaffected Labor voters unwilling to vote Liberal are the biggest source of the PUP vote.

  2. So who becomes the official Opposition if it is tied 5-5? Surely there is only room for one; considering things such as parliamentary procedures, ceremonial duties, allowances staff wages and privileges, guaranteed placement in pre election debates (!) and probably a raft of other things I haven't thought of or that are unknown to me.....Who gets to make the call and under what criteria would they use?

    1. In the 2002-2006 parliament (14-7-4) the Liberals and Greens were granted a sort of co-opposition status for resource purposes by Labor under Jim Bacon, and Peg Putt was sometimes called the Leader of the Greens Opposition. Something similar could be applied again. Labor did not repeat this arrangement under Lennon after 2006. (I think the decision was taken when it looked like the Libs might win only six seats.) To the extent that it's not covered by legislation it would be the government's call. There might be precedents in Queensland.

      Election debate participation in Tasmania isn't determined by being the official opposition or not as there is no regulatory body overseeing debates.

  3. I hung-up on the ReachTEL robot and I'm wondering how many other people did. I have no evidence ReachTEL ring mobile phones although they claim they do. Potentially, there is a 20% distortion in their results as more conservative voters would tend to have landlines and answer questions asked by a robot.
    The Examiner and Mercury have accepted the polls at face value and have extrapolated results accordingly.
    On the face of it a Hodgman government seems likely and Dr Bonhams' analysis is excellent. Interestingly, the election coincides with a full moon. Of course this does not change the outcome but it could change peoples responses to it.

    1. I'm not sure why conservative voters would be more likely to answer automated questions. It might for instance be argued that young voters would be more used to automated voice interaction. Calling or not calling mobiles is not an issue if your scaling is accurate as proven by Newspoll last election. Still, this election is a big test for how well robopolling works in Tasmania. It had some issues at the federal election (despite tipping all five seats correctly) but improvements seem to have been made.

  4. The similarity of The Mercury and The Examiner polls is bizarre. Does that mean the Examiner poll was a subset of the Mercury poll? Did the Mercury pay more for a larger poll? I must admit I'm surprised at the size of the Green vote in Bass in this poll. Maybe it has to do with Kim Booth's distancing himself from the Labor/Green partnership.

    1. It looks like a subset but there might be some other explanation. I'd be interested to hear from anyone who was polled the Examiner's pulp mill questions to see if they also got the Mercury's schools and parks questions.

  5. How do you figure that one out David? The Greens have ( as you would expect ) presented a united front since the election was called and Mr Booth hasn't gone off script, and this small window in time is the only time we have observed a poll rise for the party in Bass. Besides, whilst I'm sure Mr Booths recalcitrant ways over the last 4 years were a genuine expression of his individuality, I am sure there was a broader strategy behind the scenes within the Green camp...a strategy for one voice to maintain the green 'rage' in order to placate the die hards, whilst sensible and dedicated MPs such as Mckim and O'Connor went to business in the tougher and more productive roles as ministers, conscious of consensus building, getting results and building up trust of the Greens in the broader community. The latter is very important!

    This hero worship of Kim Booth as the "real deal" is on par with some in the Labor movements adoration of Brenton Best as the "real deal". Booth at least has a bit of style but they are both serving a purpose for their respective parties , which seems not to be understood by some. But then again, it's really not meant to be!

  6. Kevin do you think the electorate is more polarised than usual in this election? Can you see the level of people just voting 1 to 5 being greater than last time around. Would the level of exhaust favour any particular party in the final result?

    1. I think the larger number of candidates is likely to slightly to modestly increase the exhaust rate in cases where a party's last candidate is excluded. This might slightly advantage the Liberals in Franklin if Green preferences exhaust at a higher rate than normal when Green #2 is cut. I expect a high exhaust rate of PUP votes.


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.