Thursday, March 1, 2018

Tasmania 2018: Commissioned Pokies ReachTEL

On 14 Feb 2018 the Australia Institute Tasmania conducted a ReachTEL poll for the 2018 state election (see guide), some questions from which have been on public display for some time.  Some results on company tax cuts and donations disclosure were published some time ago and a further tranche on poker machines was published a few days back.  As well as this I have obtained (with thanks to the reader who passed it on) the primary voting figures from the poll, which was taken on the night nominations for the election was announced.

Voting Intention

The voting intention figures are old rope now but may be of some interest in terms of discussions of how the campaign has unfolded. Voting intentions were Liberal 41.7 Labor 30.5 Green 11.0 JLN 4.6 Other 8.0 Undecided 4.2.  What would be a staggering 68% of undecided voters said they were leaning to the Liberals, except that the effective number of undecided voters after scaling would have been just 39, giving that 68% figure a margin of error of at least 15% (even more after scaling).  After redistributing the undecided the results were Liberal 44.6 Labor 31.1 Green 11.1 JLN 4.8 Ind/Other 8.4.

The main difference between that poll and the Mercury ReachTEL of 22 Feb is that the raw Ind/Other vote crashed from 8.4% to 2.0%.  The most likely reason for this is that voters who might have been looking for a Wilkie-type independent, or for One Nation or a religious-right party, discovered that none of those were running.  Aside from the Shooters, Fishers and Farmers the only really notable "Ind/Other" running is Brenton Best in Braddon.

With that taken into account the differences between the two polls seems very minor - one could just assume these intending Ind/Others voters had gone to other parties more or less in proportion.  Except that since the TAI poll found a very large 82% of all minor party voters said they would preference Labor over the Liberals in a forced choice between the two, so it might be inferred these Ind/Others were mostly Labor-leaning.  On this basis the Mercury ReachTEL is slightly better for the Liberals than the Feb 14 TAI poll, but the difference is not statistically significant as proof of improvement through the campaign.  (An aside: forced-choice preference polling for Hare-Clark is silly when 35-40% of Greens voters and even higher numbers for other parties will exhaust their vote rather than preferencing anyone. In any case, most Greens preferences will never be distributed.)

I already had an vague idea of what the primaries for this poll could look like because I had had a go at reverse-engineering the primaries and found that the issues breakdowns worked best when the raw Liberal lead over Labor was 11-12 points.  (They also worked best with the Lambie Network vote on about 2% and the Greens on about 15% but this sort of thing rarely works perfectly!  That said it is more likely to work with a ReachTEL because they publish data to one decimal.)


Although not an issue rated as "most important" by nearly as many voters as "jobs and the economy" and health, poker machines have become a defining issue of the campaign because of the extent of external campaign spending (and the unknown but apparently large extent of campaign donation) over the issue.  Vested interests threatened by Labor's/the Greens'/JLN's policy to remove poker machines from pubs and clubs are making a massive attempt to convince voters to vote Liberal, while other interests supporting the policy are working the other way on a smaller scale.  The pro-pokies campaign seems so far to be resonating with working-class voters and in communities concerned about possible job losses, in a similar way to the way the forestry issue has worked for both major parties from time to time in the past.   It doesn't seem to be playing so well in poker-machine ground zero, Glenorchy in the seat of Denison.

There are a few interesting points from the pokies polling.  Pokies are, in isolation, an easy punching bag for poll respondents and most voters who have an opinion will say they're bad.  In this case the design of the opening question in the published pokies PDF  is in no way skewed and still 53.5% of voters say pokies have a negative impact on the community, 22.9% a neutral effect, 12.9% positive.  Labor voters (73.5%) and Greens voters (91.1%) overwhelmingly rate the impact as negative, but it is worth bearing in mind that this figure for Labor voters excludes those who have been scared off supporting Labor by this specific issue.  Even Liberal voters are more likely to rate them as having a negative impact.

Gaming revenue (I expect this includes not just pokies but also Keno and so on) contributes around 1% of state revenue (source) but the next question finds that of those voters who take a punt at the revenue share nearly half overestimate it greatly.  This is common in polling on everything from the percentage of Muslims in Australia to the percentage of Tasmanians who work in the forest industries.  What is especially odd here is that Labor and Green voters are more prone to overestimate the contribution of pokies to state revenue despite being more likely to think that pokies are bad.  So this suggests a failure of the left to communicate a useful argument (that pokies are only a modest source of state revenue) to even its own voters.

The other pokies results can be seen on the links above and I don't think they require further comment from me.  I will say that asking a question about donations after a barrage of questions on pokies could contaminate the response, but the idea that parties should be disclosing donations is so hard to argue against that it would have majority support anyway.  (That's not to conclude, as the Greens Leader does, that if someone has a different view on this they're not a grown-up and don't care about democracy.)

Why Isn't Booting Pokies A Major Vote-Winner?

It might seem baffling that if pokies are so detested by voters, and given that the Liberals do not seem to be serious about solving the terrible problems caused by gambling addiction, Labor is so far getting a rocky response on this issue on the hustings.  One problem here is that the "yuck!" response voters give in isolation might be different when a voter is asked to make a judgement about pokies versus jobs, even if that judgement is actually based on a false conflict.  (Anecdotally from multiple sources, those who respond negatively to Labor's policy at first will often change their minds following discussion, but most voters will not be engaged in such discussion.)

Another is that, in a way similar to same-sex marriage, pokies may be an issue on which Liberal voters will grumble about their party's policy but not feel motivated to switch their vote - while some potential Labor voters might well be scared off by job loss fears.  Scare campaigns about the thin end of the wedge to banning horse race betting, alcohol, smoking or whatever else may also be having some effect when the pokies debate, like the greyhound debate in NSW, is framed as an elitist attack against working-class pastimes.  (Frankly I'm surprised those running this line have not made more of the fact that Labor has at least one candidate who has been involved in attempts to get rid of smoking as well.)

Another reason is that some voters may agree that pokies are a big problem, but think that Labor's solution goes too far.  It's difficult to say how widespread this view is, because the polling we've seen so far only gives a forced choice between Labor and Liberal policies, and there is a very big gulf between those.

While Labor's position is "evidence-based" in terms of harm minimisation and overall community benefit, this is not solely an evidence issue; it is also a philosophical issue.  In that respect the growing trend of calling certain policies "evidence-based" as if that ends all rational argument is trite.  Even if getting pokies out of pubs and clubs is the best solution in terms of overall community benefit, is that sufficient reason to impinge upon the freedom of the hypothetical Burnie pensioner who only plays pokies sporadically with their friends, isn't addicted, doesn't suffer severe financial losses and doesn't want to have to go to Launceston casino (which would be an even worse bet) in order to play them?  My feeling is Labor's struggling to deal with that sort of argument.

(Personally, I don't like banning things in general - and this is a ban of sorts whatever its supporters might disingenuously claim otherwise.  I can't get outraged about restricting pokies to casinos though, because I freely admit to a snobbish contempt for all forms of purely luck-based gambling.)

I think the strategic question of why Labor decided to go with the policy it has - rather than, say, some cocktail of reduced bet sizes, reduced machine availability, tighter regulation, shorter licence periods and severe penalties for gambler-grooming -  is a very interesting one, and one that has not been at all well explored in the media.   Is it about rebranding the party for a break with the past?  Covering off against the Greens? There are many possible explanations.

If the party really thought that the high disapproval ratings of poker machines in polling made the issue an easy vote-winner, then that was a serious mistake, but I suspect that there is much more to it than that.

PS: It is also interesting to note that the pokies part of the poll was funded not by TAI (who conducted the poll) but by Impact Investment Group.


  1. News this morning that the Liberals have tried to hide their plans to relax Gun laws. Could this have an effect? Interesting they didn't have the policy on their website.

  2. The late-breaking news on Liberal gun policy is not a good look for them as the failure to release such a significant policy plays into the already present charge of secrecy over donations, and as Tasmania is quite sensitive to gun issues because of Port Arthur. All the same
    I generally find late-breaking events have little impact on voting (finding smoke signals from them in differences between votes cast before and after is generally impossible). Also any impact will be blunted by the electronic advertising blackout and the print media blackout tomorrow.