Poll-Shaped Object Fails To Prove Opposition To Proposed Hobart Stadium

There is quite a deal of noise currently about a poll supposed to show opposition to a proposed new AFL stadium in Hobart.  Anecdotally, the concept is opposed by many northerners on the usual parochial grounds and by lefties (I'm suspecting it is not just lefties) who think the money should be spent on social priorities. It might be no surprise to find nobody much liked it then, but does the poll provide any actual evidence of this? I was one of those who was surveyed in this poll and I was not impressed.  

The poll has been hyped as a "leaked poll", which means that the source commissioning it gave it to journalists for free.  It was a robopoll of mobile and landline phones commissioned by Tasmanian Labor and conducted by Community Engagement, who are not at this stage an Australian Polling Council member and are hence not subject to public disclosure requirements.  

Community Engagement is a fairly well known minor pollster that does commissioned polling from time to time for the New Zealand and Australian Labo(u)r parties among other clients.  During the 2019 Australian federal campaign the Daily Telegraph accused the pollster of push-polling but the Telegraph's evidence was as flimsy as one would expect from the Telegraph: the best they could come up with was that the poll had asked respondents if  “Liberal Party disunity in Canberra” would influence their vote.  

The stadium poll was conducted two months ago (!) from October 6 to 9.  Partial results have been circulated among media, but only of one question.  The question was:

"3.1 The Tasmanian Liberal Government plans to build a 750 million stadium on Hobart's waterfront.  Do you support or oppose the plan?"

(The published results are 8.1% strongly support, 8.5% support, 16.1% neutral, 14.9% oppose, 52.4% strongly oppose.)

However, there were actually several more questions, the general thrust of which was unsubtly leading the respondent to agree that the stadium was a waste of money better spent on other priorities such as health and housing.  As a respondent I felt that I was being very strongly pressured down a certain line, even though it was a critique that I agree with.  The poll was not a push-poll as such (the publication of results alone demonstrates that) but that does not mean it was useful or commendable research.  Overall it was what I call a skew-poll (a poll with wording likely to lead respondents to respond in a particular way so that the client could get a result they could use for political purposes) but there was probably also an element of message-selling in it as well.

It's possible that the design of later questions could have caused some respondents to hang up, but the report in circulation provides no information on how this was handled.  Indeed mid-call termination rates are a subject it's almost impossible to find public information about for any pollster.  The poll used primitive weighting (age and gender only) and there was obviously some pretty heavy reweighting on that basis as the effective sample size was 1078 from an actual sample of 2541.

The lead question may seem benign enough but I am not convinced it is.  Firstly it highlights that the "Liberal" government has this plan.  There is no reason for mentioning this information, and it could trigger a partisan response (in either direction).  Using "The Tasmanian Government" would certainly lose no relevant information, but I am not even sure there is any need to mention whose plan it is.

Secondly the lead question alludes to an argument against the stadium - it will cost a lot of money.  But it does not provide any information about what the stadium is intended to be for, or any other information that might balance out mentioning the cost.  It would be better just to not mention the cost, since many respondents would be unaware of it, and by making respondents aware of the cost, the poll renders its sample unrepresentative of the Tasmanian voter base.  This renders the poll invalid as a measure of what Tasmanian voters currently think.

The polling report also contains a bizarre statement that the sample was drawn from a population of 863,969 across Tasmania.  Recently it has been reported that Papua New Guinea has been vastly undercounting its population but this news of lost tribes of 300,000 uncounted Taswegians lurking with their pet thylacines in Sensation Gorge and Tongatabu (not "Mt Tongatabu", by the way) will surely excite demographers everywhere.  (Or perhaps they were estimating the number of heads instead?)  Perhaps it is a typo for 563,969 but even then one would hardly count infants as part of a genuine sampled population.

We have been round this tawdry old block too many times before in this state.  During major debates such as the Tasmanian pulp mill, the kunanyi/Mt Wellington cable car and the old-growth logging debate it has been very common for lobby groups to put out polling supposed to show public support for their view but that actually has serious design flaws, most commonly the use of skewed wording.  I really don't know what the point is.  If someone commissioning such a poll is confident that (i) the public supports their case and (ii) the wording of the poll that purists like me complain about actually makes little difference, then why not use as scrupulously neutral a poll wording as you can?  Your poll claim is then protected from criticism, and if what you think is correct, the results won't be much different anyway.

I've seen still worse but given the amount of hype out there about this poll already I can do no more but promptly declare it to be a poll-shaped object and award it a Wirrah Award For Fishy Polling.

image source