Lambiemania: What Should We Make Of The Tasmanian YouGov Poll?

YouGov Tasmania (state) Liberal 31 Labor 27 JLN 20 Greens 15 IND 7
If recorded at an election (unlikely) would result in a hung parliament, probably leading to a Labor minority government
High JLN figure likely to represent a bubble, with response options contributing - but consistent with JLN being in the mix for seat wins

Populist minor party bubbles often happen in state elections.  In the late 2017 SA Newspoll, Nick Xenophon's SA-Best was polling 32%, ahead of both major parties, and Xenophon himself was 5% ahead of both major party leaders combined as Better Premier.  Was South Australia heading for Premier Xenophon?  Just a few months and one of the worst ads in political history later, the juggermaut had ground to a halt, with the party polling 14.2% on election day and winning zero lower house seats.   

And while One Nation did manage 22.7% in Queensland 1998, there have been plenty of other bursting bubbles: One Nation Queensland 2017, Palmer United Tasmania 2014 and Queensland 2015 and yes Jacqui Lambie Network Tasmania 2018 are other cases where minor parties that threatened to win several seats ended up with one seat or nothing.  (Another one is Territory Alliance in the NT 2020 but that is a weaker example since the only poll that pointed to success there was an internal poll.)  Campaigns often result in fourth parties being exposed as rabbles full of terrible candidates and contradictions.  They can be outspent by big parties or simply sidelined from media debate because they won't be governing.  

The startling finding of the first ever YouGov Tasmanian state poll released today should be viewed with that cautious framing, among many others.  Some others include:

* The poll was taken between December 21 and January 4 when many voters would be on holidays.  This is less of a problem for online polls like YouGov, since people don't need to be home to take a poll - but it could still lead to less representative responses.  (Note that only the last day in field includes John Tucker's recent threat to confidence and supply.)

* The sample size of 850 (effective 769) is on the small side, but that said it is not that much smaller than EMRS and it is not small enough to by itself invalidate the broad pattern of the poll.  

* JLN have already announced some candidates for the election but they have been a relatively minor presence in the election lead-up so far with very little advertising.  Lambie herself will not be on the ballot and voters will need to vote for individual Lambie Network candidates (they can't just place a 1 in the Jacqui Lambie Network box like they did in the Senate).  

* YouGov is a panel-based pollster and the range of Tasmanian voters it would have access to would be smaller than for EMRS.  We saw some panel pollsters (including Resolve and Morgan) be way off with their Tasmanian samples for Voice, though Pyxis Newspoll's sample was successful.   

* The leaders of the major parties have been specified to make it clear to voters that they should be thinking about state and not federal politics.  But given that the leaderships of both Rockliff and White are often criticised, this might have the effect of making disgruntled party followers more likely to pick some other option.  

* The only options offered to voters were Liberal, Labor, Green, JLN and a single named independent in each seat.  (No Shooters, Fishers and Farmers for example).  This is likely to have inflated the JLN vote substantially.  Voters were allowed to be undecided after a double pass and still stay in the survey but those who were undecided at that point were excluded from the voting results - the same method of dealing with undecided voters used by Newspoll, but Newspoll will generally offer more voting options.

I don't believe JLN would actually get 20% in an election "held now" and I'll be amazed if they get that when this bumpy term is finally submitted to the judgement of a baffled electorate.  But while there could be responses that this poll is simply rubbish, the idea that something like that percentage might pick JLN when offered a limited range of options is not one I think should be dismissed out of hand.  

The 2022 Senate race gives a sort of idea of this.  JLN's primary vote was 8.6%.  But on a four-party preferred basis that rises to 13.4% (compared to 35.3% for the Liberals, 30.1% for Labor, 19% for the Greens and negligible exhaust).  When voters are given a limited choice list it's pretty easy to get JLN well into the teens.

Even assuming this is, as it seems to be, a bubble further boosted by a restricted list of voting options, it's still way higher than what JLN need to win seats.  In 2018 they ran in three seats, got 3.2% of the statewide vote, and would have been lineball for a seat in the 35-seat system.  A vote of, say, 8% statewide would give them good chances of winning three seats.  

YouGov has released individual seat breakdowns but given that they are based on samples of only about 170 electors per division they should be treated with extreme caution (as the individual sample size is tiny), and are mainly useful for pointing to broad response patterns.  The YouGov projection of 11 Liberal, 10 Labor, 7 JLN, 6 Green and 1 independent seems close enough to correct to me if these numbers were actually polled (and nobody else ran) but many warnings apply.  JLN have previously suggested they're not running in Clark (which they might rethink), the Green vote in Braddon tends to lag their state total a lot more than this (six seats off 15% would be very fortunate), and it would be very surprising if Lambie Network got more in Franklin than Braddon.  In short almost no weight should be placed on these seat breakdowns (though the 2% for Lara Alexander in Bass is bound to have a high approval rating among pollwatchers.)

An 11-10-7-6-1 parliament is fantasy-football land at this stage but for those determined to play such a game, I'd expect that that would be the end of the Liberal Government, either immediately or within a year.  Even if any deal could be negotiated between the Government and the whole of the JLN grouping, it would be only a matter of time before someone from the latter grouping split and brought down the government, either putting in a Labor-led mess or else leading to another election.  While we are talking here about numbers that I don't think we'll be seeing on election day, the poll does point to a potential problem for the Rockliff Government - if it is short of a majority, it may struggle to find anyone to form government with. This has become more acute with the recent threat to confidence and supply from John Tucker.

This result from the poll is very damning if correct:

It needs to be kept in mind that the Liberal vote numbers are radically different from EMRS, which has a good record in Tasmanian elections.  EMRS indeed is finding almost no appetite for minor parties other than the Greens (most of its IND/Other vote is supposed independents, but some respondents would be using that term incorrectly), however it has not to this stage included JLN in its list of party options.  

What do we know about YouGov?

As YouGov has not been seen with a Tasmanian voting intention poll before, some background may be useful.  YouGov is a major international firm that in Australia is best known through its former association with Newspoll.  This association ended in 2023 with the main Newspoll modellers departing to form Pyxis.  YouGov recorded an excellent result at the Voice referendum but in that case it was reproducing the questions that had been used in Newspoll, rather than writing its own.  As an entity independent of Newspoll it has yet to be tested at a regular Australian election, but has been doing Queensland polling that is pretty similar to what other polls are getting.  

YouGov is to my knowledge unique among the major Australian pollsters in running an open panel that anyone can sign up for (whether they will be picked for any specific poll among the deluge of brand surveys and so on is another matter entirely).  While this is a risk in terms of over-capturing politically engaged voters, YouGov have recently added a useful insurance against that by using self-reported Voice vote as a weighting.  YouGov is an Australian Polling Council member that publishes basic methods details.  This poll does not include a weighting for education, though I would not expect that to be a problem in the case of the JLN vote because JLN voters are typically not highly educated.  

I expect YouGov will poll again at some stage before the next election.