Thursday, June 17, 2021

The Major Parties Are Not "Neck And Neck" In Victoria

RESOLVE PM (Victoria state) Labor 37 Coalition 36 Green 9 IND 12 Others 5

2PP Estimate 54-46 to Labor.  If numbers repeated at election, Labor would win easily (c. 50 seats)

"Independent" vote very likely to be overstated

-------------------------------------------------------------

I've had too little time for writing here in recent weeks, largely because of a backlog of contract work that I had to clear after it built up during the Tasmanian snap election.  There are a few pieces I have been working on that I do hope to finish some time but they will be well behind the news cycle should they actually appear.  However, I wanted to make some comments about reporting of today's Resolve Political Monitor poll of Victorian state election voting.  This furthers a concern I have had about some responses to the 2019 federal polling failure - that some media sources that commission or work with polls have responded with trendy solutions that lose information and then lead to worse reporting of what polls are actually claiming to show.  

In my initial coverage of Resolve's entry to the major polling markets I noted that fingering two-party preferred figures as a major culprit in the 2019 federal polling failure was simply incorrect: polls were wrong overwhelmingly because their primary votes were wrong, with errors in preference estimation making only a small contribution to the failure.  I doubted the claims that this would deliver readers "something deeper" than "the “horse race” nature of the way we reported the results of TPP questions" and suggested that what this would actually lead to was journalists reporting the horse race off primary votes.  In the case of the reporting of this specific Victorian poll, the horse race commentary hasn't gone away, it's just got worse.  It's like being told just the relative positions of the horses near the end of a race without being told one of them is flagging and the other is charging home strongly.  


In this case Labor has a 1-point primary vote lead over the Coalition, compared to 7.67 points at the 2018 Victorian state election.  However, while Labor won the primary vote by 7.67 points in 2018, it officially won the two-party vote by 14.6 points.  It would have been over 15 points had the Liberals not piked on contesting the seat of Richmond.  In Victoria, the Greens are by far the most significant minor party and their preferences give Labor a massive edge.

It's difficult to know what to do with Resolve's estimate of 12% for "independents" (who actually polled only 6.07% in 2018) and really it would be useful to have some insight as to which side's vote-parkers these might be (it could be a lot are soft Labor voters).  Some voters confuse "independents" with candidates for micro-parties and others might want to vote for an indie but be unable to find one who is prominent enough for them or shares enough of their views.  In 2018, 34 of the 88 seats did not even have any independents running.  (Comparable figures were 53/93 for NSW 2019 and 77/151 for Reps 2019, and these are also polling markets where Resolve is getting far higher Independent votes than what they actually get.  In contrast Newspoll doesn't use the word "independent" in seats where independents did not run at the last election; it offers "another party or candidate" in all seats).

But even treating these as generic "others" I estimate these primary vote numbers would produce about a 54-46 2PP in Labor's favour, based on my estimate of 2018 election preferences after adjustments for (i) the Resolve totals adding to 99 not 100 (ii) Richmond and (iii) three-cornered contests.  Labor would win easily if these primary numbers are accurate.  On current seat boundaries an effective uniform 3.9% swing (after adjusting for Richmond) would cost Labor around eight seats, leaving them with around 47 seats out of 88.  However (i) personal vote effects from the 2018 election would protect Labor in its marginal seats and would be likely to reduce the damage by a seat or two (ii) an impending redistribution is also likely to benefit Labor to the tune of a seat or two.  Labor would be a good chance to keep 50+ seats on the new boundaries if this poll is correct.  

None of this is more than hinted at in the Age's reporting of the poll.  The lead article contains barely a hint to readers that the contest painted by these numbers isn't really close ("the Coalition doesn’t appear to be in an “election winning position”" is about as close as it gets).  Another article dealing with the Liberals' current leadership situation refers to "the major parties neck and neck", and "the tight poll margin" which could "embolden Mr O’Brien’s strategy and quieten detractors."  Now, for sure, only being behind 46-54 and only trailing 23-49 on the skewed preferred Premier metric is polling better than the Victorian Liberals might have feared, and is polling better than Labor were getting in NSW prior to their leadership change.  However, an election held right now in Victoria would not be tight on these numbers.  None of this is to argue that the Liberals should or shouldn't be in a hurry to change leaders.  I for one am not convinced that going back to Matthew Guy so soon after the 2018 debacle would be wise.  

I should note that the Age report refers to "a reduction in preferences flowing to [Labor]" but it is not clear whether this is simply inferred from the reduced share of the preference pool that comes from the Greens, or else from respondent preferencing.  (The "About the data" section does report that "ranked preferences" are used in the voting intention polling via a sort of simulated ballot, but there is not enough detail of how this works - detail that Resolve would be publishing if they were an Australian Polling Council member.) In any case, respondent preferences are unreliable.  And on these numbers, it doesn't matter: on any realistic preference flow from these primary votes, Labor wins easily.  It's not so different to the last major poll (YouGov in late Oct 2020) which had Labor up 55-45.  

On the matter of leaderships, the poll finds fairly good but decidedly sub-Gladys ratings for Daniel Andrews at net +10 (42-32) and Acting Premier James Merlino at +15 (30-15). The most striking impression re the ratings of Michael O'Brien is the level of indifferent responses.  Resolve's personal ratings have been previously reported as "favourability" but in fact come from the question:

"For each leader, please tell us whether you have heard of them and, if so, whether you have a positive, neutral or negative view of them?"

A high proportion of survey respondents don't know who an Opposition Leader is by name without being told that person is the Opposition Leader (to which their response might well be "oh, that guy"). I suspect this explains much of the difference between these Resolve polls (which have been finding barely a third of voters with a positive or negative view of Opposition Leaders) and YouGov state polls where usually at least 60% of voters, sometimes a lot more, are either satisfied or dissatisfied with the Opposition Leader.   Perhaps telling voters that someone is Opposition Leader provides too big a partisan prompt to some voters who actually have no idea about that leader, and Resolve are laudably seeking to avoid that, but it's rather difficult to compare a net rating of -8 (14-22) with YouGov's October net rating for O'Brien of -27 (26-53).  It's also possible that those voters who say they have never heard of an opposition leader would tell YouGov that they were not satisfied with that leader because they had never heard of them.  

We've Been Here Before

It's painful to see wheels of bad poll reporting reinventing themselves just because there has been a polling failure.  The reporting of poll results through the lens of primary votes in ignorance of preference flows became more of a bugbear in the early 2000s as both the level of third-party voting and the flow of preferences to Labor increased.  Here's an article by Peter Brent from the time pointing out that the Australian was continually reporting Labor as behind in the national horse race when based on the numbers provided, after preferences, Labor would quite often be ahead.  Around this time Newspoll started routinely publishing 2PP estimates, but during 2004 they made the error of using respondent preferences, and partly because of that had Mark Latham much closer to the Lodge than he actually was.  Since then, last-election preferences of some kind have generally worked well at projecting what elections would look like if the primary votes in polls were accurate.  That "if" is an important proviso: the 2PP estimate is only as good as the primary vote sampling underlying it, and if the primary votes are way out then the 2PP also tends to be nonsense.  

The Victorian Liberal leadership is a current matter of public speculation, some Victorian Liberals' political tactics even more so.  It does public debate about said leadership and tactics no favours for media to incorrectly report a poll as if it shows a very close battle between the major parties, when if the poll is remotely accurate then it shows no such thing.  Rather than getting rid of "horse race" commentary, in this case the reporting is keeping it ("neck and neck" and all!) but also dumbing it down and distorting it.  

This is not the only aspect of this poll's reporting that is suboptimal.  The lead report says that:

"Victorians were also lukewarm on the state government’s latest budget with one in four voters believing the measures, which including billions for mental health funded by a big business tax, would be bad for the State as a whole. Only 28 per cent of Victorians believe the budget will leave them better off."

I haven't seen the text of these questions or the precise results of them yet, but what isn't mentioned here is that voters are usually pessimistic about whether budgets will leave them better off!  In the context of federal Budgets, Newspoll has found more than 28% of Australians believing a Budget would improve their lot only seven times in 34 years, and this year (19%) wasn't near being one of them.  Likewise, more than a quarter of Australians have thought a federal Budget would harm the economy 21 times out of 34.  So those are not lukewarm responses to a Budget, they're quite good.   

Why is it so?

Opponents of Daniel Andrews' government may be mystified that polls keep showing it in decent shape when Victoria's experience of the COVID-19 outbreak has been worse than other states (and Victorians give their state much worse ratings for handling of COVID too).  Not only has Victoria had more serious outbreaks, but it also struggles to contain them without lockdowns in comparison with New South Wales.  However, firstly, the polls do reflect this to the extent that the Andrews government has a substantial polling swing against it from the last election, the only state for which this is true.  (Western Australia's government had an immense swing to it at a recent election, Queensland's and Tasmania's had small swings on a 2PP or 2PP-equivalent basis, NSW's government has a large swing to it in polling and South Australia's is around about where it was.)  Those who expect that COVID and lockdowns alone should be driving a double digit swing back in Victoria should probably look at some of the overseas governments that are romping in the polls - the Tories in the UK in particular - in spite of COVID death rates that are fifty times Australia's. The other point is that Victorians do not only mark their state government down when there are outbreaks - they blame the federal government too.  

No comments:

Post a Comment