Thursday, November 14, 2024

Poll Roundup: Labor Loses The Lead!

 

All things must pass.

2PP Aggregate (Last Election Preferences) 50.1-49.9 to Coalition (+0.4, first Coalition lead of term)
With One Nation adjustment 50.6-49.4 to Coalition
Election "held now" could leave Labor with about 70 seats

--

It finally happened!  After endless months of Labor's 2PP lead being painfully whittled towards zero, last weekend's Newspoll finally put the Coalition into the lead on my last-election 2PP aggregate.  It is true that I said they were eight weeks away from breaking Kevin Rudd's record for the longest aggregated 2PP lead since there was nothing much to aggregate, and they actually lasted nine.  So there is that, but it's not much consolation, and a lot of the other aggregates flipped two or three months back anyway.

Historically, it's no big deal, and perhaps not even a medium one.  Almost every government falls behind in polling at some stage in every term, except the first Hawke government which went to an election not long after the half-time siren.  Governments, albeit the other side's are better at it, have frequently recovered from being well behind and tend to poll badly in the last year before elections.  Labor has the benefit of a friendly pendulum from last time and the Coalition needs to win a lot of seats in some very different places to get another sniff at government in 2025.  

And yet, the results of the US election have seen an upswing of punter interest in the idea that Labor is really going to lose, to the point that they are now betting underdogs, with an implied chance of victory according to one bookie that bounced around between 40% and 49% for no sane reason yesterday before settling down around 43-45% so far today.  

Tempting as it is to see this as punters taking too strong a lead from Trumpism (ignoring differences between the US and Australia) or from polls underestimating Trump again, there's more to it.  The US results have shone a spotlight on what a lousy year governments are having worldwide, with every incumbent government in a developed nation that has gone to the polls losing vote share, which is unprecedented for years with sufficient data.  The best swing results in the world this year would put Labor down to a primary of below 30%, although not necessarily out of government.  Social upheavals connected to high inflation, cost of living and high immigration rates have made it a fraught time to govern.  2025 is another year and things may be very different then but the idea that one or two interest rate cuts would put all the voter pessimism to bed is one I'm having a lot of trouble buying.  Even if governments have good economic stories to tell, if it's not connecting with enough voters' own experiences, it doesn't help.   

The latest batch of polls has been as follows:

* This week's Newspoll was a second straight 51-49 to Coalition off primaries of Labor 33 Coalition 40 Greens 11 One Nation 5 Others 11.  The Coalition primary was the highest from Newspoll of the term.

* This week's Resolve was very similar at Labor 30 Coalition 39 Greens 11 One Nation 5, generic "independent" 11 and others 4.  I got this at 50.1 to Coalition by last-election preferences treating independent as independent, but 51.3 to Coalition if lumping "independent" and others as generic Others.  (That said the latter assumes a large UAP component in Others).  This was the first Resolve poll of the term that I converted as a Coalition lead.

* This week's Morgan was 50.5-49.5 to Coalition by its headline respondent preferences (51-49 to Labor by last-election) off primaries of Labor 30.5 Coalition 37.5 Green 12.5 One Nation 6.5 IND 8.5 other 4.5.  By my own last-election estimates this was actually the worst Morgan for Labor since late August, marginally worse than three others that also rounded to 51-49.

* Last week's Essential had the Coalition ahead 49-47 on its respondent preferences "2PP+" but the primaries were far more favourable to Labor at ALP 31 Coalition 34 Green 12 One Nation 9 UAP 2 others 8 undecided 5.  Essential is the only pollster still getting anything that looks much like the last election from time to time (something its use of past vote is prone to).  By last-election preferences I got this one at 51.7 to Labor, the second largest difference between Essential's respondent and last election preferences of the term.

So, four in a row with the Coalition ahead based on the pollsters' headline figures; the reason my aggregate doesn't have the Coalition further ahead is that I use last-election preferences and therefore treat Morgan and Essential as ALP leads.  And the reason it has them ahead at all rather than having Labor on about 50.4 is that both Morgan and Essential have become Labor-friendly on last-election preferences in recent months compared to the aggregate overall, and adjustments are applied to both of them.  This isn't noticeable in their headline respondent-preferences figures which are far from Labor-friendly.  

In terms of leaderships, Newspoll had Albanese at a new personal low of net -15 (40-55) and for the first time had Dutton with a netsat of -11 (40-51).  This was the first time Newspoll has had Albanese with a worse netsat than Dutton, but every single non-Newspoll since mid-June has had Dutton better and by an average of 12 points.  Newspoll also had Albanese's lead as Better PM cut to an anaemic four points (45-41).  Resolve had the leaders tied as preferred PM, and also had Dutton at a new high for that pollster of net +5.  It had Albanese net -14.  

Poor as Albanese's ratings are, it is common to see exaggerations of them from people who don't bother doing their research.  Supposed professor of politics Peter van Onselen, writing for the august scientific journal the Daily Mail, declared that "[..] the current numbers suggest [re-election is] up in the air, even though no first term government has failed to win re-election since 1931.  But none have been saddled with the sort of unpopular leader this government now is."  In fact, in his first term in office John Howard got down to net -25 in October 1997 and -31 in late June 1998.  He was re-elected on 3 October 1998.  He did lose the 2PP 49-51, but Albanese could plausibly do that and be re-elected too.  Malcolm Turnbull also got down to Albanese's current net level as the second PM of a different first-term government and won an election two weeks later.  

There have also been two reported WA federal intention polls, one by DemosAU with Labor ahead 52-48 and one by Redbridge with Labor ahead 54.5-45.5 - collectively rather benign and the latter copping some disbelief.  The Redbridge poll is however quite old (Oct 1-10).   

Aggregation Notes

A number of issues in polling aggregation that I should note here quickly.  Firstly, I've introduced (though I'm not graphing it) an alternative reading that includes a One Nation adjustment.  One Nation preferences were significantly stronger to the Coalition than at past elections in both the Queensland state election and the Fadden by-election.  This might just be a Queensland thing but I'm assuming it is nationwide and hence modelling it as a preference shift that I believe there is good evidence for and that makes sense.  The Queensland election however did not show any shift in Greens preferences.  I will continue to use the last-election tracking as a headline for consistency, but I will be using the One Nation adjusted version for projections as my best assessment of where things are at in seat terms.  

Since the start of September there seems to have been a shift in how Morgan last-election preferences relate to my aggregate.  My own conversions off Morgan had been running around 0.4 points above the aggregate for Labor for the first eight months, but this has suddenly jumped to an average of 1.2.  About 0.2 points could be attributed to the house effect being applied to Essential in the same time.  The remainder appears to be real; in a time in which Labor clearly isn't improving based on other polls, the average Morgan last-election 2PP has gone up from 50.4 to 51.0.  

The Guardian recently launcbed its own aggregate which gives very different numbers to mine, BludgerTrack, Mark the Ballot etc.  In particular, it models the Labor primary around the bottom end of what any pollster ever gets for Labor, and has the Coalition 2PP currently at 51.4, which is worse than every published 2PP bar the odd Freshwater.  Different perspectives are healthy - aggregators, like pollsters, shouldn't herd!  The main basis for the difference is that the Guardian's aggregate attempts to gauge the house effects of pollsters from past federal elections - in both 2010 and 2019 pollsters overestimated the Labor 2PP (quite badly so in 2019) and in 2022 they overestimated the Labor primary, an error which in a 2PP sense was cancelled out by stronger preference flows to Labor.   The danger for 2025 is that if pollsters overestimate the Labor primary again and there is a weakening of preference flows, then the average polled 2PP for Labor could come out too high by a few points.  

This is particularly a hazard for aggregates like mine that ignore respondent preferences (because of their poor predictive record to date and the fact that not all pollsters release them).  Since the middle of the year Morgan's respondent preferences are running 1.2 points behind last-election preferences for Labor (compared with 0.7 in the first half of the year).  Essential's are running 1.65 points behind (in the first half of the year it was 1.85).  In the Queensland state election the preference shift seems to have been modest, around 0.7 points, with most of that down to One Nation, but Green voters could be less happy with Albanese's Labor government than they were with Steven Miles', and federal Labor is coming off a record flow from the Greens in 2022.  

All this said the same polls do not show house effects to Labor in state elections, and that's one reason my aggregation philosophy is 'this is what the polls collectively are saying, once adjusted for those that seem to have a lean one way or the other and emphasising the polls with the best records' and not 'the polls will probably be wrong and by we think by this much.'  

I am still slogging through constructing my best estimate of the 2PP for Queensland from fragmentary data so as to put up my final Queensland poll review including a review of final polls.  Pending this I have already updated my aggregate weightings for Queensland as it appears that no reasonable 2PP estimate is going to change the order of polls on the dreaded table.  Resolve has been upweighted for its return to form with an excellent final poll, though the difficulty of knowing what to make of Resolve's mid-term polls with their high Independent scores compared to their final polls remains.  

Barring something dramatic in the meantime, the next poll review will be an annual summary posted in the final days of the year.  

2 comments:

  1. Thanks for the very thorough work and for providing so much detail on methodology as always. One thing I don't understand is the statement "So, four in a row with the Coalition ahead" if the most recent Morgan Labor 2PP by last election preferences is 51 (rounded) and the most recent Essential Labor 2PP by last election preferences is 51.7.?

    ReplyDelete
    Replies
    1. The Coalition was ahead on the headline figure released by the pollster in all cases - Morgan uses respondent prefs as its headline figure but also releases last-election prefs, while Essential only releases a (mainly) respondent based figure and I calculate a last-election estimate myself. (If a respondent has a preferred minor party for their primary vote but doesn't have a 2PP preference, Essential allocates them by last-election preferences). Have reworded to make clearer.

      Delete

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.