Poll Roundup: Coalition In Freefall

2PP Aggregate: 52.3 to ALP (+1.2 in two weeks)
With One Nation adjustment (optional) 51.7 to ALP
If polls are accurate, Labor would win if everyone voted now, with a slim majority or close to it

(Figures above updated for YouGov)

The election has started already (people are now voting by post, even if the Greens HTV website doesn't seem aware of that) and the Coalition campaign appears to be in big trouble.  The polling swing back to Labor that started in late January and became more noticeable in late February has accelerated in the past two weeks and Labor is now polling majority government numbers from several pollsters.  It's too early to be confident that that will be the result, since elections are on average closer than even the final polls have it, but what we are seeing at the moment is a polling meltdown.  The Coalition primary vote is plummeting, and while this may yet turn around or be underestimated they now face the reverse version of Labor's problem during Labor's long decline last year.  There will be a bottom to this somewhere but no-one yet knows if this is it or where it might be.  By the time an underdog effect kicks in (if it does) will it be too late?

Overall what is happening at the moment according to most polls is the Coalition is shedding polled primary vote rapidly.  The lost primaries are spraying but enough would end up with Labor as preferences that Labor is surging on 2PP even without that much of a primary vote pickup.  The boom in the non-major vote at the Coalition's expense is such as to produce this startling statistic.  Prior to last Monday there had been no case all term where the combined non-major primary beat both the Labor primary and the Coalition primary. In just four isolated polls scattered through the term the non-majors had tied for the lead.  Since last Monday the non-majors have led both majors in five out of the seven polls that have been released!

To illustrate what has happened with the Coalition primary, in last week's Morgan it was down two points in a week, before rebounding by half a point this week.  In Newspoll it was down a point last week and another point this week (four points all up since early March).  In YouGov it was down 1.5 points in a week.  In last week's Redbridge, down two in a week.  In Essential down two in a fortnight.  In Resolve down three in a fortnight.  It was already falling before that.  We have yet to see Labor with a primary vote lead from anyone, but if this continues even that might not be far away.  The one black sheep in this story is Freshwater, which this week has the Coalition's primary stable on 39%, four points higher than anybody else.  

Aggregation Details And What Is The Deal With Morgan?

Once again I don't propose to replicate other services that spell out the primary votes, but just discuss how my last-election prefs aggregate sees the released polls.

Last week's Morgan was 53.5 to Labor by respondent preferences and 54.5 by last-election prefs.  This week's is 54.5 by both.  My aggregate has been heavily adjusting Morgan because of its very high recent numbers for Labor and aggregated these at 52.5 and 52.2 respectively.  Morgan's very high numbers for Labor have led to a lot of comment (have they changed something? Is this Morgan returning to its historic ways?) but I have a theory on what's happening here.  Morgan respondents seem to be highly engaged and Morgan's use of SMS may result in enthusiasm bias.  Morgan was actually slightly worse for Labor than my aggregate in general for months leading up to Labor starting to really improve in late February, but since then it's diverged rapidly.  However the reaction in Morgan weeks ago is now being picked up by other polls - consistent with something starting in more engaged voters and gradually spreading to more representative samples.  Of course, it's hard to know as Morgan is so non-transparent, but this is what may be going on.  

Last week's Redbridge was 52 to ALP, which I converted as 52.1 and aggregated as such.

Last week's YouGov was 52.5 to ALP by YouGov's modified preferences, which I got as 53.6 by last-election preferences.  YouGov by last-election preferences has been strong for ALP since it switched to generic ballot so I aggregated this as 52.3.

Newspoll's 52-48 by modified previous elections preferences seemed a little lucky for Labor based on the primary votes but this could be either rounding or the impact of, eg, a low TOP and high independent reading.  I aggregated this per my last-election conversion at 52.2.

Resolve came in as a rather startling 53.5 (method not stated) off the same major party primaries (ALP 31 L-NP 34) that they had in their very good final 2022 poll.  Resolve has been volatile lately having thrown a 45-55 outlier only six weeks earlier (though that one was partly a respondent preferences thing).  Because Resolve has been bad for Labor lately on average my aggregate took this one as a pretty big signal and aggregated it at 54.

This week's Essential came out with 50-45 to Labor (=52.6) by Essential's mostly respondent preferences; I got 53.2 by last-election.  My aggregate processed this as 52.6 given Essential's recent fairly good results for Labor.  

Finally this week's Freshwater came out with 50.3-49.7 to Labor prior to rounding. This was off the same primary votes that produced the previous 49-51; I have since found out that Freshwater is using respondent preferences.  I aggregated this at 50.9 to Labor given Freshwater's recent fairly strong form for the Coalition (this one being further away from the aggregate than other recent Freshwaters.)

By last-election preferences the nine polls released since the previous edition come out at an average of 52.8, but this is a sample including two Morgans (the strongest poll for Labor) and only one Freshwaters (the weakest).  There's a case that my current last-election prefs aggregate figure of 52.1 is actually a little overcautious (because of the way it adjusts for a poll's form relative to the aggregate in a situation where voting intention is now changing pretty fast) and it could reasonably be a few tenths higher.  Anyway I'm not going to fiddle with that now.

As I write Redbridge/Accent have issued a new MRP taken February 3 - April 1.  The very long time frame and the age of the data mean that I haven't included it in my aggregate, but will have a look at it soon and may edit some comments into this article.  The fact that it dates from a time before the Coalition vote went flop should not deter looking at it completely as it may be useful as concerns relative trends, especially if the polls come back.  

Leadership stats are also continuing to back in the trend as Peter Dutton's fate now looks sealed as the second Opposition Leader to record an entire term of negative netsats (Bill Shorten also did this but in his second term as leader).  Anthony Albanese has beaten Peter Dutton on net approval in every poll since the previous Freshwater, by an average of ten points (even the current Freshwater which is very different to the others, though his advantage there is only one point).  He even polled a net +2 in  Resolve which in its outlier a couple of polls back had him net -22.  While Albanese's preferred prime minister leads are still underwhelming by historic standards of such things skewing to incumbents, he did at least get double digit leads in three of his last four readings, after having only one such since May last year.  

The Shrinking Preference Shift

Through the term there has been a general view that Labor's preference flows in 2022 could not possibly be matched and Labor might even do far worse on preference flows in 2025, as Greens and One Nation voters especially became disillusioned with the government from their various perspectives.  But in the two polling series that regularly issue respondent preferences results, the gap between respondent and last election preferences in the Coalition's favour has been falling.  First, Morgan:

Secondly, Essential:

Morgan at one stage had the gap at 1.5 points and now has it as half a point.  Essential once had the gap at nearly two points and now has it less than one.  For whatever reasons the Coalition is tanking both on primary vote share and on respondents saying they will preference it.  Furthermore, the latter started well before the former.

I have been using a One Nation adjusted figure that routinely chops 0.5-0.6% off the last-election 2PP for modelling purposes and recommending that such a preference shift be assumed, but for the time being I'm more of the view that a meaningful shift isn't a lock to occur.  While I do still think One Nation preferences will shift, several pollsters are still amalgamating Trumpet of Patriots and Independent under "other".  The TOP vote will probably be well below the 2022 UAP vote as TOP are running in only two-thirds of divisions.  The Independent vote is likely to be higher than 2022 as there are more indies and more well resourced ones.  This suggests the use of a last-election "others" flow for these polls, which my aggregate does, could be underestimating flow from "others" to Labor.

For the time being, Labor appear to be winning and to have prospects of winning quite strongly.  If current polls were reflected they would be quite likely to remain in majority and might even increase it, while the Coalition's weak primary (if accurate) could see them drop some more seats to those pesky tealoid entities.  If Labor did lose their majority, it would probably be a manageable minority a la NSW, not a messy one.  But there is still much that can happen and the ghosts of federal elections past always suggest caution about ALP polling leads.  

What's All This Then?

It's common for governments to make up ground in polls if trailing during campaigns, but a government taking the lead and continuing to surge like this is quite unusual (it happened in 2004 but only in the last couple of weeks).  A government polling at its previous election result at this stage is rare; we last saw this in 2001 (when a massive national security bubble following the 9/11 attacks was diminishing) and last for a first term government in 1984 (when the polls of those days were rubbish anyway; they greatly overestimated Labor's lead).  What has been happening the last two weeks is unusual in Australia. 

Obituaries are already being written for the Coalition's campaign. It's hard to tell how much of the polling slump has to do with various errors and a remarkably lopsided tally of candidate problems, and how much has to do with forces beyond the Opposition's control, in particular economic uncertainty caused by Donald Trump.  But while the Coalition can't stop Trump from existing they can decide how to react to him,  The problem here is they have been all over the map.  People who hate Trump think the Peter Dutton is way too Trumpy and people who like Trump think he is not nearly Trumpy enough.  Making Jacinta Price a shadow minister for government efficiency when they already had one may have seemed like a fun tip to the Trump movement when they were ahead but it is not so funny now, especially when Price accused the media of being obsessed with Trump then was found to have been wearing a MAGA cap.  Some Labor supporters are enjoying the Coalition suffering its own dose of how Labor gets stuck (or gets itself stuck) between the Greens and anti-Greens.

To some degree this election reminds me of 1990.  The Coalition then seemed to be competitive in the leadup but their campaign was pretty silly from slogan down ("The Answer Is Liberal") and they ended up caught in an absurd xenophobia fight about the imaginary Multi-Function Polis and getting their leader branded "Peacock a danger in the Lodge" in The Australian.  Peacock's ratings were way worse than Dutton's.  In that case though the Coalition, generally written off by election day, did perform surprisingly well on the night, to the point that it wasn't immediately clear what the result was (leading to the introduction of indicative 2PP counts).  

Seat Polls

The last two weeks saw the most ridiculous seat poll claim possible with Compass Polling claiming to have found a 29% primary vote swing against Chris Bowen in McMahon with ex-Liberal independent Matt Camenzuli on 41%, and for some reason The Australian published this.  No methods details were available.  

Much of the action in seat poll reporting has been in Dickson where at last count four non-neutral seat poll results have been claimed, with only one of these (a conservation group uComms that had Labor ahead 52-48) surfacing with partial detail (the raw primary votes anyway).  The others claimed were a Coalition Freshwater at 57-43, a Labor internal at 50-50 and a uComms for independent Ellie Smith supposed to have Labor 51.7-48.3 ahead (a flip of the current margin).  Without seeing far more details of this shell game it's hard to say much about any of these. As the Coalition's polling overall has worsened, Peter Dutton's slender margin in Dickson has been getting attention lately outside the more excitable ALP faithful.  I would expect Dutton to get a boost for becoming leader (particularly in Queensland which has not had many home-state leaders) but that can be muted by the fact that he had a very high profile as a long-term major minister and leadership contender in previous elections.  Dickson as a seat is not that demographically easy for the Coalition at the moment as it has a fairly high income and also a high education level, and such seats are in general transitioning away.  The other thing to bear in mind here is that Queensland federal polling has a long history of being nonsense.  

I have commented on recent Tasmanian seat polls on my Tasmanian guide here but don't have the time or energy at present for a full roundup of others out there.

Betting

Betting isn't reliable but can be interesting to watch.  Betting sites now have Labor around a 77% chance; while this may appeal to people who make dimwitted "follow the money" posts on Twitter (I am thinking of a policy of blocking on first sight for those) this is still below where Labor was on election day 2019 when they lost.  Type of government markets have Labor minority slightly ahead of Labor majority, which at one stage was at a ridiculous $15 and is now around $2.50.  

Seat betting was very slow to follow the headline odds and polls up but over recent weeks has come through.  When I checked on April 10, markets had flipped Lingiari and Robertson back to Labor, made Labor a unanimous but long favourite in Brisbane and become split about Bradfield.  When I checked again yesterday, Paterson, Chisholm, Bullwinkel and McEwen had joined the flood back to red (though with Bullwinkel tied in one market) and independents were now favourites in Curtin (with one tie), Bradfield and Cowper, the latter two being the first seats the Coalition was favourite to drop.  This made Labor favourite in 75 seats, Coalition 62, Greens 2 and others (mosrly independents) 12.  After considering close seats, a reasonable read of what the markets "think" is 74.8-61-2.8-11.4 - the Coalition now has more seats where it is wobbly than Labor.   The long list of seats markets expected Labor to lose has been pared back to Aston, Bennelong, Gilmore and Lyons with Bennelong very close to flipping.  This is not very different to the Australian Election Forecasts take (which is lightly influenced by betting) though AEF is more bullish on the Greens,  having them favourites in Ryan and Brisbane albeit not odds on, and on a median result of four seats.  

There is too much going on at this election!  I am struggling greatly with the problem of there being only one of me but hopefully the Easter break from other work will give me more time to get some more pieces written (and also deal with the Tasmanian Legislative Council Pembroke hot war!)

Election Night

Finally my election night plans are still unresolved; I've been too busy to do anything about it!  If any established reliable non-Murdoch media source with capacity to handle high visitor numbers wants to hire me for live blogging, preferably not behind a paywall, I'd be interested to discuss (link to email in profile).  If nobody does, I'll do it here with an extra tedious level of begging for donations, and possibly pay for a live cover style feature to avoid the need for refreshing, and just hope it doesn't overload this ancient platform!  

YouGov update

The new YouGov saw a major party tie on the primary vote (33-33) with a 53-47 2PP to Labor by their modified preferences.  I got 54.5 by last-election preferences which was aggregated at 53.1.