Poll Roundup: Coalition Support Slides In Pre-Budget Polling

2PP Aggregate (Last-Election Preferences) 51.1 to ALP (+1.5 in last four weeks)
With One Nation Adjustment (Recommended) 50.5 to ALP
If polls are on average accurate, Labor would almost certainly win election "held now", probably in minority

In the four weeks since my previous instalment there's been a substantial shift in national polling and Labor has recaptured the lead in both versions of my 2PP aggregate.  Although not every poll in that time has supported the shift, the trend is overall so well supported that when Morgan came out with a 54.5-45.5 to Labor outlier on Monday, the three polls later in the week did scarcely anything to peg back the gain in my aggregate that Morgan produced.  Looking at primary vote aggregates the culprit here is the Coalition primary.  Of the six polls that polled at least once both prior to 25 Feb and since 25 Feb, on average the Coalition primary is 1.6% lower since 25 Feb, though much of the gain went to independents and non-Green minor parties.  The most recent polls are carrying a heavy weighting in my aggregate because there are so many of them, and the suggestion for now is that Labor's lead is continuing to build. However it is Budget week, and these are not the best of times for trying to use a Budget to fuel electoral success as the Coalition has often done in the past. One of the reasons that I want to put this article out now (and update it with polls that come out entirely before the Budget) is to have a clear baseline for where things stood before the Budget did its thing.  (A brief refresher: Budget bounces in polls rarely happen - on average following a Budget a government goes slightly backwards).  

There are a few likely reasons why the Coalition is tanking.  One is that there have been a series of policy announcements that have been poorly received and/or badly communicated - insurance company breakups, bans on public servants working from home and a very off-key proposal to hold a referendum on deporting dual citizens.  What's strangest about the latter is that when Peter Dutton shelved his proposal to hold a second referendum on Indigenous recognition should the Voice fail, he said "it’s clear that the Australian public is probably over the referendum process for some time", and slammed the government for wasting money on the referendum that could be used on cost of living issues.  Now throwing money at another confusing and doomed referendum would be good?  Overall the Coalition has seemed in the last few weeks to sorely lack campaigning match fitness, and this is no surprise since while Dutton has held the show together remarkably well for much of the term, he has been carrying a frontbench who are mostly just not up to it or going through the motions. 

The second is that there could be some "Trump drag" effect.  The first Trump presidency was chaotic but fairly harmless where Australia is concerned, but version 2 is destabilising expectations by slapping tarriffs on historic allies (including us) and sucking up to conventional rivals and Cold War foes in Russia.  Some voters may be nervous about an opposition that they fear might be too compliant with Trumpism. Dutton isn't actually all that Trumpy lately (he was very sharp in distancing himself from Trump over Ukraine for example), but some will still think it's an act, or overplay passing similarities.  But at the same time while Trump is at least faintly toxic for most Australian voters, there are voters on the right who want the Coalition to go down the MAGA path and are supporting minor right parties because it isn't doing so.  These votes will not reliably return as preferences.  For this reason while some would have seen Clive Palmer buying into the pre-existing Trumpet of Patriots party as bad news for Labor, I think it was actually more annoying to the Coalition.

The third is that Labor has had a pretty good run of things for several weeks now with the interest rates cut followed by a series of trouble-free policy rollouts, although general voter sentiment polling (right direction/wrong direction) still remains bleak (a dire 29-51 result in the recent Redbridge).  Going forward, I think the cyclone-induced shift away from an April 12 election has actually helped Labor in that they can make going the full term look like the right thing to do rather than desperation.  While the Budget may be a very hard sell, to put out a Budget anyway and say 'this is how it is and we are making the mature decisions' should look better than running away from the Budget for no easily explainable reason.  

This is flowing through to leadership polling too.  After a long run of mostly having worse net satisfaction ratings than Peter Dutton, Anthony Albanese has had a better rating than Dutton in the most recent Newspoll, Freshwater and Essential polls with the latter his first net positive rating from anyone since the Voice referendum loss (albeit only +1 and from Essential which tends to be mild).  Dutton has however continued to beat Albanese on net rating in YouGov except for one tie.  Albanese is also tending to do slightly better on preferred PM polling compared to earlier in the year, though his leads remain in single digits (historically below par) with a best so far of 9% in Newspoll.  

Aggregation Details

I don't think it's necessary for me to keep posting primary vote details from individual polls when these are mostly available on Poll Bludger, Wikipedia etc but I do want to give some examples of how my last-election preferences 2PP aggregate is handling some of the current polls.

This week's Morgan at 54.5-45.5 (by both last-election and respondent preferences) raised eyebrows concerning Morgan's bounciness and historic cases of Morgan skewing to Labor at election time, although for much of the term Morgan has not been that bouncy and hasn't skewed to Labor at all.  My last-election preferences conversion for this poll was also 54.5, but especially following a few other good Morgans in recent weeks for Labor, my aggregate decided Morgan could have developed a leaning compared to other polls and processed the poll at 53.4.  A reason why Morgan's house effect could vary over time is its partial use of SMS polling which may be prone to motivated response bias compared to online surveying.   At times when Labor voters are more motivated and Coalition voters less, it may be the former are more likely to take part.

This week's Essential was 47-47 by their "2PP+" method using mostly respondent preferences, which is the same as 50-50.  My last-election conversion was 50.5 to Labor but Essential has been strong for Labor lately.  My aggregate therefore credited Labor with just 49.9. 

Last weekend's Freshwater was 51-49 to Coalition.  It is unclear what preferencing method Freshwater is currently using.  My last-election conversion for their primaries was 50.4 to Labor, from a poll that has been somewhat Coalition-leaning though in recent polls that effect on last-election converted estimates has reduced.  My aggregate credited Labor with 50.9 by 2022 preferences.

This week's Redbridge was 51-49 to Labor. My last-election estimate was 51.1 and currently my aggregate doesn't think this figure for Redbridge has a house effect either way, so it went in at 51.1.

This week's YouGov was 50-50 by YouGov's modified preferences (based on respondent prefs and an unstated number of previous elections).  My last-election estimate was 51.5 but YouGov has for whatever reason (or perhaps no reason) been good for Labor since they switched from Newspoll-style readouts that list parties that ran last time in a seat to a new ranked generic ballot method similar to Resolve.  My aggregate therefore credited Labor with only 50.5.

The Newspoll of two weeks ago was unusual in not showing any real recovery in Labor's position, producing another 51-49 to Coalition by modified previous-elections preferences.  This went into my aggregate as 49.8 for Labor, the same as the previous Newspoll.  

If my aggregate did not include adjustments for polls that appeared to be falling on one side of the current estimates or the other, and did not include accuracy or frequency ratings, its estimated last-election position for Labor could be even higher (around the mid-51s).  That said, last-election preferences are likely to be overestimating Labor's position, and for seat modelling purposes I am using the One Nation adjusted version (currently 50.4 to Labor after at one stage falling to a precarious 49).  

Anyway aside from a very brief and small lift for the stage 3 tax cuts, this is the first sign of a real lift in Labor's polling position all term at a time when there were enough polls for it to be meaningful.  This is significant in that if Labor falls behind again, at least we know now they are capable of bouncing back.

What would this mean?

If the polls are broadly accurate, the government is now again in a winning position - potentially an ugly win but the chance of the Coalition forming government if the current polls were reflected would be low.  An election "held now" with a 2PP of 50.4 to Labor (my One Nation adjusted figure) would be expected to deliver Labor 2PP wins in 73 or 74 of the seats included in my classic-2PP model, assuming swings varied only randomly with the exception of personal vote effects.  Even if the swings landed badly for Labor as both the most recent YouGov and Redbridge MRP models predicted, Labor would still be likely to win the 2PP in about 70 of these seats.  There would be risks of actually losing a few of these to the Greens, but on the other hand there would be prospects of taking some of the Brisbane-area seats from said party.  It's just very hard off a 2PP of 50.4-49.6 to get the Coalition to as many seats as Labor, and if we get a really hung parliament with both sides mucking around in the high 60s I suspect Labor stays in government anyway.

Believers in the theory that Labor is going to get whacked in the outer suburbs have taken succour from large swings in the WA state election in some areas, and have especially argued the case for the Liberals taking Pearce.  Thst's fine but if you take from Column A you have to take from Column B.  The WA election must be seen in the context of an eight-year old federally dragged government coming off a remarkably lopsided pandemic-politics election.  A massive swing back to the Opposition was inevitable and for the Liberals to only pull back about 12% 2PP in the circumstances was poor  For sure, state factors including leadership disunity and poor strategy played into it, but the result said nothing bad overall about the condition of the Labor brand in the state.  If one wants to argue that the results in the outer suburbs were warning signs for Labor, then at least by the same token the inner-suburban results in Perth were dire for the conservative forces, who failed to win back seats that should have been absolute gimmes and barely got over the line in Nedlands and Churchlands that they should have won 60-40 in their sleep.  Magnify that failure nationally and it means that a shift away from Labor primaries supposed to cause teal seats to fall won't happen and that the Coalition should be nervous about its own remaining inner-city marginals (classic-2PP or otherwise) at least outside Victoria.  I'm not saying this is how things will actually play out - just that one can't spin the WA result as a clear positive for the Coalition federally, and I don't recommend trying to spin it as a positive at all.  

Labor's deep problems in Victoria mean that even on a national swing of less than 2% there could be casualties in the range of Bruce, McEwen (already widely written off alongside Aston and Chisholm), Holt, Hawke, Dunkley.  But if Labor is getting 50.4% 2PP but copping the swing needed to really get into the second-tier seats like Holt, Hawke and Dunkley then that would most likely mean little or nothing was happening outside Victoria and on that basis it's very hard to get the Coalition into government.

Polls Are Snapshots Not Predictions (etc)

The surge in Labor's polling is likely to give rise to optimism and even premature hubris among some of the faithful online but we still have between six and eight weeks to go, which is still heaps of time for things to change back.  Or get better.  Or the current polls may not collectively be super-accurate anyway - federal polling has overestimated Labor in some way (primary and/or 2PP) in four of the last seven elections, though there is no evidence of a similar effect at state level.  What I do hope the recent shifts back and forth highlight is that there has been overconfidence throughout the term about the chance of a hung parliament.  A hung parliament of some kind still looks likely, but isn't a certainty yet and never was.  Also there are hung parliaments and hung parliaments, and while the range of possible non-majority parliaments includes some that could be unstable messes, it could also be a NSW 2023 like situation where no-one gets a majority but it's obvious on the night which side has won and the dramatic connotations of "hung parliament" don't really apply. 

Betting - What On Earth Isn't Going On

I like to keep an eye on betting markets although they are not reliably predictive.  The behaviour of the markets lately has been quite strange, with no response to Labor's improving polls until very recently, and even then the markets still have the Coalition narrow favourites.   It's possible the lack of movement in Newspoll is a factor here and bettors aren't taking the other polls seriously (in some cases with good reason) but it was surprising to still last week see Labor at only around a 36% implied chance (now around 45%).  An ALP majority has come in from a ridiculously long $15 to $6.50 yet there has been virtually no movement in seat betting at any time since it commenced: markets still expect Labor to lose Aston, Chisholm, McEwen, Bennelong, Robertson, Paterson, Gilmore, Tangney, Bullwinkel, Lyons and Lingiari and the Coalition to gain Curtin, Ryan and to have the best chance in Brisbane (though no-one is odds on in the latter).  

It seems that overall betting markets took the Trump victory in the US as a big signal (it happened there, could happen here) but don't seem to be taking so much of a counter-signal that it happening there could disadvantage the chances of "it" happening here.  I struggle to see what isn't going on here as logical and suspect the seat markets especially don't have a lot of money in them.  

Seatpoll-Shaped Objects

While seat polls or bracketed polls of groups of seats are starting to emerge there is not much to see at this stage as most are not neutrally commissioned and all are being poorly reported by usual suspects.  The News Corp tabloids breathlessly announced polls of seats of Green interest conducted by Insightfully for Advance.  While I have seen some people praise Insightfully's internal polls, I can only examine public polling and the two Insightfully public polls known to me (one of rural seats in the NSW election and a Tasmanian one of the Voice) were both inaccurate; the Tasmanian Voice one was right re the state result but with electorate breakdowns that were obviously way too similar to each other.  

The numbers are supposed to show the Greens winning only Melbourne but these are presumably based on respondent preferences, which are especially likely to be unreliable in seats where Labor candidates are excluded because of the higher rate of how-to-vote-card following among Labor voters.  On my read of the pathetically incomplete primary vote data supplied by the perennially useless Herald-Sun, the Greens would retain Griffith and win Macnamara provided that Labor did not recommend preferences to the Coalition, and probably retain Ryan unless the Labor flow to the Greens weakened sharply, while Brisbane would be clearly won by Labor (it seems the poll only bothered canvassing a Liberal/Green 2CP).

News Corp tabloids also published a poll of six teal seats by Freshwater which collectively found what they claimed to be a 5% swing (taking into account the redistribution it was actually 6%).  They reported primary votes of teals 33 Liberal 41 Labor 7 Green 7 but hopelessly failed to report the identity of the remaining 12%.  The overall sample size of 830 doesn't enable any useful conclusions about specific seats and the YouGov and Redbridge MRPs had significantly different outlooks for these seats, so I wouldn't place too much weight on this one at this stage.  Although the poll is described as "conducted exclusively for News Corp", I've heard they were given it for free as sometimes occurs with Redbridge, rather than paying for it.  

Climate 200 seat polls have also been "reported" with the Saturday Paper uncritically acting as a dropoff point for claims that Climate 200 backed candidates are winning Cowper 53-47 and Bradfield 52-48 and trailing in Forrest 49-51 and Flinders the same.  However the Forrest mention refers to the vote for Nola Marino, who is well kown to be retiring.  The Saturday Paper doesn't even bother to state the name of the pollster or provide any method details whatsoever.  Climate 200 had some especially successful Redbridge polling with advanced modelling at the 2022 election but if this is uComms automated polling with simplistic age/gender weightings it requires a lot of caution as it may be over-capturing engaged/educated voters.  That said it has not appeared in uComms' APC disclosure statements yet.

I'll update this article and the graph at the top with tonight's Morgan and anything else that comes out that is entirely pre-Budget and then start a rolling post-Budget roundup post once the first post-Budget polls appear.  

Morgan Update Monday:  Morgan came out with another very strong poll for Labor if such things can be believed, this came to 54-46 by last-election preferences (53 respondent-allocated).  With two of these in a row my aggregate is now even more convinced that something has sent Morgan on a red team bender, so only credits Labor with 52.3 for this one, but even that put it up another tenth of a point to 51.1.