South Australia 2022: Can Even COVID Save The Marshall Government?

We're just three weeks away from the 2022 South Australian election and the first Newspoll of the campaign has just given a reminder of how challenging the Marshall Liberal Government's task is.  The poll has the Coalition trailing 47-53 off primaries of Coalition 37 Labor 39 Greens 10 Others 14.  Strikingly, Marshall trails Opposition Leader Peter Malinauskas 39-46 as Better Premier (an indicator on which opposition leaders rarely lead unless governments are in big trouble).  Marshall's own rating is a shadow of its former COVID-boosted self at a still reasonable net +1 (48-47) but Peter Malinauskas is doing much better at +20 (51-31).

Perhaps the latter offers some hope for the Liberals since they may be able to dent the Opposition Leader's image, but that the Marshall Government appears struggling to avoid a first-term defeat at this stage is not historically surprising.  Firstly, the government was elected with a small majority and is federally dragged, so it would be expected to more likely than not lose some seats.  Secondly federal drag also predicts how state governments will go compared to their leadup polling, which for the SA Liberals wasn't crash hot anyway.  Thirdly the government has lost a majority it started with, and this is often associated with defeat (though I am a little more cautious about this at present given that voter contempt for hung-parliament games has seemed in recent years to be rising, both in Australia and the UK.)  At some time analysts were wary of predicting state governments would lose after a single term, but first-term losses in Victoria 2014 and, far more spectacularly, Queensland 2015, reminded us that it does happen.  

During 2020 and 2021, Australian governments enjoyed a wave of electoral success on the back of the pandemic, winning the Western Australian, Queensland and Tasmanian elections (with a huge swing, a modest swing and the equivalent of a modest 2PP swing respectively), the NT and ACT elections and loseable by-elections such as Eden-Monaro.  The NT government even managed to win a by-election from the opposition.  The recent NSW by-elections saw the first meaningful loss for an incumbent government in the seat of Bega, which might be taken as some signal that the days of free votes could be over.  Even that is not even near conclusive though, as the average 2PP swings in the NSW by-elections were below the average for by-elections in that state in the previous 23 years.  The NSW results were patchy, had a lot to do with candidate factors, and the Perrottet Government has been far more criticised for its handling of COVID than Marshall's has.

Mike Rann's win in the 2010 election is the only previous case (from 8 attempts) of an Australian Premier winning an election while trailing as "Better Premier" (Rann trailed by 2 points, and won despite losing the 2PP vote).  The current poll, however, is presumably not the final poll and therefore comparing it to that history is not comparing like with like.  As far as I know, though, Isabel Redmond's 45% Better Premier score in that 2010 poll is the highest score any Opposition Leader has polled in any state Newspoll and gone on to lose the election, so Malinauskas would break a record if he failed to win from 46%.  Malinauskas is, incidentally, the only state Opposition Leader to have served continuously through the pandemic, and in general this has seemed for good reason as his bipartisan approach to COVID has so far paid off.  

The seat layout

The SA election is interesting for the impacts of the redistribution, the number of independent MPs contesting (nearly all of whom have either moved seats or defected since the last election) and the likely disappearance of SA Best, who at this stage have announced candidates only for the Upper House.  SA Best polled 14.2% in 2018 with its voters' preferences splitting 50-50.  

The 2018 result was 25 Liberal 19 Labor 3 IND, with two independents in Liberal-2PP seats (Geoff Brock and Troy Bell) and one in a Labor-2PP seat (Frances Bedford).   Brock and Bedford have both switched to target incumbents in new seats following the redistribution, with Brock tackling Dan van Holst Pellekaan in Stuart (Liberal 11.7) and Frances Bedford running in first-term Liberal Richard Harvey's ultra-marginal Newland (Liberal 0.1%).

I am assuming Bell retains Mount Gambier pending polling evidence otherwise.  If that is the case then given that Brock is vacating Frome, the Liberals start with 26 Liberal-2PP seats, of which three are currently held by the ex-Liberal independents Sam Duluk (Waite), Fraser Ellis (Narungga) and Dan Cregan (Kavel).  If they gain one more seat from independents than they lose to Labor, Bedford or new independents, then the Liberals are back in majority.   It used to be safe to assume that candidates like Duluk and Ellis would lose, but after Troy Bell's victory and hold on his seat, we have to wait and see. 

Labor's easiest path to a majority is to win its 20 2PP seats (including regaining Florey) and also pick up the low-hanging marginals Newland (0.1), King (0.6), Adelaide (1.0) and Elder (2.0).  The next brace of seats are all above 6%.  However Labor's task is complicated by the presence of Bedford in Newland and by personal vote effects for the Liberals in Newland and Elder (won from Labor in 2018) and King (new in 2018).  With those taken into account, Labor's task becomes more difficult, and I get Labor being more likely than not to win the 2PP in a majority of seats at 51.4% 2PP (this is based on my usual variable swing model with personal vote adjustments), whereas the Liberals become more likely than not to get 24 2PP wins (not counting Mount Gambier) on only 49.7%.  The difficulty for the Liberals is that they are far more vulnerable to independents taking their 2PP seats (eg if any of Duluk, Ellis, Cregan, or Brock wins, or Bedford should Newland remain Liberal-2PP) while the only such threat that is obvious for Labor is if Bedford stops them from winning Newland (in which case she would be expected to support Labor anyway).  So the target for an actual majority is likely to be higher.

I convert the current Newspoll to around 25 Labor 2PP wins to 21 for the Liberals (22 including Mount Gambier), which shows that a modest 2PP advantage would not necessarily win the election outright for Labor, though they would have good prospects of being the largest party. There has been a long history of 2PP results not converting very well to seats in SA, which tends to have relatively few marginal seats, though mostly this history has to do with (now removed) advantages in conversion for Labor.

Aside from the Newspoll, there has been one poll by Dynata (which is not an Australian Polling Council member and has yet to establish a track record of successful election forecasts).  The huge 17% Independent/Other vote in this poll makes it hard to read much into it anyway.  And there have been very small sample uComms seat polls that have Bell holding on in Mount Gambier and Brock falling way short in Stuart, but it will be impressive if these are all that accurate.  A particular concern in Stuart is that Brock has polled very well in the past in parts of the seat, so a random sample might or might not capture those parts successfully.

See also Poll Bludger, Tally Room, ABC

I will be covering the count live on the night of March 19, and post-counting points of interest over subsequent days.