South Australia 2026

So here we are, on the eve of the South Australian state election. In some respects, this looks like a very straightforward race to predict; in others it's quite unusual. Polling across the board agrees that tomorrow should be a resounding success for the Labor state government. On the other hand - and by no means uniquely in South Australia - the long standing dominance of the Coalition parties appears to be heavily threatened by a large One Nation primary vote.

A Couple of Brief Excuses

It's pretty typical around these here parts for me to focus overwhelmingly on whichever house has the multi-electorate races partly because of the granularity that data often provides and mostly due to time constraints - I give myself extensive latitude to procrastinate with the excuse that you never know if more polling might precipitate from the aether.

In this election, though, there are two other "houses". One is the First Nations Voice. However, due to a lack of attention from media outlets, there is no polling I can find. Further, the vast majority of candidates do not have any party affiliation. So my excuse for not spending more time on this race is simply that I have no meaningful data to play with.

In the case of the Legislative Council, my excuse is more typical. The state-wide proportional representation (an increasingly popular approach to upper house elections) makes the predictive math relatively simple. (I say simple, not necessarily accurate.)

There are 22 seats to fill, and due to the staggered terms, 11 will remain in place:

For the remaining 11, the latest state-wide polling might be applied. The last upper-house-specific polling I saw came from 23 February and had Labor at around 33.8%. (N.B. this was reported as 38%, but excluded 11% of undecided voters; 38% or the remaining 89% is just over 33.8%). One Nation polls just under 18.7%, Liberals just under 13.4% and Greens around 9.8%.

More recent polling is, of course, available for the lower house. Even if this is representative of upper house voting intention, it doesn't shift the dial that much: Labor 34.4%, One Nation 21.4$, Liberal 15.8% and Greens 10.2% (after accounting for an excluded 7% undecided).

Any party polling over 1/12 (8.333%) of the vote is secured a seat, so this locks in 4 Labor seats, 2 One Nation, 1 Liberal, and 1 Green.

With a remaining primary vote between and 5.1% and 7.5%, the Liberals are well lined up to pick up another seat on preferences. The last two are a wild guess, with around one seats worth of support strewn across minor parties which could flow in any direction and another seat remaining in the "undecided" camp.

Nevertheless there is no mathematical possibility for the Liberal party to take a majority, and the possibility of Labor picking up the last two spots to secure 11/22 is purely theoretical. That means a hung upper house with Labor as the largest player, and a path to majority on any vote either with the Greens and one cross bencher to their left, or the Liberal party to their right.

The main game

We now turn our attention to the lower house. And in that "we" I of course include Antony Green, who published an electoral pendulum for the Legislative Assembly last year. I've altered this slightly, reverting Black and Dunstan back to pre-by-election margins, and dropping out a some against Independents including the seat of Mt Gambier which is vacant following the resignation and imprisonment of it's previous incumbent.

(I have assumed Kavel will return to a 2-party race with the retirement of Cregan, and while this year is a good year for a Liberal-turned-Independent to hold their seat, this is the first Independent race for MacKillop's McBride so it's easiest to treat the seat as Liberal for now.)

Now typically, we'd calculate the swing from last election to the latest two party preferred poll (54.6:45.4 to 59:41 is a swing of 4.4 percentage points to Labor) and apply that uniformly here.

But in this election, the Liberal Party isn't necessarily the party to be looking at in two party preferred terms. One Nation also polls 59:41 against Labor. Does this mean that We can construct a hypothetical ALP v ONP pendulum by just replacing all Liberal numbers with One Nation? Probably not. Do I have a better way of calculating such a pendulum? Nope.

But that same polling does suggest the Liberal primary votes exceed ONP in "Inner Metro Adelaide". I can't find a strict definition of how that term was used in the polling, but I assume that includes Bragg, Colton and Morphett. The other six? That'll come down to preferences.

Here's what we know on a more granular level: In Chaffey, One Nation came in third in 2022 despite the party's reasonably lacklustre showing across the state. Schubert saw a fourth place finish, behind the two major parties and the Greens who we may assume have negligible overlap in constituencies. In Finniss and Hammond, Independent candidates were more successful (though, of course, ONPs credibility as a viable third option has shot up significantly since then).

Chaffey certainly seems the most vulnerable to a ONP takeover, followed by Schubert. MacKillop feels ripe for an IND hold-out, as do Mt Gambier where we have the only district-specific poll, Narungga and Stuart. None of these are credible wins for Labor; the fall in Liberal popularity means they won't return to the blue; and the viable option of an independent largely removes the push-factor driving people to One Nation.

Final Prediction