South Australian election
South Australians voted on 21 March 2026, three days after the final DemosAU wave closed. The March briefing scores that wave against the count.
Intelligence Unit · Series IU-03
Poll Briefings
Each briefing takes one DemosAU wave and reads it: the primary vote against the last election, the cohorts carrying the movement, the measures behind a government's standing. Every one is a page, and the briefings whose PDF has been filed offer it for download.
Every wave in the cycle, read in the order it was polled.
Each coloured line is one jurisdiction's series and each dot on it is a briefing, placed by the dates the wave was in the field. Open a briefing to see what it connects to: the briefing itself, its PDF where one has been filed, and DemosAU's own release and full data for the same wave. The federal MRP model runs on its own branch beside the monthly federal poll, because the two are different products with different samples.
Victorian election
Victorians vote on 28 November 2026. Every wave on the Victorian branch leads to this point.
The Liberals lead on 28%, Labor is on 22% and One Nation on 21%, and no party has moved more than two points since June/July. On a seat benchmark no pair of parties reaches 18 of 35 seats, and EMRS's August poll places One Nation at the same 21.
The Coalition's two-party lead narrowed to 52–48 from 55–45 in August, and the preference flows this wave revises account for two-tenths of a point of that move. No party's first preference shifted outside sampling error, and Labor records its highest primary vote since October 2025.
One Nation has fallen to third for the first time this year, leaving Labor on 26%, the Coalition on 25% and One Nation on 24%. The fall is concentrated among women, and DemosAU's seat ranges put the Coalition on 32–50 and One Nation on 26–39, against 11–22 and 53–66 in July.
Seven months from the election, Labor is steady on 32% and leads 57–43 after preferences. The Coalition is up five to 25% and back in second, and One Nation is down four to 23% from a record June reading.
Released three days before the Secret Harbour by-election. Labor is steady on 33% and leads 54–46, while the Liberals have slipped to a tie with One Nation on 21% — a party that polled 4% at the 2025 election.
The first wave since Ben Carroll replaced Jacinta Allan. The Coalition leads 55–45 after preferences, unchanged from June, and both major parties lifted two points taken from the minor parties rather than from each other.
The third federal MRP is the first in which no party's seat range reaches a majority: Labor 68, One Nation 60, the Coalition 13, and the top of Labor's band is 72 against the 76 needed. One Nation rises in all twenty-six groups the model reports.
A sample of 2,694 still cannot separate One Nation's 29% from Labor's 26, which is the third wave running to decline the question. Of seven policy positions put to the same respondents in One Nation's name, one has majority support and five are rejected.
One Nation's six-point rise to 27% is the only movement in the wave outside sampling error, and it takes 37% of regional voters against Labor's 23. The government's rating improved on all seven issues it was tested on while its vote fell.
The Coalition falls five points to 18%, the only movement in the wave that clears sampling error, and its projected seat range collapses from 16–28 to 4–11. Offering the same respondents the Coalition under a different leader leaves its vote exactly where it was.
The two-party measure now runs Labor against One Nation, at 56–44, three months after an election One Nation finished second in. This is the first South Australian wave whose intervals establish the full ordering of the top three parties.
Labor's primary vote falls to 33% and its two-party lead narrows four points to 53–47. The government's rating fell on four of the eight issues it was tested on, and One Nation takes 29% of the regions against the Liberals' 14.
Every change since February is inside sampling error, and the two points between One Nation's 23 and Labor's 21 do not establish an order. One Nation takes 31 per cent of TAFE-qualified voters, where Labor takes 14.
No party's first preference moved outside sampling error since February, and the single point between Labor's 25% and One Nation's 24% establishes no order. The LNP's two-party share rose two points anyway, because 74% of One Nation preferences return to it.
One Nation is printed first at 28%, Labor second at 26 and the Coalition third at 23, and none of the three gaps between them is large enough for this sample to establish an order. Asked to imagine Pauline Hanson had retired, the same respondents give the Coalition twice the seats.
The final poll before the 21 March election, scored against the result three days later. Every party landed within 1.9 points of the final poll and One Nation within 0.1.
One Nation enters the lower-house question at 21% and the any-other-candidate share falls from 20 to 7, so thirteen of those points were already in the poll. The 61–39 two-party figure is struck on the 78% of the ballot that does not exhaust.
One Nation's modelled seat estimate goes from 12 to 52 in four months and the Coalition's central forecast of nine sits at the floor of its own 9–17 range. The model also stops borrowing two electorates' preference flows and measures them from respondents instead.
One Nation is listed for the first time and takes 17%. Nine of those points come from the Liberals rather than from the any-other-candidate category, which is the reverse of what the same change did in New South Wales.
One Nation rises seven points to 21%, the only movement outside sampling error. The LNP's primary vote falls three points and its two-party share rises two, which the published preference flows account for exactly.
One Nation enters the lower-house question at 19%, level with the Liberals, and eleven of those points come from the collapsed any-other-candidate category rather than from a party. The poll declines to order One Nation and the Liberals one point apart, and the election five weeks later separated them by four.
The Liberals fall six points to 35%, the only movement among the four main parties that clears sampling error, and the seven points the two major parties lose between them turn up across the four other columns. The poll orders the first two places and leaves third unresolved between independents on 17% and the Greens on 15.
One Nation enters the lower-house question at 21 per cent. The Coalition's primary vote falls eight points and its two-party share rises two, because the report assumes three in four One Nation preferences reach it.
Eight months after the election no party had moved more than 2.0 points and the two-party figure was 1.1 from the result. The ballot did not offer One Nation, which took 17% once it did.
Labor loses primary vote and gains four seats, because the Coalition loses more and One Nation takes it. Twelve seats from 17% would be the largest third-party result in the House of Representatives since 1934, and the report sets out four reasons for it that are not the vote share.
DemosAU's first Victorian wave of the cycle, and the only one it generated itself. The Coalition leads 51–49 after preferences, and One Nation was absent from the lower-house question while taking 11 per cent in the upper.
Three months after the July election no party had moved more than 2.4 points, on the smallest effective sample in the series at 500. The Liberals take 53% of mortgage holders against 25% of renters, the one division in the wave wide enough to clear its interval.
Labor leads 59–41 after preferences on a lower-house question that did not offer One Nation, which took 15% of the upper house. Two-party support runs from 70% among voters aged 35 to 54 to 50% among those 55 and over.
The LNP's two-party lead is 0.2 points from its election result while its primary vote has fallen 4.5, Labor's 3.6 and One Nation's risen 6.0. Compulsory preferences return three in four One Nation votes to the LNP.
The lower-house ballot carried no One Nation line while the upper-house question in the same poll put the party at 12%. Labor leads 47–21 on first preferences and 66–34 after them, and ramping and the cost of living are the two issues the government is rated worst on, both at 42 points negative.
| Branch | Briefing | Fieldwork |
|---|
What a briefing is, and what it is not
The polling is DemosAU's and so is the methodology. A briefing is Ace Strategies' reading of a published wave, written after the fact and filed under a document number so it can be cited. Each one names its pollster and its sponsor, and the method note on every page separates the two claims.
The pages carry the figures as text rather than as pictures of tables, so they can be searched, quoted and checked. Where a briefing's PDF has been filed it is the same briefing in the form that goes into a meeting, and its download button appears on the briefing and in the timeline above. Five of the thirty are filed so far.
Related
- The polling archive holds every wave, with the full data release for each.
- Methodology sets out how each wave is built, weighted and modelled.
- How to read a poll covers margin of error, preferences, seats and outliers.
- DemosAU is the research and polling partner behind every wave in the series.