How accurate are the Victorian election polls?
This page was last updated on . The rows are rebuilt from the same register as the Victorian election polls 2026 tracker, so a new poll from any house replaces its row here on the same update.
After the Victorian Electoral Commission declares the result of the 28 November 2026 election, this page scores each house's final poll against it, by party and on two-party preferred. Until the count, it lists the poll each house would be scored on today and sets out the scoring, so the ranking that follows the result has its method on record before the result exists.
The poll each house would be scored on today
Each row is the newest poll from one house, with its figures exactly as published, and it is the poll that house is scored on unless it polls again before 28 November 2026.
| House | Fieldwork | Sample | L/NP | ALP | ON | GRN | OTH | 2PP L/NP–ALP |
|---|---|---|---|---|---|---|---|---|
| Newspoll | 23–26 Jul 2026 | 1,035 | 31 | 28 | 19 | 13 | 9 | 52–48 |
| The Australian; The Conversation; The Poll Bludger, 28 Jul 2026 | ||||||||
| Resolve Strategic | 6–12 Sep 2026 | 967 | 28 | 25 | 23 | 12 | 13 | — |
| Resolve Strategic for The Age, reported 14 Sep 2026; The Conversation (A. Beaumont), 15 Sep 2026; The Poll Bludger, 16 Sep 2026, for the sample and the undecided share Read the companion post | ||||||||
| RedBridge / Accent | 1–14 Sep 2026 | 2,371 | 29 | 24 | 25 | 15 | 7 | 56–44 |
| RedBridge Group / Accent Research for the Financial Review; The Poll Bludger, 16 Sep 2026 Read the companion post | ||||||||
| Freshwater Strategy | 31 Jul – 3 Aug 2026 | 1,020 | 30 | 25 | 22 | 14 | 9 | 52–48 |
| Freshwater Strategy for the Herald Sun; The Conversation; The Poll Bludger | ||||||||
| Roy Morgan | 5–7 Aug 2026 | 2,084 | 26 | 26 | 23.5 | 12.5 | 12 | 51–49 |
| Roy Morgan, special SMS survey, published 18 Aug 2026 | ||||||||
| DemosAU | 6–11 Aug 2026 | 1,007 | 32 | 23 | 22 | 13 | 10 | 55–45 |
| DemosAU / Premier National release, 6–11 August 2026, as reported by DemosAU and The Poll Bludger, 17 Aug 2026 | ||||||||
| YouGov MRP MRP model | 16 Jun – 10 Jul 2026 | 4,003 | 26 | 23 | 24 | 13 | 14 | 53.8–46.2 |
| YouGov MRP for Common Threads; The Poll Bludger, 10 Aug 2026 | ||||||||
A modelled estimate, such as an MRP, is scored as published and marked as a model, because a reader comparing houses wants to know which figures were a survey and which were a model.
How the scoring works
The arithmetic is fixed now so that the ranking after the result cannot be shaped by the result.
- Error by party
- Each party's error is the poll's figure less the result, in percentage points, so a poll that had Labor on 24 against a result of 27 scores −3 for Labor.
- Mean absolute error
- A house's mean absolute error is the average of its five party errors with the signs removed, and it is the single number the ranking sorts on.
- Two-party error
- The two-party error is the poll's Coalition share of the two-party vote less the result's, shown beside the ranking rather than folded into it, because houses derive their two-party figures by different methods.
- The ranking
- Houses are ranked from the smallest mean absolute error to the largest, and a house whose final poll is a modelled estimate is ranked with the rest and marked as a model.
The frame the result fills
The two result columns stay empty until the Victorian Electoral Commission declares the count, and the same arithmetic then runs on every house's row above.
| Party | Post-change average | 2022 result | 2026 result | Error |
|---|---|---|---|---|
| Coalition | 28.9 | 34.4 | — | — |
| Labor | 24.4 | 37.0 | — | — |
| One Nation | 22.9 | — | — | — |
| The Greens | 13.2 | 11.5 | — | — |
| Others | 10.7 | 17.1 | — | — |
| Two-party, L/NP–ALP | 54–46 | 45–55 | — | — |