Why the annual results tell you almost nothing.
Every August the results land and the coverage says scores are up, or down, or stagnating. ACARA publishes a measure of how big each of those movements is. By that measure, almost none of them are movements at all.
Of the 860 year-on-year comparisons of average NAPLAN score in the published file — every state and territory, every domain, every year level, three consecutive year-pairs — 100 per cent are classified “close to”, ACARA’s lowest category. Not one clears the bar.
Computed from ACARA’s own published dataset, using ACARA’s own classification column. Nothing here is an outside critique of the assessment — it is the authority’s arithmetic, read.
What ACARA publishes that nobody quotes
Alongside every comparison in the results file, ACARA reports two separate things. The first is whether a difference is statistically significant— whether it is large enough to be distinguished from sampling noise. The second is the effect size— how big the difference is relative to how much students vary from one another anyway.
Those answer different questions, and in a dataset of this size they answer them very differently. With hundreds of thousands of students, a difference of one or two points is comfortably significant. On a scale whose standard deviation is around a hundred points, it is also nothing. ACARA resolves the two into a single classification it calls the nature of the difference:
| ACARA says | When |
|---|---|
| “Close to” | not significant, or an effect size below 0.2 |
| “Above” / “below” | significant, and an effect size of 0.2 to 0.5 |
| “Substantially above / below” | significant, and an effect size above 0.5 |
Coverage quotes the significance and drops the size. That is how a movement ACARA itself files under “close to” reaches the public as a result.
What the file actually contains
Take every year-on-year comparison in the published dataset for the all-students figures: nine jurisdictions including the national total, five domains, four year levels, four measures, and three consecutive year-pairs. That is 3,600 rows, of which 3,440 carry a classification — the rest are suppressed. Sorted by what ACARA called them:
3,395 of 3,440 — 98.69%— are classified “close to”.
Of the remainder, 44 reach “above” or “below” and exactly one reaches “substantially above”, in the whole file, across three years and every jurisdiction in the country.
The measure the headlines use is the emptiest one
Split those comparisons by which measure they describe and the picture sharpens considerably. What movement exists is concentrated almost entirely in the proportion of students reaching Exceeding— the top proficiency level. The average score, which is the number the coverage reports, contains nothing at all.
| Measure | “Close to” | Anything more |
|---|---|---|
| Average NAPLAN score | 100.00% | 0 |
| Strong or above (%) | 100.00% | 0 |
| Developing or above (%) | 99.19% | 7 |
| Exceeding (%) | 95.58% | 38 |
Each of those rows covers 860 comparisons. So the headline number — average score, the one that becomes “results are up” or “results have slipped” — has not produced a single classified movement anywhere in Australia across the entire period the current scale has existed.
What this does not mean
Two overclaims are available here and both are wrong, so it is worth naming them.
It does not mean nothing happened. “Close to” is not “identical”; it means the movement is too small, relative to the spread of student performance, for the measuring authority to call it a difference. Real changes in classrooms may well be underway and simply too gradual to surface in a two-year window.
And it does not mean NAPLAN measures nothing. The same file contains differences ACARA classifies as substantial in every single comparison— between students by parental education, by Indigeneity, by remoteness. Those are large, stable and consequential. The assessment is detecting something real and serious. It is the annual up-or-down framing, specifically, that carries almost no information — and it crowds out the findings that do. The 2026 Victorian results works through one of those.
What this means for your child
The practical upshot is short: when the results-season coverage arrives, you can safely ignore it, and you should not let it change anything about how you think your child is going.
- A state or a school going “up” or “down” this year is not evidence of anything. It is not a reason to move schools, to worry, or to relax. If you would not have changed your view without the headline, the headline should not change it.
- Your child’s own result is a different question from the national one.The noise finding is about year-to-year movement in aggregate figures. A proficiency level on your child’s own report still tells you roughly where they sat on one day — see what the four levels mean.
- Watch the things that do move.Whether your child writes more fluently than last year, reads harder books than last year, revises without being asked — those change on a timescale you can actually observe, and none of them appear in a national dataset.
How this analysis was done
Every figure is read directly from ACARA’s published NAPLAN national results dataset for 2026, sheet “Year on year comparisons”, filtered to the all-students rows. The classification counted is ACARA’s own “nature of difference” column, not a recomputation: this page tallies the authority’s verdicts rather than forming its own. The 0.2 and 0.5 thresholds those verdicts use are stated in the NAPLAN 2025 Technical Report and were checked against the file’s own coding before being quoted.
Two limits. The window is 2023 onward, because four proficiency levels replaced the older ten-band scale that year and the two scales do not join up; nothing here should be extended backwards across that break. And 160 of the 3,600 rows carry no classification because they are suppressed — almost all of them Tasmania’s 2026 figures, which are withheld in the published file. They are excluded from the percentages rather than counted as zeroes.
Sources
- Australian Curriculum, Assessment and Reporting Authority (ACARA), NAPLAN national results dataset, 2026.Every score, percentage, effect size and classification quoted across these pages.
- ACARA, NAPLAN proficiency level descriptions.Definitions of the four levels: Needs additional support, Developing, Strong, Exceeding.
- ACARA, NAPLAN 2025 Technical Report, pp. 86–87.The effect-size and nature-of-difference methodology: the 0.2 and 0.5 thresholds separating “close to”, “above” and “substantially above”.
- ACARA, NAPLAN (National Assessment Program), and the Victorian Curriculum and Assessment Authority, NAPLAN assessments.What the assessment is, who sits it, and how it is administered in Victoria.
Something that does tell you where your child is
Twenty minutes on your child’s actual writing — what it does well, what is slipping, and what would move it. Free, and without obligation.
New to all this? Start with what NAPLAN actually is →