The Test Got Smarter. Nobody Was Given the Job of Noticing.
Your child sits NAPLAN in Years 3, 5, 7 and 9. Somewhere between 2016 and 2022, without much announcement, the thing they sat changed — not just how it was delivered, but what it was doing while they sat it.
It has been administered annually since 2008 as a synchronous census of more than a million students, across 10,000 schools, testing reading, writing, language conventions and numeracy.
It helps to have two words rather than one. Brennen and Kreiss distinguish digitisation, converting an analogue process to machine-readable form, from digitalisation, the restructuring of social domains around digital infrastructure. NAPLAN Online, the 2016–2022 conversion of the national assessment, contains an uneven mix of both. It digitalised an assessment instrument while not embedding the institutional proficiency or appetite to leverage newly enriched data.
That distinction is not academic hair-splitting. It is the difference between putting the same test on a screen and building a different test.
The test that watches you take it
Centralisation of a single online platform enabled accelerated reporting and the infrastructure for digital innovations, such as machine learning algorithms and tailored testing. Tailoring allows students to be routed through skill-based testlets according to earlier responses, thereby changing the mechanism through which proficiency is elicited beyond the channel of delivery.
Read that last clause slowly, because it is the whole story in one sentence. Two children in the same room, sitting the same NAPLAN, no longer sit the same questions. What each answers early determines what each is asked next.
This is a genuinely better instrument. The digitalisation of NAPLAN brought real strengths in faster reporting, particularly in adaptive precision confirmed through randomised trials, and in engagement and disability-access gains. It is also an instrument nobody outside the centre can see the workings of.
Why it was done
The reform was advanced to stabilise a policy environment already turning against NAPLAN's high-stakes paradigm — a decade of union, academic and parental discourse which cast the test as onerous and narrowing. The complaints were specific: a narrowing of the curriculum towards tested content; teachers' work burdened with administrative data production; mounting costs to student and professional wellbeing.
Digitisation was advanced as the answer on three grounds. Efficacy — tailored testing providing a "better measure of the Australian Curriculum, more precise results and a significantly faster turnaround of those results". Efficiency — retiring paper assessment enables a faster turnaround, reducing outsourced logistics and costly interpretative work, such that classrooms can quickly action findings. Centrality — concentration of response-level data in a single platform enables keener cross-jurisdictional analysis for policy, and a richer diagnosis tool for teachers.
That is, the reform negotiated a middle ground wherein NAPLAN's inordinate burden on teaching would be ameliorated by efficient, cost-effective assessment, further tempered by arming teachers with data that advantageously steers their practice.
Notice what that bargain requires. It requires someone to be tasked with turning richer data into better teaching.
Six years, one deadline, one outage
A national deadline was set by the Education Council in 2016 against jurisdictional readiness, with "all schools expected to transition by 2019". However, states deferred "to gain a greater level of confidence in the assessment platform". Thus 2018 — which held dual-mode paper and digital tests — produced a dispute on comparability without a mechanism ingrained in the governance, and lacked an external certification of platform confidence. On 14 May 2019, a nationwide disruption forced a commissioned root-cause review, which was withheld, and transition extended to 2021. Following COVID, NAPLAN was "fully administered online for the first time in 2022", six years after the deadline.
This timeline represents a weak inter-institutional commitment to the terms of digital reform. Change could scarcely be reached, let alone institutionalised.
Who was watching
Six actors carried the reform, but one relationship set the pattern: Education Services Australia, which built and operated the platform, was simultaneously the Education Council's own legal entity until 2021–22. Overseer and operator were one body, an arrangement built and maintained for the delivery period.
The consequences of that arrangement are architectural. The item bank, algorithm and audit were centralised and closed to public oversight. Subsequently network infrastructure, device readiness and incident response lacked a central control, and were muddied through a devolution of responsibilities to the states — in practical terms the eight test administration authorities — and were thereafter handed down to schools without the technical expertise to resolve issues or interpret outcomes.
Nor was there much appetite for finding out. The reform sat within the Auditor-General's mandate, and I have not been able to find a governance-centric performance audit of it. This instantiates an organisational will to assure only the high-stakes paradigm of NAPLAN, while shirking responsibility of failures in the sites of innovation — adaptive testing, question-level analysis — into "more of an IT thing", as it was put to the Queensland Parliament in 2019. The NSW Secretary's inability that year to state the cause repeats the sub-federal prerogative to escape association with the NAPLAN Online black box, and culminated in the root-cause review which never saw public scrutiny.
What each party was promised, and what arrived
The beneficiaries of the digitalisation occurred in proportion to their distance from the classroom.
| Who | Promised | Delivered |
|---|---|---|
| The national centre | Concentrated response-level data; faster reporting | Delivered — item bank, algorithm and data centralised |
| Ministers | Cross-jurisdictional evidence | Partial — speed; longitudinal comparability abandoned in 2023 |
| Researchers | Richer national datasets | Refused — item-level request declined |
| Schools | Faster school reports | Partial — 2023 results ‘cannot be directly compared’ to prior series |
| Teachers | Actionable diagnostics | Split — per-item data delivered, items withheld; IT readiness devolved unfunded |
| Students | Better experience; precision at the tails | Split — engagement gains; regressive penalty on low-income households |
The teachers' row is the one to sit with. User engagement was thin. User-research techniques should run from discovering needs through to operating, measuring and improving, and yet ACARA's consultation was massively front-loaded — heavy at the design stage, thin once the thing was running. Outcomes remain opaque: the NSW Centre for Education Statistics and Evaluation notes that "item content for the majority of online test items will not be released to teachers and school leaders", which "prevents deep analysis of item performance" formerly routine on paper.
A teacher can now be told which question a class failed, without being told what the question was.The scale that had to hold
There was one more thing the reform had to manage: two tests, on two media, reported on one scale. Equating the 2018 dual-mode results, ACARA's technical report concedes the initial adjustment "did not sufficiently resolve inconsistencies", in that scores below the minimum standard were "upwardly adjusted by the equivalent of around one standard error", and that a Year 9 writing mode effect was left uncorrected. To forge the continuity of data relevant to parents — more so to policymakers interested in macro-quantitative outputs — ACARA scaled adjustments that, in the McGaw review's words, "obliterated any real change that might have occurred". Perelman and Haney judged the modes "inherently incompatible"; ACARA disputed this and sustained the paper-era scale across successive adjustments until 2022, the last dual-mode year.
Then the adjustments that were supposed to equate the modes were discarded in a 2023 reporting reform that replaced ten bands and the national minimum standard with four proficiency levels, announced by media release.
So the continuity that justified the equating was abandoned once the equating was no longer needed.
The judgement
ACARA's chief executive has maintained that "the purpose has never been diagnostic assessment" — a position recorded in Wilson and colleagues' 2021 review. Perhaps not. But the uncapitalised digitalisation represents a jointure between federal educational policy and the educators delivering education. Without the promised precision that would see NAPLAN as a productive diagnostic tool at a classroom level, construction became less interrogable, while consequences remained fully enforceable.
That is the sentence a parent should carry away. The test is harder to inspect than it used to be. What it does to your school is exactly as consequential as it ever was.
Since the government has been able to weather the discontent of unions and scholars discrediting any use for NAPLAN itself, the digitalisation didn't have to succeed in anything but digitisation. The appetite for digitalisation had effectively dried up.
NAPLAN Online delivered a digitalised instrument through a digitised service while institutionalising nothing that would adversely affect the marketisation of school comparison. The reform entailed an efficient rehoming of a national assessment platform — notwithstanding the nationwide disruption of 14 May 2019 — but ultimately left unattended the appetite for newly enriched data, to empower commonwealth policy and teachers whose practice was subordinated to NAPLAN preparation. By concentrating the benefits of reform in fine-grained data furthest from where it is deployed to public value, the governance chain secured NAPLAN as the insulated connective between education policy and neoliberal accountability mechanisms.
The instrument got better. Nobody was given the job of noticing.