Skip to content

Science1 publisher3 min readPublished

Australia's 72-point PISA science gap narrowed by 16 points since 2022

Australia's reported PISA gaps are accurate averages of two different populations, and the trend lines inside them move the other way. A matched study of NAPLAN data locates schools the averages miss.

The Scientist · Science desk

Illustration accompanying Australia's 72-point PISA science gap narrowed by 16 points since 2022

What happened

  • The latest PISA round put Indigenous students in Australia 72 points behind non-Indigenous students in science, 76 behind in reading and 74 behind in maths.
  • Within those same results, the average Indigenous science score rose 16 points since 2022 while the non-Indigenous average did not move.
  • Anderson and colleagues' 2022 synthesis of 51 studies of Australian schools found a pervasive lack of teacher readiness, curriculum readiness and school culture readiness to teach Indigenous students.

Compiled by The ScientistSomething wrong?How this is made

Why it matters

  • decision A pooled gap statistic cannot identify which school moved its students, and program funding decided on gap reporting is decided on that statistic.
  • exposure Four in five Indigenous test-takers sit in the bottom two socioeconomic quarters, so the reported gap is entangled with socioeconomic position. Read as a measure of student capability, it is a composition effect.
  • capability The matched within-group comparison ran on ten years of NAPLAN data that already exists, so education systems can ask which schools improved outcomes without commissioning new testing.
  • constraint PISA scores against the OECD's definition of applying knowledge to real-life challenges. Against that definition, instruction in an Indigenous language registers no attainment, however well it is delivered.

The 72-point science gap and the 16-point rise come out of the same dataset, and putting them together fixes the 2022 gap at about 88 points [1][3][15]. The narrowing since then is roughly a fifth of where the gap started [15]. PISA averages sit near 500 points [2], so the 16-point move is about 3% of the scale [21].

Whether the same population improved is a separate question. PISA draws a fresh sample of 15-year-olds each cycle [7], so a subgroup average can shift when the composition of the subgroup shifts. The analysis reports the point differences without standard errors or the size of the Indigenous sample [18]. In reading and maths the pattern is different again: Indigenous scores held steady while non-Indigenous scores dipped [4], so the narrowing there came from the non-Indigenous side falling [19].

The 2024 study led by Peter Anderson changed the comparison group. A pooled comparison asks whether two populations score differently, and when they start from different places it answers yes every time. Matching Indigenous students to Indigenous peers who shared similar starting points asks whether a given school moved its students. Across a decade of NAPLAN data, some schools and systems showed real, sustained gains that the conventional comparison had missed [5][6].

The starting points are documented in the PISA data itself. Eighty per cent of the Indigenous students who sat the most recent test came from the lowest two socioeconomic quarters, and half from the lowest quarter alone [8]. The analysis attributes the difference to access and resourcing, not aptitude or ability [9]. That is a claim about confounding, and it is the reason a raw gap between two populations fails as a measure of what students can do.

PISA's construct belongs to the OECD: how well 15-year-olds apply what they know to real-life challenges [7]. The analysis proposes a different benchmark, drawn from the UN Declaration on the Rights of Indigenous Peoples. The General Assembly adopted the declaration in September 2007, and Australia voted against it before endorsing it in April 2009 [12]. Article 14 sets out the right of Indigenous peoples to establish and control their own educational systems and to teach in their own languages and cultural methods [13]. The questions that follow from it are answerable: whether Indigenous languages are resourced as languages of instruction, and whether Indigenous knowledge is in the curriculum as a distinct and equally valid form of knowledge [20]. Studies cited in the analysis find Australia has not adequately incorporated those rights into its education system [17]. Anderson and colleagues' 2022 synthesis of 51 studies found a pervasive lack of teacher readiness, curriculum readiness and school culture readiness to teach Indigenous students [11].

What to watch

  • Whether the national PISA report publishes standard errors and Indigenous sample sizes for the latest cycle, which is what would separate the 16-point science move from sampling noise.
  • Whether the next PISA cycle repeats the direction of the Indigenous science result or reverses it.
  • Whether state education departments adopt matched within-group NAPLAN comparisons in their public reporting alongside the gap figures.
Loading claim ledger
Loading source directory links
Loading share composer
Loading topic controls
Loading related stories