Our evidence base — and how we enforce it.
An AI is a probability machine; it has no idea what helps a child learn. So we constrain it with research-informed principles — and, unlike a reading list, each one is tied to something we actually check in the generated resource. We also say plainly what the evidence says to avoid.
Explicit instruction over minimal guidance
Novices learn best when concepts are clearly explained, modelled and practised — not discovered unaided.
How we apply it
- State the skill plainly and model it before asking students to do it independently.
- Sequence from easier to harder; don't open with the hardest item.
How we check it
- Flag tasks that expect a novice to discover a procedure with no prior modelling or worked example.
Honest limit: Explicit instruction is most important for novices; as students gain expertise, heavy scaffolding can become redundant.
Manage cognitive load
Working memory is limited — chunk content, keep one new idea per step, and avoid clutter.
How we apply it
- Introduce one new idea at a time; keep instructions short and unambiguous.
- Pair a visual with the prompt only when it reduces load (don't decorate).
How we check it
- Flag questions that combine several new demands at once for the year level.
Honest limit: Load is relative to the learner's prior knowledge; the same task is heavier for a novice than an expert.
Worked examples with faded support
Show a fully worked example, then gradually hand over the steps to the student.
How we apply it
- Where a procedure is taught, include a worked example before independent practice.
- Fade support across items: model → guided → independent.
How we check it
- Flag a procedural practice set that gives no worked example or scaffold for the first item.
Honest limit: The worked-example effect weakens as expertise grows (expertise-reversal).
Retrieval practice (low-stakes recall)
Recalling information strengthens memory far more than re-reading it.
How we apply it
- Favour questions that require students to generate an answer from memory over recognition where appropriate.
- Include a short warm-up that retrieves prior learning when it fits the topic.
How we check it
- Prefer generative recall items; flag a set that is entirely copy-from-the-page recognition when recall is feasible.
Honest limit: Retrieval is most powerful with feedback and spacing; a single quiz in isolation is weaker.
Spaced & distributed practice
Spreading practice over time produces more durable learning than massing it.
How we apply it
- Where natural, revisit a prior concept alongside the new one rather than only the latest topic.
How we check it
- (Advisory) Note when a revision set could interleave a previously taught concept.
Honest limit: Optimal spacing depends on the retention interval; there is no single best gap.
Interleaved practice for discrimination
Mixing problem types teaches students to choose the right method, not just execute it.
How we apply it
- For revision sets, mix related problem types rather than blocking all of one type together.
How we check it
- (Advisory) Note a blocked practice set that would benefit from interleaving related types.
Honest limit: Interleaving helps discrimination tasks most; for a brand-new single skill, blocked practice first can be appropriate.
Elaboration & self-explanation
Asking 'why' and 'how' deepens understanding beyond surface recall.
How we apply it
- Include at least one prompt that asks students to explain their reasoning, not only state an answer.
How we check it
- Flag a set pitched at Years 3–6 that contains no 'explain why / how do you know' prompt at all.
Honest limit: Explanation prompts work best for students with enough prior knowledge to reason from.
Activate prior knowledge
New learning sticks when connected to what students already know.
How we apply it
- Open by connecting to a familiar idea or everyday example before introducing the new concept.
How we check it
- (Advisory) Note when an explainer/lesson dives into new content with no link to prior knowledge.
Honest limit: The link must be accurate; a misleading analogy can create misconceptions.
Built-in feedback
Practice improves learning most when students can check and correct their answers.
How we apply it
- Provide a complete, correct answer key so students/teachers can give immediate feedback.
How we check it
- Flag any verifiable question whose answer key is missing or not independently checkable.
Honest limit: Feedback helps only if it is acted on; an unread answer key changes nothing.
Systematic phonics in the early years
Early reading is built on explicit, systematic phonics and decodable text.
How we apply it
- For F–2 English reading, keep vocabulary controlled and decodable; introduce only a few high-frequency irregular words.
How we check it
- Flag F–2 reading text that relies on long, irregular or low-frequency words a beginning reader couldn't decode.
Honest limit: Decodability is one dimension; comprehension and vocabulary still matter and aren't guaranteed by decodable text alone.
Concrete → representational → abstract
In maths, move from materials to diagrams to symbols as understanding grows.
How we apply it
- Use a concrete/representational visual (ten_frame, array, fraction_bar, number_line, bar_model) to support a new abstract idea, especially in F–4.
How we check it
- Flag a new F–4 maths concept presented only as bare symbols with no supporting representation.
Honest limit: The aim is to fade to the abstract; staying concrete forever is not the goal.
Explicit vocabulary teaching
Key terms are taught directly, in context, and used by students.
How we apply it
- Introduce subject vocabulary in context and give students a chance to use each term.
How we check it
- (Advisory) Note unexplained technical vocabulary that the year level is unlikely to know.
Honest limit: A word list alone is weak; terms need to be used to be learned.
Standards-referenced A–E judgement
Assessment is judged against the achievement standard, with distinct A–E qualities — not points.
How we apply it
- When a marking guide is requested, write one criterion per content descriptor with genuinely distinguishable A–E qualities of student work.
How we check it
- Flag A–E levels that aren't genuinely distinguishable, or a criterion citing a descriptor not selected for the task.
Honest limit: A well-formed guide still requires teacher moderation; it does not replace professional judgement.
We reject 'learning styles'
Matching teaching to a supposed visual/auditory/kinaesthetic style does not improve learning.
How we apply it
- Differentiate by readiness and prior knowledge — never by an alleged 'learning style'.
How we check it
- Flag any content that classifies students by learning style or tailors a task to one.
Honest limit: Rejecting learning styles is not rejecting variety — multiple representations of the same content are still valuable.
References
- Kirschner, P. A., Sweller, J., & Clark, R. E. (2006). Why minimal guidance during instruction does not work. Educational Psychologist, 41(2), 75–86.
- Australian Education Research Organisation (AERO). Explicit instruction — practice guides and the 'how students learn best' model of learning. (CC BY 4.0)
- Rosenshine, B. (2012). Principles of instruction: Research-based strategies that all teachers should know. American Educator, 36(1), 12–39.
- Sweller, J. (1988/2011). Cognitive load theory. (Developed at UNSW, Australia.) In Psychology of Learning and Motivation, 55, 37–76.
- Renkl, A. (2014). Toward an instructionally oriented theory of example-based learning. Cognitive Science, 38(1), 1–37.
- Dunlosky, J., et al. (2013). Improving students' learning with effective learning techniques. Psychological Science in the Public Interest, 14(1), 4–58.
- Roediger, H. L., & Butler, A. C. (2011). The critical role of retrieval practice in long-term retention. Trends in Cognitive Sciences, 15(1), 20–27.
- Cepeda, N. J., et al. (2006). Distributed practice in verbal recall tasks: A review and quantitative synthesis. Psychological Bulletin, 132(3), 354–380.
- Australian Curriculum v9.0 (ACARA) — English, phonics and word knowledge; systematic synthetic phonics in the early years. (CC BY 4.0)
- Queensland Curriculum & Assessment Authority (QCAA) — Guide to making judgments and standards-referenced A–E assessment. (CC BY 4.0)
- Willingham, D. T., Hughes, E. M., & Dobolyi, D. G. (2015). The scientific status of learning styles theories. Teaching of Psychology, 42(3), 266–271.
- Australian Education Research Organisation (AERO). Tailoring teaching to 'learning styles' is not supported by evidence. (CC BY 4.0)