There’s no shortage of explanations for the ‘boy problem’. They don’t read enough. They rush. They don’t revise properly. They need shorter tasks. They need more competition. They need texts they can relate to. None of these ideas are implausible. They might even be right about boys in general but they’re certainly not true of every boy (just as they might also be true of many girls!)

We do what we always do when we see a pattern: we start explaining it. The trouble is that our explanations bypass other sources of evidence. Before we’ve considered attendance, prior attainment, reading age, class allocation, teacher absence, behaviour, curriculum time, option choices, or whether the difference is meaningful once other variables are considered, we default to the most obvious cause: there’s something about gender that causes boys to do less well than girls in school.

Whenever the ‘boy problem’ is raised, someone is bound to suggest a “boy-friendly” curriculum: more action, more choice, more movement, less writing, greater relevance. Of course, no one wants boys drifting through school bored, alienated and underperforming but good intentions don’t translate stereotypes into causes. We move from observation to cause, and from cause to policy, almost without noticing. The graph shows an effect; we think we can see the cause.

In the 1940s, the Belgian psychologist Albert Michotte showed how easily human beings mistake sequence for causality. In The Perception of Causality, first published in French in 1945, he demonstrated that very simple visual sequences create the appearance of causal connection. If one shape moves, touches another, and the second shape moves away we don’t merely see one event followed by another, we see one thing causing another thing happen.

Causality cannot be directly observed, only inferred. Yet some sequences are so compelling that our inferences feel like perception. We we see something happen we believe we can see its cause. Classrooms produce the same illusion. The behaviour in Miss Crumb’s class is unruly, so Miss Crumb must be ineffective. Gavin doesn’t complete his work, so he must be feckless and work-shy. Parvinder hands everything in on time, so she must be conscientious. We may sometimes be right, but the ease with which we jump to conclusions should worry us. In reality, life is dense with competing causes: prior knowledge, relationships, routines, attendance, sleep, anxiety, curriculum, peer status, home life, task design, teacher absence, reading fluency, classroom norms, health, habit, memory and chance. Because this is all a bit messy and hard to make sense of, w e prefer simpler stories.

This is how poor reasoning often works in schools. We begins with something visible: a gap, a pattern, a graph, a behaviour, a lesson that went badly and then we add in the most convenient, plausible explanation because it fits what we already believe. Once an explanation feels plausible, it becomes true. From there it’s a just a hop, skip and jump to policy.

Once we’ve settled on a cause, we’re quick to make moral judgements. In A Treatise of Human Nature, published in 1739-40, David Hume noticed that writers often begin by describing how the world is, then suddenly start telling us how it ought to be, without explaining the move. One minute they’re making factual claims; the next they’re making moral ones. The grammar changes from “is” to “ought”, but the argument supplying that change has gone missing.

Hume wasn’t saying facts are irrelevant to moral reasoning but that facts don’t contain values inside them. A description of the world doesn’t, in itself, tell us waht we should approve, resist, tolerate or change. To reach an injunction, we must add a value judgement. We have to say what we care about.

Boys are more restless, therefore school should be redesigned around restlessness. Some children arrive in school with lower prior attainment, therefore academic expectations must be lowered. A pattern exists, therefore it must be accepted, accommodated and declared inevitable. But the existence of a tendency doesn’t tell us how to respond to it. Reality offers constraints but doesn’t give us commandments.

One version of this mistake is the naturalistic fallacy: because something exists, or appears natural, it must be desirable, inevitable or right. In Principia Ethica, philosopher, G. E. Moore objected to attempts to define “good” as some natural property such as pleasure, usefulness, survival, evolutionary success or social approval. Even if something is pleasurable, adaptive, efficient, popular or widely desired, we can still ask whether it is good. Natural facts can inform our moral reasoning, but they can’t reason for us.

In education, “natural” can so easily become an excuse. Children are naturally curious, so direct instruction must be deadening. Adolescents naturally test boundaries, so firm discipline must be unrealistic. Boys are naturally more active, so schools must become more “boy-friendly”. Claims begin as observation, but harden into policy.

Another version of the same mistake runs the other way. Because something ought to be true, we convince ourselves that it is true. This is the moralistic fallacy. It feels nobler than the naturalistic fallacy because it usually begins with a decent moral impulse. We want equality, so we deny difference. We want parenting to determine children’s futures, so we resist evidence suggesting its effects are more complicated. We want schools to overcome disadvantage, so we treat any persistent gap as proof of failure. We want merit to be rewarded, so we overstate how much success is under an individual’s control and how much is down to luck.

The problem isn’t our the values — equality is worth defending, parents do matter, schools should make a difference and merit should count — but when we allow values to settle empirical questions in advance. If a finding threatens our preferred moral story, we don’t get to make it disappear by disliking its implications. Denial leaves us designing interventions for a reality that doesn’t exist.

Both fallacies are ways of avoiding difficulty. Where the naturalistic fallacy turns reality into an instruction, the moralistic fallacy turns morality into a filter on reality. One says, “This is how things are, so this is how they should be.” The other says, “This is how things should be, so this is how they must be.” Sadly, the world is not always as we might wish it was, and the way it is doesn’t decide what we should do.

We can believe children should be creative, independent, collaborative and self-regulating, but those aspirations don’t tell us how novices learn. We can believe all children should achieve highly, but that doesn’t mean all children arrive at school eually likely to acheive.

The OECD’s Building Strong Foundations for Life offers a useful recent example. Its 2025 International Early Learning and Child Well-being Study reports that, at age five, girls outperform boys across the three broad dimensions assessed: foundational learning, executive function, and social and emotional development. The executive function chart is especially arresting. In country after country, girls are way ahead of boys.

It’s easy to look at this and think we’ve seen the cause. Five-year-old girls have better executive function than boys. Therefore boys are less ready for school. Therefore boys need a different kind of early years provision. Therefore settings should become more boy-friendly: more movement, more play, more choice, more competition and more activities designed around what boys supposedly are.

But this is correlation not causation. It tells us that girls, on average, score higher on particular measures in particular jurisdictions but not whether the gap is biological, maturational, cultural, pedagogical, artefactual, or some tangled combination. It also reports that 36% of variation in foundational learning scores and 22% of variation in executive function and social and emotional development scores occurs between early childhood education and care centres or schools. Clearly, context has a lot to answer for.

Executive function sounds like a thing children possess in greater or lesser quantities, but it’s something we can never observe directly. We infer it from performance on tasks: waiting, switching, remembering, inhibiting, complying, attending. A five-year-old’s score may reflect cognitive capacity, language, familiarity with adult-led tasks, sleep, confidence, socialisation, or willingness to please an unfamiliar adult. The scores are real but their meaning is far from self-evident.

This is not to say there isn’t a gender gap — there clearly is and to deny it would be to fall into the moralistic fallacy — but we should be cautious about what it represents. A measured difference isn’t an explanation. We can say that girls, on average, scored higher on certain measures at age five. We can’t leap from that to confident claims about what boys and girls are naturally like, or what schools should do in response.

Read more

It was a privilege to be asked to serve as lead drafter for the English and Drama curriculum framework for Northern Ireland. It was also a rare pleasure to work with the knowledgeable and passionate teachers, school leaders and experts who made up the English and Drama Subject Working Group. I’m immensely proud of what we’ve created. No document is ever perfect, and few are likely to please everyone affected, but this one captures my beliefs about what the study of English can be.

Curriculum drafting is an act of stewardship. It asks us to think carefully about what young people are entitled to know, experience and be able to do. It asks us to balance inheritance with possibility: the literature, language, stories, arguments, traditions and forms that have shaped us, and the knowledge that will allow the next generation to take up that inheritance with confidence, judgement and joy.

My hope is that the framework will have a powerful and positive impact on the teaching of English across Northern Ireland. I hope it helps teachers think with greater clarity about the structure of the subject. But, most of all, my hope is that it helps more children experience English as an invitation into language, literature, thought and culture.

Here is our vision for English, taken from the TransformED Northern Ireland Curriculum 2028.

Vision Statement for English

English is the study of language both as a means of living and as a tool for human thought. A high-quality English curriculum allows pupils to participate fully in the world and fosters their sense of belonging within it. Through accurate and fluent speaking and writing, pupils can communicate ideas, experiences and emotions. Through confident reading and attentive listening, pupils not only allow others to communicate successfully with them, they enter the wider conversation in which literature records and reshapes human experience. The study of English becomes a gateway to deep wells of thought and experience beyond their own.

A strong English curriculum lays the foundation for a lifelong habit of reading across many forms and genres. The intensely personal pleasure of immersion in the flow of well-crafted prose makes a text a place to live inside and where a pupil makes meaning at their own pace. Whether through the narrative tension and fully imagined world of a novel or through an intricately crafted argument, a literary text captivates the reader’s mind. Literature gives space for beliefs, emotions and relationships to be examined with care. Through the rich range of specific and connected texts that they study in school, through the passion of their teachers from a lifetime of reading and discussing literature, and through the lasting desire to keep reading that these experiences form, pupils are given one of the most important means that humans have ever produced for personal development. As individuals, as members of reading communities and in wider society, they grow culturally, emotionally, socially and spiritually.

English thus extends beyond practical utility. Pupils encounter a rich and varied diet of whole texts that require patience, sustained attention and informed interpretation. With support to build stamina for extended reading, they learn to follow argument, trace narrative and recognise patterns, fostering their own creativity as they learn from and adapt stylistic and structural conventions. Wide reading reveals how writers connect across time, genre and tradition. Through knowledge of context and intertextuality, pupils learn that stories do not stand alone but belong to a shared cultural landscape in which ideas are inherited, contested and refined. Through encounters with literature from different periods, cultures and traditions, pupils gain the knowledge and experience to give them genuine choices and informed judgement.

Vocabulary runs through every aspect of the subject. Words function as conceptual tools that enable pupils to make finer distinctions, perceive pattern and hold more complex ideas in mind. As vocabulary expands, so does the reach of thought. For this reason, vocabulary, background knowledge and context are taught deliberately. This matters as much in early reading as in later study, since without it, pupils cannot make sense of texts or the ideas they contain.

A cumulative approach shapes the teaching of reading, writing and spoken language. Reading fluency and disciplinary knowledge must be built step by step. In Foundation and Key Stage 1, this means that phonological awareness and decoding ability are priorities until automaticity, accuracy and prosody are achieved by all. Success in both spoken and written language requires early and regular participation in rich sound-worlds – poems, rhymes and songs, memorable folktales and stories – with plentiful oral recitation and rehearsal as well as enjoyment of the social and emotional benefits of discussing them. It requires practice in purposeful listening and clear contribution. Writing depends on explicit teaching of handwriting, syntax, spelling, sentence control, vocabulary, structure and argument. Together, these strands enable pupils to communicate with increasing confidence and precision.

The aim of English is to give young people an entitlement to:

• knowledge that helps them understand the world, including knowledge of letter and print that renders the world of text accessible

• language that helps them express ideas with precision

• literature that illuminates human experience

• habits of attention that support thoughtful and creative engagement.

In this way, English prepares pupils not only for academic success but also for participation in the ongoing conversation of humankind, understanding the traditions they inherit and contributing to their future. Reading for pleasure emerges as a long-term outcome of a curriculum that builds the competence, confidence and shared cultural entitlement required for all aspects of English to become genuinely enjoyable.

If this piques your interest to review the substance of the curriculum, you are most welcome to download and review it. The curriculum is currently in its public consultation phase and, should you have any comments to make, you can do so here.

Fire is one of humanity’s oldest and most essential technologies. By warming homes, cooking food, hardening clay, and later sterilising equipment and powering engines fire kickstarted civilisation. But it also burns down buildings.

Long before there were cities, schools, books or wheels, our ancestors were learning how to use flame. Traces of burned bone and ashed plant remains at Wonderwerk Cave in South Africa suggest that hominids were using fire around a million years ago. At Gesher Benot Ya’aqov in Israel, burned seeds, wood and flint fragments point to controlled use of fire nearly 790,000 years ago. Later evidence suggests not just burning, but cooking, tool-making, protection and social life organised around the hearth. Fire changed what we could eat, where we could live, how long we could stay awake and what we could make.1

To be useful, fire must be controlled. It requires attention, boundaries, tools, habits, knowledge and rules. A hearth left unattended might burn your house down; a kiln not properly maintained will destroy the pots it was meant to harden; giving a child a box of matches is asking for trouble. Out of control, fire is terrifying: indiscriminate, hungry, impossible to reason with, as wildfires show when they turn forests and towns into ash.

No-one is pro-fire or anti-fire. We’re pro-hearth and anti-arson. In a kiln, a furnace, or a carefully managed campfire, fire is wonderful. But obviously, no one wants it in the curtains, the wastepaper bin or the roof space. The question isn’t whether fire is useful but how it’s power harnessed and how its risks are mangaged.

We need to think about AI in school in the same way by working out how we control it and how we manage its risks. For some, AI is about to democratise expertise, personalise learning, solve workload and make every child’s education infinitely adaptive. For others, it’s a cheating engine, a hallucination machine and an existential threat to thought itself. Both responses contain some truth, but both are too simplistic.

AI will make some things in schools better, some things worse and quite a lot of things merely faster. That last possibility should worry us more than it does. Schools are already full of tasks that can be completed without anyone thinking very much. Just as students can copy notes without understanding them, teachers can produce resources that look attractive but don’t teach anything particularly well. Likewise, leaders can generate policies, reports and improvement plans which look impressive without addressing the reality of classrooms.

AI hasn’t created these problems — they stem from all-too-human desires to save time, avoid effort and look good — but it has accelerated them. And, as with fire, pouring accelerant over a poorly controlled system is never a good bet. It’s pointless to discuss whether AI should be allowed into education — it’s here, baby! The real question is whether we can control it and what we need to guard against

The findings that follow are a sequence of warnings about the consequences of AI on learning.

1. Students know there’s a cost to using AI but use it anyway

2. Students who use AI best need it least

3. The best AI tutors are deliberately obstructive

4. AI can strengthen participation or replace it

5. The better AI gets, the more human judgement matters

Taken together, these findings suggest a simple test. AI is useful when it helps students think more clearly, practise more deliberately and become less dependent over time. It’s dangerous when it makes weak understanding look strong, makes effort feel pointless, or allows adults to outsource decisions they ought to own.

1. Students know there’s a cost but use it anyway

Discovering that use use aof AI is endemic amongst students is about as surprising as discovering that they dress badly or wash infrequently. In 2025, Oxford University Press reported that around 80% of 13- to 18-year-olds said they used AI in their schoolwork, while only 2% said they didn’t use it at all. More interestingly, many of them worried that AI might be damaging their learning and skill development. The survey reports, “Nine in ten students say AI has helped them develop a skill—problem solving, idea generation, and exam prep top the list. At the same time, 62% report negative effects, such as reduced creative thinking and over-reliance.”

• 62% reporting concern that AI reduced creative thinking and over-reliance

• 60% worry AI encourages copying over original thinking.

• 51% fear it may reinforce bias or stereotypes.

• 48% suspect peers are secretly using AI to complete schoolwork.

• 47% worry teachers can’t detect it.

• 47%, worry they can’t identify accurate AI-generated information.

• 32% admit they can’t tell if AI content is true

In other words, students appear to understand the bargain. AI helps with problem solving, idea generation and exam preparation, but they also worry that it encourages copying, weakens creative thinking, fosters over-reliance and makes truth harder to judge. That is a much more interesting picture than the usual story of students either cheating shamelessly or embracing the future with innocent enthusiasm.

Think of the student who starts with a reasonable request: explain this question. Because the explanation is helpful, they ask for a few ideas. Then they ask for a better opening sentence. Then they ask for the whole paragraph to be rewritten. Each step is only a little nudge further from independent thought. By the time the work is submitted, the student hasn’t cheated in the old-fashioned sense, but nor have they done the thinking the task was meant to provoke.

Schools already incentivise this. We tend to reward completion more than we do thinking, polish more reliably than struggle, and compliance more immediately than learning. AI fits beautifully into that system because it manufactures the appearance of effort. It makes students seem productive, capable and conscientious, even when the mental effort has been outsourced.

Of course, we shouldn’t be too smug about this. We use satnav and lose our sense of direction. We use autocorrect and become less certain about spelling. We save phone numbers so we don’t have to remember them. Offloading is a normal human response to convenience. The different with AI is that the offloading reaches into explanation, judgement, expression and thought itself.

A student doesn’t need to be lazy to use AI badly. They only need to be under pressure, short of time, unsure what good looks like, and surrounded by tasks where the submitted product matters more than the thinking that produced it. The cheating panic catches only the most obvious abuse. The deeper problem is dependence disguised as efficiency.

2. AI most helps those who already know most

Read more

The usual explanation is comforting: someone must have failed. Leaders weren’t strong enough, teachers weren’t consistent enough, students weren’t motivated enough, and parents weren’t supportive enough. But what if the more troubling possibility is also more useful? What if schools are systems that can produce poor outcomes even when most of the people inside them are well-intentioned, hardworking and competent?

We know from ordinary life that intentions and outcomes fail to match. Someone tidies your desk and you can’t find a vital document. A friend organises a party in your honour and creates an evening you dread. A policy designed to help people ends up making their lives harder.

History is littered with larger, grimmer examples: the Treaty of Versailles, Prohibition, the Partition of India, the War on Drugs and No Child Left Behind. Each was, at least in part, an attempt to solve a serious problem. Each created consequences its architects either didn’t foresee or couldn’t control. Good intentions pave the road to hell. But why are we so surprised that good intentions are so obviously insufficient?

In schools, the gap between intention and outcome is often hidden because the intentions are so virtuous. Who could object to attempts to improve attendance, raise standards, keep children safe, develop reading, reduce workload or close gaps? But virtuous aims can still create vicious incentives. An attendance policy can improve the numbers while making school feel more hostile to the families most in need of trust. A data system can promise clarity while generating hours of low-value labour. A behaviour policy can create calm corridors while pushing the most vulnerable children closer to exclusion. None of these things happen because people want them to happen but because systems convert aims into incentives, routines and consequences.

A common response is to assign blame. If we could just remove the incompetent or malicious, all would be well. This is the logic behind school inspection: if a school is underperforming, its leaders must be the problem. Expose them, replace them, and improvement will follow. But this isn’t just wrong, it’s foolish. Replace the leader and the same pressures remain: recruitment problems, unstable budgets, weak local services, inherited curriculum gaps, hostile accountability, fragmented data, exam incentives and the daily weather of human need. Sometimes new leadership helps enormously, but only when it changes the conditions in which people work.

There may be a few schools led by tyrants or incompetents, but most schools, regardless of outcomes, are led by people doing their best. Leading a struggling school — especially one in a deprived area — is punishingly hard. Success and failure are always influenced by luck. Punishing the unlucky for failing to control every variable deters others from taking on such roles.

Good intentions aren’t enough but they do matter. If we assumed that well-intentioned leaders could be supported, mentored, and developed into effective ones, we might create a system with higher morale, less burnout, and better outcomes, especially for the most disadvantaged students.

To do this, we must move away from a culture of blame and towards one of responsibility. Schools are complex networks, not only of people, some of whom may unintentionally act against their own best interests, but also of policies, specifications, software, and other interrelated elements. This complexity means parts will always misalign. Schools aren’t tidy jigsaw puzzles; they’re often cobbled together with string. Sometimes literally.

Leaders have to juggle all of this while also responding to external pressures: government mandates, media scrutiny, societal crises, emerging technologies. It’s inevitable that schools will sometimes struggle. But while unintended outcomes aren’t always a leader’s fault, they’re still their responsibility.

All of which is a long preamble to Stafford Beer’s famous maxim: “The purpose of a system is what it does.” POSIWID, for short. Beer argued that systems should be evaluated on their observable results. A system’s true purpose is revealed by its consistent outputs, regardless of what its designers claim.

No one intends for a school to produce poor outcomes. But if it consistently does, then intention is functionally irrelevant. Focusing on what the system does, rather than what it should do, helps us address discrepancies between aims and outcomes.

Applying the principles of POSIWID to Ofsted, England’s school inspectorate, reveals something about these discrepancies in the inspection process.

What are Ofsted’s stated aims?

• “To improve lives by raising standards in education and children’s social care”

• To conduct inspections that help services improve

• To apply proportionate regulation focused on risks

• To inform policy and practice through inspection insights

• To adapt to ensure relevant and effective oversight

• To be transparent and engage stakeholders

To what extent are these aims fulfilled in practice?

What does Ofsted do?

• Publishes a rubric for judging school performance

• Conducts two-day inspections assessing leadership, behaviour, curriculum, and safeguarding

• Publishes inspection reports with brief recommendations on how to improve on future inspections

• Releases subject reviews and clarifies inspection practices¹

• Shapes school priorities - schools align teaching and assessment with what they believe Ofsted wants

• Encourages performative compliance - schools prepare evidence to impress inspectors

• Promotes high-stakes accountability - risk aversion, exam focus, and data obsession follow

• Increases workload and stress - inspection anticipation spikes pressure, sometimes with tragic outcomes

• Constrains curriculum innovation - schools avoid novel approaches that might not align with Ofsted frameworks

• Reinforces system hierarchies - inspections influence parental choice and school reputation

• Shapes public perception - inspection grades serve as shorthand, often lacking nuance

• Fuels a consultancy industry advising schools on how to pass inspections. Though diminished, it still exists.²

Each of these can reasonably be argued to be consequences of Ofsted’s activities even though many of them are unintended. The mechanism is depressingly familiar. Once inspection carries high stakes, schools begin to optimise for what can be seen, evidenced and defended. Curriculum intent statements, book looks, data drops, lesson routines and behaviour scripts become safer than the more elusive business of finding out what students actually know and can do. The proxy becomes the target. The target becomes the work. Eventually, schools are not merely preparing for inspection; they are being reorganised by the anticipation of it.

What does Ofsted not do?

• Improve student outcomes directly — this is the work of those who work in schools

• Close performance gaps — despite intentions, these remain stubborn or widen

• Make schools happier, safer places — while safeguarding is important, cliff-edge judgments likely don’t help

These repeated outcomes reveal Ofsted’s true functional purpose. Whether intended or not, inspection arguably distorts the system more than it improves it.

Ofsted presents itself as the gold standard for school evaluation. But through the lens of construct validity, problems emerge. Construct validity asks whether a test genuinely measures what it claims to. Although inspectors believe they measure “quality of education,” a vague, complex construct, instead, they assesses proxies: lesson snapshots, curriculum maps, brief conversations with leaders, teachers and students etc. But can we meaningfully assess the quality of education a school provides in two days?

When inspections are supposed to reveal something as complex and long-term as the effectiveness of a school, we should ask: do Ofsted ratings correlate with other meaningful indicators? The answer is, at best, inconsistently. While high Ofsted grades sometimes align with good academic outcomes, the relationship is far from perfect. Blunt, high-stakes judgments are antithetical to consistent and valid measures of actual learning. Ofsted too often relies on unreliable proxies like short-term outcomes or performance during inspections, which tell us little about what pupils have genuinely learned or retained. This exposes the emperor’s new clothes: what passes for accountability is often little more than data theatre. If we’re serious about improving education, we need to stop mistaking performance for learning and start measuring what really matters.

Instead, too many inspections end up evaluating whether schools are good at being inspected. That is not the same thing as being good at educating. We mistake performance for learning. Worse still, inspection outcomes skew against disadvantaged schools. These schools are less likely to receive ‘Good’ or ‘Outstanding’ ratings, even after adjusting for prior attainment. If judgments track pupil intake more closely than teaching quality, we’re evaluating demographics, not education. This is a failure of discriminant validity: inspections measure the wrong things. A school might be good at being inspected but poor at educating. That’s a dangerous confusion.

To compound this, Ofsted’s impact is hardly neutral. High-stakes inspection distorts practice. Jane Perryman paints a bleak but familiar picture: under the watchful eye of inspectors, schools stop being places of learning and become stages for performance. The inspection regime, she argues, doesn’t just evaluate schools, it reshapes them. Teachers begin to teach not for understanding, but for optics. Leaders write policies for show, not for substance. In essence, schools are being judged on how well they simulate the illusion of effectiveness. It’s not about what works, it’s about what looks like it works. Perryman’s work reveals the corrosive impact of high-stakes accountability: it warps priorities, distorts identities, and rewards surface over depth.

None of this is to say that accountability is unimportant. Inspection exists because schools are public institutions entrusted with children’s lives and futures. Parents deserve information, governments need assurance and children should not have to wait years for adults to notice that a school is failing them. The case for accountability isn’t trivial, and those who oppose Ofsted too glibly can sound as if they’re really opposing scrutiny. The problem isn’t that schools should be answerable for what they do but that the current model confuses public judgement with useful feedback, and too often rewards the appearance of quality over the slow, difficult work of improvement.

But if we are to hold schools accountable, we need to be clear about what we’re holding them accountable for, and how we’re measuring it. If our tools can’t validly and reliably do that, then they’re not just ineffective, they’re dangerous. A poorly calibrated system of accountability risks incentivising the wrong behaviours, punishing the wrong schools, and rewarding performativity over substance.

In truth, Ofsted does achieve some of its aims: public visibility, pressure to improve, and a semblance of consistency. But if we’re serious about the quality of education rather than the appearance of it, then we need an accountability system with far stronger construct validity. Otherwise, we risk mistaking noise for signal and performance for learning. For all its good intentions, inspection has become a system that too often rewards the management of appearances.

What might a better system might look like?

Read more

That is the basic idea behind affordances. An affordance is the environment’s invitation to act: the possibilities it opens, the actions it makes available, the behaviours it seems to ask of us. Sometimes this is straightforward: a handle affords pulling, a chair affords sitting, a path affords walking. Sometimes it’s less clear: a low wall might afford sitting to one person, climbing to another and an obstruction to someone else. Affordances are relationships between people and environments, not properties that objects simply possess.

James J. Gibson, the psychologist most associated with the term, defined it like this:

The affordances of the environment are what it offers the animal, what it provides or furnishes, either for good or ill. ... It implies the complementarity of the animal and the environment

He argued that we don’t perceive a neutral world to which we later attach meaning, but one already organised around possible action. We see objects as climbable, reachable, passable, graspable, avoidable, usable or dangerous. A surface isn’t just flat, brown and extended; it may be something we can walk on. A gap isn’t just empty space; it may be passable for one body but impassable for another. rather than waiting for us to decode it, the world makes some routes inviting and others daunting; it encourages some choices and makes it more difficult to select others.

Gibson was one of the central figures behind ecological psychology, a tradition that tried to put perception back into the world rather than treating it as something sealed inside the head. Much twentieth-century psychology tended to imagine perception as a kind of internal reconstruction. The world sends us fragments of sensory information, the mind processes these fragments, and somehow we build an internal representation of reality. Gibson thought this picture was misleading. Perception, he argued, is more direct, more active and more bound up with the environment than that. We don’t first encounter shapes, colours and textures and then later infer what they mean. We encounter an environment in terms of what it allows us to do.

Affordances, therefore, are always relational. A chair affords sitting for a person of the right size, strength and mobility, but not for a horse or a baby who cannot yet sit upright. A shallow step affords access to the able bodied but obstructs wheelchair users. A tree affords climbing to a child with the strength, confidence and desire to climb, but might afford shade, danger, beauty or nothing much to someone with a different set of motivations and characteristics. The same environment offers different possibilities because we all differ in our bodies, histories, purposes, knowledge, fears and incentives. Gibson’s point was that the world was already structured with invitations and constraints long before we developed the ability to reason abstractly about it.

Environmental psychology is the study of how physical and social environments shape human perception, emotion, behaviour and decision-making. It asks a deceptively simple question: how do the places we inhabit act on us? A room is not just a container for activity — it changes any activity that takes places within it. Light, noise, density, layout, temperature, sightlines, furniture, privacy, crowding, entrances, exits and the availability of tools all alter what we notice, feel and do. While the settings in which we act aren’t the whole story, they’re always instrumental to the development of that story.

This doesn’t mean the environment determines behaviour in any crude sense. We’re not billiard balls, knocked into predictable directions by the objects we encounter but we are affected by our environments. In schools, a poorly lit, unsupervised corridor invites lingering. A cramped, overheated classroom invites friction. Arranging tables in groups where students face each other, not the teacher, invites conversation rather than attention. A staffroom with no usable work space tells teachers something about what’s expected there. The design of a setting leans on tthose who inhabit it.

The design theorist, Donald Norman popularised a related idea by distinguishing between actual affordances and perceived affordances. The actual affordance is what can be done whereas the perceived affordance is what people think can or should be done. That badly designed door may physically be pushable, but the handle suggests pulling. We respond to the invitation we perceive, not to the designer’s intention. This distinction is particularly useful in schools because leaders and teachers are constantly engaged in designing systems meant to guide behaviour. But what we intend a system to do is always trumped by what others perceive the system as being for.

We create policies, design classrooms, produce resources and import technologies to support our intentions, and then wonder why people behave in ways that seem to undermine them. The problem isn’t that people ignore what we meant, it’s that thery don’t see it. People respond to what the environment makes visible, easy, safe or worthwhile. Good intentions are filtered though timetables, staffing, habit, workload, accountability, architecture, professional norms and school culture, so that — by the time they reach the classroom — they’ve become invisible and intangible.

This Twitter exchange reminded me of all this:

PowerPoint offers a revealing example. Although it doesn’t cause sub standard teaching, it does make some teaching decisions feel more natural than others. A slide deck affords sequence, display, delivery and forward motion. It makes the planned route of the lesson visible and reassuring: click, advance, continue. This is great for a trainer giving as presentation but less useful in a lesson. A slide deck can make coverage feel like progress and movement through material feel like movement in understanding. It draws attention towards what the teacher has planned to show and away from what students are actually doing with it. Bad teaching is perfectly possible without a projector but if we’re mindful about the ways a tool can shape action, we have an insight into the limitations of the slide show.

Teaching requires movement backwards — and sometimes sideways — as well as forwards. A teacher explains, checks, notices, adapts, models, questions, corrects, returns and rehearses. Understanding is recursive and moves in loops. Slide decks are essentially linear. Of course, they can be used well. They‘re great for displaying preselected images or chunks of text. If you’re sure about where you want to get to they’re a useful way to frame a hinge question. The question isn’t what PowerPoint can technically do but what it makes users think it is for. For most teachers most of the time, a slide deck affords presentation. It affords telling more readily than checking, sequence more readily than responsiveness, and coverage more readily than diagnosis. The deck says, move on regardless of what’s right for the students in the room.

Schools are full of things designed with one intention that afford another. Leaders introduce systems to improve teaching, reduce uncertainty, support students, create consistency or make the school more humane. Teachers use those systems in ways that help them survive the day, teach their classes, protect their time, avoid unnecessary conflict and do right by the students in front of them. Students read the same environment for its own possibilities: where effort is worthwhile, where it’s possible to cut corners, where rules are flexible, where adults mean what they say and where they don’t.

The danger of becoming attuned to affordances is that we move from blaming individuals for rational behaviour to a kind of fatalistic acceptance which sees people as puppets. This isn’t true. We interpret affordances, resist them, redesign them, work round them. Sometimes we make things worse, somethime better. A school creates pressures which shape rather than replace judgment.

We should think harder about what we’re incentivising, not only in the crude sense of rewards and sanctions, but in the subtler sense of what the environment makes easier to notice, easier to start and easier to justify. If behaviour is shaped by the relationship between people and setting, then improvement means redesigning the setting so that better action becomes more available, more visible and more likely. The point isn’t to manipulate people but to stop pretending that “just telling” will be enough.

Desire paths reveal the psychology of use: we don’t move through environments as designers imagine from above, but as physical beings with purposes, pressures and limited patience. When an official route feels indirect, effortful or misaligned with where we want to go, shortcuts are not just more attractive but more obvious. And, as it becomes increasingly visible with each each new person making the journey, it’s more socially permissible and easier for the next. The new path is a record of repeated situated judgement. In schools, the same principle applies when teachers create workarounds or students discover where effort can be avoided and rules negotiated. The sensible response isn’t to try to block the path, but to find out why people are taking it

Too often, policies are written far away from the circumstances in which they’ll be used. A behaviour policy may be designed in an office, but it’s enacted in a corridor when a teacher’s runing late to their next lesson and a student is having a melt down. Feedback policies may be written with learning in mind, but they’re enacted when a teacher has ninety essays to wade through, limited time and a sense that someone will check for evidence of compliance. Ecological psychology shifts us from thinking about what we ideally want to what’s possible with these people in these situations..

Improvement should be less about adding expectations and more about changing the cues, frictions and feedback loops. If we want teachers to respond to students’ thinking, they need lesson routines that make that thinking visible while there is still time to act on it. Mini-whiteboards, short written responses, live sampling, visualisers and hinge questions all have the capacity to alter the classroom environment by making understanding more visible and misunderstanding harder to miss.

If we want feedback to improve learning rather than decorate books, we need to change what the environment asks teachers to produce. Book scrutiny should look less for the visible trace of teacher work and more for whether students’ work has got better over time. Meeting time is better used to examine samples of work, identify recurring errors, compare explanations and decide what students need next. The unofficial desire path of superficial tick-and-flick marking is a sign that the official path is too concerned with evidence and not concerned enough with practicality.

If we want professional development to change practice, we need to design settings in which practice is the obvious thing to do. A room facing a screen invites listening. A department meeting built around examples, non-examples, rehearsal, student work and subject-specific misconceptions invites professional judgement. This changes what people think the space is for. Teachers only become better at explaining difficult ideas when the environment gives them time, materials and colleagues with whom to refine those explanations.

The same thinking applies to behaviour. The environment has to make the desired action easier than the workaround. If teachers are expected to follow up disruption, the process needs to be quick, visible and — crucially —supported by leadership. If students are expected to move calmly through corridors, routes, timings, thresholds and adult presence need to make calm movement normal. If leaders want consistency, they need to reduce the number of points at which adults are forced to negotiate on the fly.

Inclusion can also be redesigned ecologically. If our aim is that all students have access to a challenging curriculum, then the setting should make access easier than avoidance. Teaching assistant deployment should support participation rather than create dependence. Adaptations should help students meet the intellectual demands of the subject rather than lower those demands. The focus must be on how to arrange people, tools, routines and expectations in order to make successful participation more likely for this student in this lesson.

Pressure shapes judgement, but it doesn’t replace it. This way of thinking acknowledges that we are responsive to our environment, but not trapped by it. We can push against it or adapt to it. We can improve our interactions or sometimes make them worse. But an environment that is specifically designed to acknowledge stresses and encourage optimal behaviour is one in which everyone finds it easier to be their best.

The practical response is to audit the invitations and affordances offered by our schools, study the desire paths by walking through a teacher’s week. Consider what the calendar, marking load, meeting structure and data system make easier. Follow groups of students through their day and notice what each threshold, transition, task and adult response seems to promote. Look at department meetings and ask whether thay are making intellectual work likely or just focussing on administrative compliance. Look at the responses to behaviour incidents and ask whether the system makes swift certainty easier than prolonged negotiation. Look at curriculum documentation and ask whether — and how — it makes teachers think about learning or merely getting through content. Wherever students or staff have worn unofficial paths through the system, try to appreciate what’s in it for them and, wherever possible, make these desire paths official.

Schools don’t need more initiatives. For the most part, they need better alignment between intention and environment. If we want students to work hard, the setting has to reward effort. If we want teachers to think, the timetable has to protect thought. If we want curriculum to matter, meetings have to normalise discussion of subject content. When the environment is aligned with our intentions and values, no one has to struggle to do the right thing.

Our environment teaches us how to behave. Staff learn what’s really valued, students learn what’s actually possible. We’re all fallible, thoughtful, anxious, hopeful, self-protective, generous, ambitious people trying to work out what our school is really asking of us. Working out what you want people to do is simple. But working out how to make it reasonable for people to do what you want them to do is the hall mark of great leadership.

In 1970, psychologists and psychiatrists were invited to a lecture on “Mathematical Game Theory as Applied to Physician Education.” The lecture, supposedly given by Dr Myron L. Fox, a graduate of the Albert Einstein College of Medicine and a student of the great John von Neumann, was actually given by an actor who knew nothing about either game theory or physician education. The audience of MDs and PhDs were, in fact, unwitting subjects in a study conducted by Donald Naftulin, John Ware and Frank Donnelly on “educational seduction.”

The audience were divided into two groups; one group was given a lecture by an actual scientist about something relevant and interesting, the other group listened to Dr Fox pedal his nonsense. In the first experiment, Dr Fox was instructed to lecture in as boring a monotone as he could manage. The audience was then tested on how much they’d retained and, surprise surprise, it wasn’t much. In a second experiment, Dr Fox went to town, using the full range of his thespian skills: he had his audience laughing, concentrating and nodding along. Even though the content of the lecture was absolute pap, filled with what Deborah Merritt describes as “double talk, neologisms, non sequiturs, and contradictory statements,” students rated the lecture as interesting, stimulating and valid.1

This phenomenon, that a charismatic speaker could fool a knowledgeable audience into believing any old rubbish was meaningful and worthwhile, became known as the Dr Fox Effect. Dr Fox bamboozled three separate audiences of professional and graduate students. Merritt, in a critique of student surveys used to evaluate lecturers, put it like this:

Despite the emptiness of his lecture, fifty-five psychiatrists, psychologists, educators, graduate students, and other professionals produced evaluations of Dr. Fox that were overwhelmingly positive. … The disturbing feature of the Dr. Fox study, as the experimenters noted, is that Fox’s nonverbal behaviors so completely masked a meaningless, jargon-filled, and confused presentation.

The problem isn’t that audiences enjoy good speakers. Of course they do. The problem is that we’re surprisingly bad at distinguishing the feeling that we’re learning from the reality of learning. We may enjoy fluency, warmth, pace, confidence, humour and theatricality, and mistake these things for understanding. In schools, we do something similar whenever we use engagement, enjoyment, participation, confidence or lesson polish as proxies for learning. These things aren’t worthless, but they’re not the same as students knowing more, remembering more, or being able to do more later.

This may help explain the wild popularity of the late Sir Ken Robinson’s 2006 TED Talk on schools killing creativity. It’s had something like 76 million views, despite being, in my view, utterly and completely wrong-headed. Robinson was funny, charming, fluent and seductive. He told a story people wanted to believe: that schools crush children’s creativity and that the answer is to liberate them from the dead hand of traditional academic instruction. But creativity is not released by reducing the amount children know; it’s cultivated through knowledge, discipline, imitation, practice and the gradual acquisition of the very traditions Robinson was so breezily willing to caricature.

Now, although other researchers have confirmed the existence of the Dr Fox Effect, it appears that although people in the audience rate a good speaker positively regardless of what they say, little actual learning takes place unless the speaker possesses considerable subject knowledge. Psychologists Eyal Peer and Elisha Babad replicated the original study in 2012 but added an additional item to the questionnaire people who attended the lecture were given. The question asked whether audience members felt they’d actually learned anything. The results were interesting: even those students who had evaluated Dr Fox as a highly effective speaker were aware that they’d learned nothing from the lecture.

The problem with ‘engagement’

So, what does this tell us about effective teaching in schools? There’s still a commonly held belief in education that a good teacher can teach anything well and that subject knowledge, whilst not without value, is clearly less important than the pedagogic skills and personal charisma of the teacher. The argument suggests that it’s much better for teachers to focus on acquiring and practising generic teaching skills than on developing subject knowledge.

Although students may well enjoy engaging lessons and motivating speakers, these things don’t actually seem to make much difference to learning. Rob Coe has referred to engagement was a “poor proxy” for learning. This is one of those claims that manages to be both obviously true and weirdly provocative. Of course students need to attend to what they’re learning. Of course we’d rather they were interested than bored but engagement is, at best, merely a condition that may make learning more likely.

In The Hidden Lives of Learners, Graham Nutthall discusses his research which found that “students can be busiest and most involved with material they already know.” Students are far more likely to get stuck into tasks they’re comfortable with and already know how to do than into the more uncomfortable business of grappling with uncertainty. A classroom can look wonderfully engaged because students are busy rehearsing what’s already secure.

The same problem applies to motivation. Here’s Nuthall again: “Learning requires motivation, but motivation does not necessarily lead to learning.” Motivation and engagement may be vital elements in learning, but their impact depends on what they’re used in conjunction with. Motivated to do what? Engaged by what? If students are being motivated to perform superficial tasks, or to practise things they can already do, then the whole enterprise may be a waste of time.

The illusion of confidence

Tom Loveless made a related point in a report for the Brown Center, where he tanalysed PISA findings on student engagement and maths attainment. He found that the relationship between engagement and attainment was far from straightforward. Several high-performing countries reported lower levels of student engagement than the US, and, at national level, students’ enjoyment of maths and enthusiasm for maths lessons were negatively associated with achievement. Confidence was no more reliable: American students reported being more confident than Singaporean students, but their test performance much worse.2 The point isn’t that engagement causes poor performance, but that motivation, enjoyment and confidence are unreliable proxies for learning.

Of course, none of this proves that engagement and intrinsic motivation are actually bad for attainment, but it does cast serious doubt on policies and practices that try to boost engagement in the belief that results will follow. As Loveless put it:

PISA provides, at best, weak evidence that raising student motivation is associated with achievement gains. Boosting motivation may even produce declines in achievement.”

That isn’t a licence for dull lessons bit it is a warning against confusing the visible signs of student enthusiasm with the less visible business of durable learning. What students enjoy may not involve much learning. Working hard is, well, hard work. Grappling with uncertainty often feels uncomfortable. If learning depends on students having to think hard, then activities designed chiefly to maximise engagement may easily steer them away from the very struggle that produces retention and transfer.

There’s also compelling evidence that we’re pretty terrible at judging when we learn best. Most of us prefer blocked or massed practice: we review the basics, complete a few related exercises, reach a feeling of fluency and move on. It feels efficient because performance improves quickly during the practice itself. Interleaving, by contrast, often feels harder and produces more mistakes during instruction, even though it tends to produce better long-term retention. There are similar problems with retrieval practice. Students tend to prefer strategies that feel productive, such as rereading or restudying, while underestimating the value of effortful retrieval. Strategies that produce quick fluency make students feel as if they’ve learned more than they have whereas strategies that feel effortful, uncertain and less successful in the moment may feel worse, but often produce stronger retention later.3

Part of the problem is that blocked practice gives us the warm, fuzzy feeling of cognitive ease. It creates the illusion of knowing. We prefer the comfort of feeling successful now to the discomfort of learning something well enough to remember and use later. In that sense, engagement can become another version of the Dr Fox Effect: a seductive surface signal that makes us feel we’re making progress when the harder evidence is far less reassuring.

What great teachers know

Read more

It’s easy to see why the idea of moving effective teachers into struggling schools has such appeal. For policymakers, it feels fair, practical and admirably direct: if some children have less access to strong teaching, then paying strong teachers to work where they’re needed most looks like a straightforward act of justice.1 For teachers, the assumption is just as seductive. If we’ve been successful in one school, it’s natural to believe that success belongs to us, that our explanations, routines, standards and judgement will travel with us wherever we go. But as many teachers discover, moving schools can shatter that belief. Success in one setting doesn’t guarantee success in another. Not only are the students different, so too are the routines, the curriculum, the expectations, and — perhaps most crucially — the institutional backing we’ve taken for granted may be weaker or all together absent. Personal effectiveness may be, at least in part, a relationship between our expertise and the conditions in which we work.

Teacher quality is real

Teachers matter enormously. The difference between strong and weak teaching is one of the most robust findings in the value-added literature. A one standard deviation increase in teacher quality is associated with annual gains of at least 0.11 standard deviations in maths and 0.095 in reading, while other studies find that stronger teachers are linked not only to higher test scores but to longer-term outcomes such as college attendance and earnings.2

One of the strongest arguments for focusing on teacher quality is that there is often more variation within schools than between them. Hanushek, Rivkin and Kain’s work on teachers, schools and academic achievement used matched panel data to disentangle the influence of teachers and schools, with particular attention to the problem that students and teachers are not randomly distributed. Their findings helped establish a central point in the value-added literature: school-level averages can hide substantial classroom-level variation.

A student in a weaker school may learn more in the classroom of a highly effective teacher than a student in a stronger school learns in the classroom of a much less effective one. School quality shapes the odds, but it doesn’t determine the quality of every lesson. Students don’t experience school in the abstract; their experience is of particular teachers and lessons. Even though a school can be chaotic, individual classrooms can be oases of calm. Equally, a school’s impressive reputation can allow some teachers to coast.

Because teacher quality can — and often does — exceed school quality, it’s tempting to assume it must be stable, self-contained and portable. We imagine a good teacher as someone who embodies effectiveness, ready to be lifted from one school and slotted into another. But although teacher may be more effective than the school around them, they’re never independent of it.

What happens when strong teachers move?

Matthew Kraft and John Papay’s recent paper on teacher portability tests this assumption directly. It uses the Talent Transfer Initiative, a randomised US programme in which low-achieving schools were allowed to offer high-performing teachers from higher-achieving schools a $20,000 stipend to transfer and stay for two years. The design created a rare natural experiment, allowing the researchers to see what happens when teachers with strong records of effectiveness move into markedly different school contexts.

The headline result suggests the programme was a success: transfer teachers performed better than the teachers these schools would otherwise have been likely to hire. But, compared with their own previous performance, they became less effective, with their value-added impact falling by around 0.12 student standard deviations after transfer, enough to move the typical high-performing transfer teacher from about the 85th percentile of effectiveness to the 66th.

This makes clear that teacher effectiveness depends, at least in part, on the conditions in which teachers work. If effectiveness were simply a stable property of the individual teacher, the decline would be hard to explain. A great teacher would remain just as effective wherever they were placed, rather than having their impact altered by different students, routines, curriculum, leadership and school culture.

In the study, the nature of the work itself changed. The transfer teachers moved into classrooms where prior maths attainment was 0.40 standard deviations lower and prior English language arts attainment was 0.29 standard deviations lower, meaning they were having to teach students with substantially different starting points, gaps in prior knowledge and instructional needs. The proportion of students eligible for free or subsidised school meals rose from 68% to 92%. Teachers in the study also reported much weaker conditions for teaching: positive ratings of student motivation fell from 86% to 39%, student discipline from 81% to 52%, and parental involvement from 60% to 32%.

This confirms what many teachers discover when they move schools: expertise is both general and local. Some things we take with us — subject knowledge, habits of explanation, standards, routines and judgement — but some we learn from the students in front of us: what they know, what they struggle with, how quickly they grasp new ideas, which examples and explanations are likely to work.

We also learn the school’s habits: whether leaders follow through, colleagues share standards, the curriculum is fit for purpose, and assessment tells anyone anything useful. We learn how refusal is handled, whether consequences happen, whether line management clarifies or clutters expectations, and whether meetings solve real problems or merely produce evidence of effort. Some of this knowledge may be transferable, but most of it has to be reaquired around new students, routines and constraints.

Kraft and Papay argue that the decline in teacher effectiveness “appears to be driven by lower match quality, negative indirect school effects, and the loss of student-specific human capital.” In plainer language, teachers are more effective when they’re well matched to their school and students, when the school supports good teaching, and when they have built up knowledge of the students they teach. Change those things, and effectiveness changes too. As they put it, “Schools are critical in supporting or constraining teacher effectiveness.”

The measurement trap

For all their faults, value-added models (VAM) capture something important: whether students taught by a particular teacher make more progress than we’d expect, given their prior attainment and other observable characteristics.3 However, these models can tempt us to treat effectiveness as if it’s located entirely within the teacher. The danger isn’t the measure itself, but that we miss what it abstracts away. VAM is a proxy for teacher effectiveness; a statistical estimate of a teacher’s contribution to student test-score growth, not a direct observation of teaching quality. It depends on the effects of all the other teachers to have taught a student, as well as the personalities, beliefs and home lives of students themselves. And this is before we even begin to think about the school in which estimates are produced.

Approached with caution, this can be useful. Modelling makes it clear that teachers differ in quality and that students learn more in some classrooms than they do in others. As such, it helps separate a teacher’s contribution from the advantages or disadvantages students bring with them. Without it, we risk mistaking high prior attainment for effective teaching, or difficult starting points for weak teaching. The problem begins when an estimate of a teacher’s contribution is treated as the teacher’s essence.

School improvement narratives have long been dominated by a strangely individualistic model of teaching. Because weak teaching is treated as the property — or fault — of weak teachers, improvement becomes the process of correcting, coaching, incentivising, moving or replacing individuals. Of course, sometimes that’s necessary. Like all other human attributes, teaching ability distributes on a bell curve and some teachers need more support than others.

The problem is that this account is hopelessly incomplete. Teaching quality is produced by teachers working within a particular environment. Schools can make it easier or harder for teachers to be effective. In some schools, even the weakest teachers can teach effectively due to how efficiently systems operate. In others, even the most skilled are limited by the chaos or indifference that surrounds them. If we fool ourselves into believing teaching is purely the output of teachers, not only do we miss the most effective levers to pull to improve teaching, we contribute to the misery so many teachers experience as the day-today reality of their professional lives. Any measure is misleading if it encourages us to ignore the extent to which the system produces the thing being measured.

When we behave often behave as though improvement can be installed in individual teachers while everything around them remains unchanged we are likely to pour resources into a ‘solution’ that makes everything harder for teachers and has school leaders operating in the dark. This is the coaching trap: identifying the teacher as the unit of improvement when it’s school culture that’s the source of the problem.

My point isn’t to abandon instructional coaching — it’s still a good bet in schools with strong cultural foundations — but to stop expecting it to create those foundations without addressing the systemic pressures which make it hard for teachers to be their best. As Kraft and Papay’s portability study makes clear, if school conditions shape teacher effectiveness, then trying to improve teaching one teacher at a time is, at best, inefficient and, at worst, irresponsible.

Kraft and Papay’s earlier work on professional environments points in the same direction: teachers improve more, and more quickly, in schools with stronger professional conditions: better order and discipline, purposeful collaboration, effective leadership, useful professional development, coherent culture and fair evaluation. And even the best teachers can be diminished by a change of context.

Systems and forces

In case it seems that I’m just blaming leaders for poor teaching, I need to make it clear that principals are not responsible for the conditions in which their schools have to operate. Schools sit inside wider systems of accountability, inspection, funding, recruitment, safeguarding, SEND provision, curriculum policy and public expectation. Many of the pressures that distort school life arrive from outside the school gates. Heads may create local systems, but they do so inside larger systems that reward some behaviours, punish others and make certain kinds of leadership much harder than they should be.

It’s helpful to think of schools as systems inside larger systems, constantly buffeted by external forces. Systems include those things over which leaders have at least some control: routines, structures, habits, processes, incentives and norms. Forces are those things imposed on schools from outside.

Forces are the weather. Funding settlements, inspection frameworks, teacher supply, local demographics, statutory duties, parental expectations, political priorities, social media storms, safeguarding demands and the changing complexity of children’s needs all press on schools regardless of a headteacher’s decisions. We can’t avoid these forces; they are the background against which we must contend.

Systems, on the other hand, are the responses we engineer in order to operate within those forces we can’t control. They’re the routines, structures, habits, processes and norms we build to make it possible to survive — and, ideally, thrive —despite external pressure. Children’s natural tendency to prefer not to be made to do algebra on a Friday afternoon requires reliable behaviour routines. Overstuffed exam specifications and students’ uneven prior knowledge require an efficient, coherent curriculum. Uncertainty about what students know requires assessment that reveals patterns and suggests workable solutions, rather than merely replicates what we already know as is typical with most internal assessment.4 Organisational complexity requires line management that balances trust and accountability in ways that allow teachers to do more than merely comply with explicitly stated expectations, and our finite capacity for attention requires systems that protect teachers from the illusion that all demands can be treated as equally important.

If we blame heads for the weather, we’ve simply moved crude individualism up a level. Blaming teachers for the effects of poor school systems is wrong, but blaming heads for the effects of poor national systems is no better. Headteachers inherit shortages, incentives, legal duties, inspection anxieties, budget constraints and political noise. They are not sovereigns. They are operating inside a system that often rewards visible compliance over genuine improvement.

Once effectiveness is imagined as something possessed by exceptional individuals, school improvement can look like a matter of importing the right person. This was the logic behind the cult of the ‘superhead’: find a charismatic turnaround leader, drop them into a difficult school, give them licence to act, and wait for improvement to arrive.

There have been heads who took on chaotic schools, created order, raised expectations and improved results. It would be foolish to deny that. But the superhead story has always been more complicated than its admirers wanted to admit. Too often, it treated leadership as a portable personal force rather than a set of conditions, relationships, routines and judgements built over time.

The cleanest caution here is not scandal but sustainability. In 2016, Schools Week reported on research into ‘superheads’ which found that although results rose during their tenure — often due to a combination of exam gaming and off-rolling — scores fell by an average of 6 percentage points after they left, with larger falls when the head had been in place for longer. Worse, the subsequent clean-up costs were also substantial, coming to £11.8 million across 21 schools.5

Effective heads don’t parachute in with a bag of tricks. Instead, they build a school’s capacity to improve without depending on them. They develop middle leaders, strengthen curriculum, make behaviour routines predictable, improve the quality of professional conversations, protect time, clarify accountability and make it easier for teachers to be their best.

The worst version of the heroic headteacher model does the opposite; centralising decision-making, accelerating only high profile change, frightening dissent into silence, chasing short-term indicators and mistaking compliance for strong culture. A school may become more orderly, but not necessarily a functional place for teachers to work.

With teachers, the temptation is to imagine classroom effectiveness can be moved without changing the routines, relationships and curriculum that shaped it. With heads, the temptation is to imagine institutional effectiveness can be imported without developing the people, habits and safeguards that sustain it. A strong head in a weak governance structure, with brittle middle leadership, poor trust and exhausted staff, may be able to impose change but this is not the same as building a culture in which teachers and students can thrive.

Towards a surplus model of school leadership

The influence of headteachers, though profound, is indirect. Leaders shape what’s prioritised, tolerated, measured, ignored and allowed to drift. A headteacher can’t improve teaching through force of will. They can, however, create the conditions in which excellence becomes more likely.

A deficit model of school improvement locates failure in individuals. Because teachers can’t be trusted, students don’t care and middle leaders aren’t holding the line, staff need tighter systems to make them comply. When things go wrong the default assumption is that someone is to blame and the solution is tighter monitoring, greater prescription and less professional trust.

A surplus model begins from a different premise. It assumes, until there is good evidence otherwise, that people are acting in good faith, working hard and trying to do the right thing. When things go wrong, it seeks to identify and remove the barriers to success. This means recognising that weak practice often emerges from unclear expectations, poor feedback loops, conflicting incentives, missing knowledge, unmanageable workload or convoluted routines.

Students arriving late to lessons offer a simple example. Deficit thinking reaches quickly for character: students are lazy, teachers are inconsistent, sanctions aren’t tough enough. A surplus model looks first at the design of the school day. Is movement time realistic? Are corridors congested? Do entry routines vary from classroom to classroom? Is the policy clear enough to be enacted day to day? The solution may still involve sanctions, but it begins with design rather than blame.

Deficit leadership is an attempt to find simple answers to complex problems, whereas a surplus model asks leader to take responsibility for the systems they oversee.

These models obviously contain elements of caricature. No school operates an entirely deficit or surplus model but you make recognise elements of these approaches to leadership in schools you’ve worked in. Importantly, a surplus model is not a call for blind trust. Trust without precision becomes drift; accountability without trust becomes theatre. Teachers are most likely to improve when they feel trusted, but they also need to know they are accountable. The aim is intelligent accountability: clear enough to prevent mutation, humane enough to avoid compliance theatre, and expert enough to distinguish looking good from being good.

Effective school leadership is to build routines in which being good is easier than looking good. That means turning behaviour follow-up into something staff can trust, reducing the number of priorities, protecting departmental time, ensuring assessment meetings to look at students’ work rather than just at spreadsheets, and insisting that policies are either maintained or abandoned rather than endlessly announced and ignored.

Every school trains its staff but does it train them into better habits or teach them how to survive bad ones? A surplus model of leadership is focussed on creating a culture where better teaching is more likely. This still leaves room for teachers to take responsibility for what they can control but acknowledges the limits of what is within their sphere of influence. Teachers still need secure subject knowledge, disciplined habits, clear boundaries, and to focus on the intellectual preparation required to teach effectively but it’s much easier to develop these when not having to act as a warlord in a failed state.

Students in the most disadvantaged schools absolutely need the best possible teaching but this is only possible when they are aligned around the principles that make such teaching possible. Where a well-led school multiplies professional development, a school which is poorly led absorbs it.

The belief that school improvement is only possible when great teachers work with great school leaders is unhelpful. What we need is great teaching and great school leadership. Both are best thought of as the product of systems rather than of individuals.

Moving people who have proved themselves successful in one setting into harder schools may still be worth doing, but this transfer is not what a sustainable improvement strategy.

This is where schools make their entrance. A school is not just buildings where children are kept busy; it is an example of humanity’s most powerful tool for transmitting cultural knowledge quickly, efficiently, and equitably. In a world where the pace of cultural and technological change vastly outstrips the speed of biological evolution, schools are essential for ensuring that each generation inherits, understands, and builds upon the knowledge of the past. Without them, we risk forcing each child to start from scratch in a world that’s far too complex, far too unforgiving, for such inefficiency.1

1. Why our brains are the way they are

In order to comprehend ourselves we need to know that we have been forged in the crucible of natural selection. In our distant past, we naked apes had few advantages beyond our big brains and opposable thumbs. In order to survive we needed to band together with other members of our species and share any useful survival tricks we came up with. These ‘good tricks’ were collected, refined and passed down the generations, forming the basis of human culture.2

It’s tempting to view culture – “the extensive accumulation of shared learned knowledge and its iterative improvements over time”3 – as that which sets human beings apart from all other species. While a few other species make very limited use of tools, we have invented smart phones, 3D printers and the Large Hadron Collider. But our capacity for culture must also have evolutionary roots. And it does. The evolution of culture took place side by side with the evolution of our genes. In fact, who we are and how our brains work is the result of two and a half million years of gene-culture coevolution. The evolution of genes is painfully slow, yet culture can evolve incredibly rapidly and have the effect of galvanising the selection process on genes, resulting in far faster biological innovation. (One of the most well-known examples is the evolutionary relationship between the genetics of lactose tolerance in humans and the cultural habit of dairy farming.4) This time period saw a quadrupling in size of the human brain,5 unprecedented changes in human gene expression,6 and archaeological evidence of hyperexponential increases in the complexity and diversity of our technology and knowledge base.7 Our reliance on each other and the tools we have designed to assist us in our efforts to stay alive have shaped our brains. In the words of evolutionary biologist Kevin Laland, “Human minds are not just built for culture; they are built by culture.” But what we’ve ended up with is just good enough to survive and reproduce; in essential ways, our brains are the same as those of our Palaeolithic ancestors.

When we think about how and what to teach, we should consider the role of evolution in shaping the way humans have adapted to think and learn. In our distant past, learning was a costly strategy – time spent learning was time we couldn’t spend surviving and reproducing – so it makes sense that we have evolved to learn as efficiently as possible. Evolutionary biologists think of learning as being either social or aso­cial. Social learning is essentially copying (what’s everyone else doing?), whereas asocial learning is accrued by interacting with the environment through trial and error. Everything we know is either learned socially or asocially, through mimicry or experimentation, emulation or innovation.

Both types of learning have associated advantages and disadvantages. The advantage of asocial learning is that you get accurate, up-to-date, first-hand information about what works and what doesn’t, but the cost is high. You could waste hours on strategies that barely work—or worse, you might end up dead after eating a suspiciously funky mushroom, tumbling off a cliff in pursuit of some gourmet berries, or getting mugged by something with claws and a bad attitude.

The advantage of social learning is that it’s easier, safer and more likely to result in productive survival strategies. Human beings, like many other species, operate in groups, and so copying the behaviour that others – especially those in your kinship group – have adopted seems a sensible strategy. After all, if everyone around you is getting on well, why would you risk trying something different? Copying something badly means you are less likely to survive and reproduce than those who copy things well, and so being good at copying is selected for. The downside to relying on other people’s experiences of what’s most likely to be effective means that we might not understand why we do what we do. When environmental conditions change rapidly or a new predator is introduced, old strategies may prove ineffective and those who can adapt the quickest are those who will survive.

It used to be believed that the numbers of social to asocial learners in a group would be fairly evenly balanced – that environmental change would favour asocial learning, whereas stability would favour social learning – but it turns out that social learning forms the basis for the remarkable growth and success of human culture.8 This might seem counter-intuitive: surely, new innovations gleaned from asocial tinkering must be the most important driving force in human ingenuity? The thing is, although we need a minimal amount of asocial learning, most people get on most of the time purely using socially learned strategies, and that’s because we copy strategically. Only the most successful ideas get passed on and spread throughout the group. Each new generation hones in on optimal solutions, so as long as there’s a little bit of experimentation going on – either through asocial learning or through copying errors – culture accumulates and is continually refined. This is essentially what Richard Dawkins referred to as memetic reproduction.

In the modern world, we support a very small number of people – sci­entists, artists and the like – that they may spend a fraction of their time on asocial learning. The rest of us spend our lives directly copying those around us or accessing the vast accumulation of human culture through word of mouth, books, the internet and now, GenAI. Pretty much every moment of every day is spent engaged in tasks which are directly or indirectly copied. As the Bible tells us, “The thing that hath been, it is that which shall be; and that which is done is that which shall be done: and there is no new thing under the sun.” If we choose to engage in a brief bout of asocial learning, we do it for fun and because we’re safe enough not to worry about it going too far wrong.

Obviously, when the zombie apocalypse comes, asocial learners will be in much demand; those who work out how to survive in the new paradigm fastest will have an enormous advantage over the rest of us. But then, if humanity is to survive, it will be because we copy the new ‘good tricks’ they come up with and begin the fight back against the undead.

So, what does all of this have to do with education? Simply this: while we might all enjoy a small amount of asocial experimentation, almost everything we learn – and almost certainly everything useful – will be due to our ability to observe and emulate. A school curriculum that favours a trial and error approach to reacquiring what has previously been discovered as the result of several millennia of iterative copying is fighting against biology. We’re just not fitted to learn that way. Even Jerome Bruner, often miscast as a champion of the discovery learning cause, could see that existing knowledge and culture were not generally passed on by dis­covery. He wrote:

You cannot consider education without taking into account how cul­ture gets passed on. It seems to me highly unlikely that given the cen­trality of culture in man’s adaptation to his environment – the fact that culture serves him in the same way as changes in morphology served earlier in the evolutionary scale – that, biologically speaking, one would expect each organism to rediscover the totality of its culture – this would seem most unlikely.

Jerome S. Bruner, Toward a Theory of Instruction

By far the most effective way to pass on the fruits of human culture is to share what has already been discovered and invented as clearly and as explicitly as we can.

This gets to the heart of what we believe schools are for. Should they be safe spaces in which we allow children to tinker about at the margins of human culture, maybe discovering something useful for themselves? Or are they, as education professor Michael Young has said, places that should “enable young people to acquire the knowledge that, for most of them, cannot be acquired at home or in the community.”9 The first choice is a Darwinian jungle in which those fortunate enough to have wealthy, educated parents will thrive and the devil take the hindmost. If you believe in social justice and giving children a fair chance to escape the constraints of this lottery, using schools to promote effective social learn­ing is the only option.

2. Why some things are easy to learn but others are hard

Our ancestors needed to find food, avoid danger, mate and ensure their children survived to adulthood. The need to find food selected for minds which could attend to and remember the characteristics and behaviour of plants and animals, the resourcefulness to provide shelter and defend against predators, and the ingenuity to consider how inanimate objects might be used as tools. The ability to compete and cooperate in groups, whether for food gathering or mating opportunities, selected for minds which were able to anticipate the motives and emotions of others, and better communicate needs and ideas. These are the folk disciplines: folk psychology, folk physics and folk biology.

Read more

Most professional development is barely professional development at all. Teachers are gathered in a hall, shown a few slides, invited to discuss something agreeably vague about pedagogy, and then released back into classrooms where nothing much has changed. Teachers’ assumptions remain intact; their routines untouched, the curriculum is still what it was and the behaviour system still works or doesn’t. Teaching is simply expected to have been improved via teacher exposure to groovy ideas.

In fact, installing air conditioning is likely to do more for student outcomes than business-as-usual CPD. That sounds far fetched only if we assume the main barriers to better teaching are always located inside teachers. Often they’re in the fabric of the school itself. A classroom that’s too hot to think in is an instructional problem but not one within the power of teachers to solve.1

Instructional coaching, however, offers a compelling alternative to this bleak picture. Not only is it a specific, practical approach to professional development, it treats teaching less like a set of beliefs and more as a craft. The basic ideas is that a teacher tries something new whilst observed by a knowledgeable coach who then offers precise feedback. The teacher rehearses, tries again, and the distance between what they meant to do and what actually happened is made visible.

Teaching is full of things that sound straightforward until you try them in front of thirty children at half past two on a wet Thursday. Explaining clearly, securing attention, modelling first steps, checking for understanding, asking better questions, responding to error and making practice successful are all simple enough in theory. In the moment they can be anything but.

Coaching is great, but can it scale?

So coaching has obvious appeal. It promises what most CPD lacks: specificity, repetition, accountability and adaptation. It doesn’t just tell teachers what good practice is; it helps them do it.

The evidence is encouraging too. Kraft et al’s meta-analysis of 60 causal studies found that teacher coaching improved both what observers saw teachers doing and what students went on to achieve. Although, he effects were much larger on instructional practice than on attainment. Changing teacher behaviour is one thing; changing what students know, remember and can do significantly harder. While this should make us wonder whether the preferred changes to to teacher behaviour were actually aligned with what is most likely to improve outcomes, still, compared with the thin gruel of much school improvement, this is pretty respectable.

However, the kind of coaching that works is very hard to scale. As Kraft et al point out, “The average effectiveness of the coaching program declines as the number of teachers involved increases, suggesting the difficulty of successfully taking such programs to scale.” Coaching produces its strongest effects in smaller, more controlled programmes, where there are expert coaches, clear models, protected time, close monitoring and a strong theory of teacher development. Once coaching expands across a large system, effects shrink. It becomes harder to guarantee the quality of the coach, the precision of the feedback, the time for follow-up, and the conditions in which practice can actually change.

This is where so many schools get stuck; they take the visible structure of coaching, assume they’ve captured the thing itself, buy an expensive plug-and-play programme, create coaching cycles, allocate coaches, produce forms, ask teachers to identify targets, and add a box to the appraisal system. Although schools are ‘doing instructional coaching,’ whether anyone’s actually getting better is hard to know.

All too often, the reality is that teachers are being visited, nudged, recorded and encouraged. Sometimes this helps and sometimes it doesn’t. In the worst cases, coaching is simply judgmental lesson observation with a thein veneer of respectability.

Here’s the thing: coaching isn’t a mechanism, or a product that can be bought off the shelf. For it to be effective, instructional coaching is a relationship organised around expertise. If the expertise is removed it becomes little more than therapy. And if the relationship is removed, it becomes monitoring. More importantly still, without sufficient investment in time and resources, without the school conditions that make change possible, coaching devolves into punitive, box-ticking advice shouted into the wind.

It’s convenient to believe the axiom that “the qaulity of an education system is limited by the quality of its teachers.” This is far more truthy than it is true. As I’ve argued before, if we want better teachers we need better schools. And if we want better schools, we need better school leaders.

Is school culture the limiting factor?

If I’m right, the limits of instructional coaching are those created by school culture. A teacher can be coached to improve their questioning, but an unreliable behaviour system will still blunt the effect; they can learn to model more clearly, but an incoherent curriculum leaves the model directionless; they can check for understanding more carefully, but if assessment across the department is a mess, no one knows what understanding is supposed to look like. We can coach teachers to tighten classroom routines, but where leaders tolerate inconsistency, they’ll be trying to build on sand.

Schools routinely try to improve one teacher at a time while leaving the system that shapes their work largely untouched. Why? Because it’s far easier to blame inadequate teachers than to recognise you’ve created the conditions in which inadequacy flourishes.

Teachers improve at different rates depending on the professional environments in which they work. Teachers in schools at the 75th percentile for professional environment ratings improved 38% more over ten years than teachers in schools at the 25th percentile.2 This moves the argument away from the familiar question, “How good is this teacher?” and towards a more useful question: “What kind of school makes teachers better?

And, if we boil that down, we find that schools in which teachers improve most have the following characteristics:

• Order and discipline. Behaviour is predictable enough for teachers to teach, not constantly negotiate

• Purposeful collaboration: Teachers work together on curriculum, tasks, explanations, assessment and student work, not just planning admin

• Instructionally credible leadership: Leaders know enough about teaching to support, challenge and model good teaching

• Useful professional development: CPD is directly connected to the curriculum and the real problems teachers face

• High trust:Teachers can admit difficulty without fearing humiliation or career damage.

• Fair evaluation: Feedback is specific, proportionate and aimed at improvement, not blame.

• Coherent culture: The school’s routines, expectations and incentives all point in the same direction.

These are all factors outside of an individual teacher’s control. Teachers can’t simply will reliable behaviour systems, coherent curricula, purposeful collaboration, high trust or fair evaluation into existence. Teachers improve in schools that actively and purposefully reduce the friction of getting better.

We’ve inherited a model of school improvement that’s far too individualistic. Weak teaching is treated as the property of weak teachers. Improvement therefore becomes the process of correcting, coaching or replacing individual teachers.

Of course, there’s some truth in this. We know that teachers vary, that expertise matters and that some teachers need more help than others. Pretending otherwise would be sentimental nonsense. But it’s far from the whole truth. Teaching quality is produced by teachers working within particular environments. The same teacher can be thoughtful, calm and precise in one setting, then become frazzled basket case in another. Pretending that it’s the teacher that’s changed is pernicious.

Good schools make better teaching more likely. A good school will have clear behaviour systems, a curriculum worth teaching, assessment that reveals useful information, line management that clarifies rather than clutters, meetings that solve problems rather than generate minutes, professional development connected to the real work of departments, and — crucially — leaders who know enough about teaching to guard against nonsense.

Poor schools do the opposite. They make every improvement harder, turning clarity into negotiation, practice into compliance, feedback into threat and curriculum into documentation. They force teachers to spend energy compensating for systemic failures, then, when they inevitably struggle, offer them coaching.

This is the coaching trap. We identify the teacher as the unit of improvement when the school is the source of the problem.

Principals create school culture

Teachers remain the most important direct influence on what happens in classrooms. But heads and principals shape the conditions in which classrooms operate. They decide what’s prioritised, what’s tolerated, what’s measured, what’s ignored and what’s sacrificed. Their actions determine whether behaviour systems work. They decide whether to focus on curriculum coherence or on silly snake diagrams imposing an illusion of coherence. They decide whether middle leaders are developed or merely burdened, and whether meetings are instruments of improvement or time sinks.

The evidence on leadership is compelling. The Wallace Foundation’s 2021 synthesis, drawing on 219 studies from two decades of research, concluded that principals affect student achievement, attendance, teacher satisfaction and teacher turnover. It also identified four key practices associated with effective principals: focusing on instruction, building a productive school climate, supporting professional learning and managing people, data and processes well.3

This fits with an older Wallace Foundation Paper whiuch argued that school leadership is second only to classroom teaching among school-related influences on student learning. The effect is largely indirect, but that’s precisely the point. Principals affect students by affecting the adults who teach them and the environment in which they work.4

That indirectness is not a weakness. It is the mechanism through which school leadership operates. Heads don’t improve writing by personally teaching every Year 8 English lesson but by making sure there is an effective curriculum in place, that practice is explicit, standards are shared, assessment is useful, behaviour allows attention, and that middle leaders know how to support improvement.

Who wants to be a headteacher?

If heads are so vital, we might hope that ambitious teachers are queueing up to take on the role. Sadly, this is not the case. TeacherTapp’s 2026 recruitment and retention report found that fewer teachers at every level say they want to become a headteacher, but the most serious fall is among deputy and assistant heads. Just 37% of deputy and assistant headteachers say they aspire to headship, down from 55% in 2017.

That should make us pause; the people closest to headship increasingly look at the job and decide it’s not for them.

In addition, actual headteacher turnover is low. In 2025/26, headteacher changes were below every recent year except 2020/21, when the pandemic suppressed movement. The fall appears to be driven mainly by the secondary sector, while primary turnover looks closer to normal. Although low headteacher turnover is not necessarily bad — Schools benefit from continuity, and improvement usually needs time — when combined with falling aspiration among deputy and assistant heads, it starts to look less like healthy stability and more like institutional fragility. Although the system may appear settled because fewer heads are moving, if the pool of potential heads is shrinking, that’s not a recipe for long-term improvement.

If the single most powerful influence on school quality is leadership, and if fewer senior leaders want to become head teachers, then the system has a leverage problem. We’re asking more and more of the role while making it less and less attractive to the people most likely to do it.

Much of this can be attributed to workload, accountability, budgets, inspection, safeguarding, SEND complexity and the sheer emotional weight of the job. Reporting on the same broad pattern, TES linked the decline in aspiration to squeezed budgets, inspection pressure and shifting school structures. Against these external forces, the capacity of the system to change is limited but still, the most urgent questions in education are how do we make headship an attractive proposition and how do we help headteachers to be better at leading teaching.

This is where the coaching conversation ought to pivot. A school has one headteacher and many teachers. England has 468,258 teachers, but only about 24,000 heads. Coaching every teacher requires a vast supply of expert coaches, protected time and favourable school conditions. Coaching heads is more plausible because there are fewer of them and because their decisions shape the conditions in which all teachers work.

Rolling out teacher coaching hoping the school is ready for it is an expensive mistake. The foundation is instructional leadership. If we were to coach heads first — if we could make sure behaviour, curriculum, assessment, line management, meeting structures, middle leadership and professional development are secure — then we would have more fertile soil for coaching to take root.

Instructional leadership is not wandering around classrooms with a clipboard. It means developing the honesty and expertise to build a school in which good teaching is more likely, and more sustainable. Honesty here means more than not lying. It means doing what you said you’d do. If leaders promise consistency on behaviour, feedback, workload or curriculum, and then fail to follow through, trust steadily erodes. Heads do not need to be subject specialists in everything, but they do need to know enough about curriculum, assessment and teaching to recognise expertise in others. If they don’t have sufficient expertise, why should anyone trust their judgement? Trustworthy leadership requires both. Honesty without expertise ends up with a well-meaning muddle, whereas expertise without honesty results in manipulation. Instructional leadership depends on being worth listening to and worth believing.

The point of headteacher coaching is to help heads become more instructionally powerful: more honest in what they promise, more reliable how they follow through, and more expert in judging the work of teachers. That means helping them see where school systems are undermining classroom practice, where policies are producing perverse incentives, where middle leaders need more support, and where confident performance is being mistaken for genuine improvement. Instructional power is the ability to make better teaching more likely across the school.

A model for head teaching coaching

A good model for head teacher coaching should start with the school, not the leader’s personality. This is not about developing admirable leadership traits. Schools don’t improve because heads become more charismatic, reflective, resilient, authentic or any of the other popular leadership industry adjectives-du-jour. Schools improve when leaders build better habits of instructional leadership: noticing the right things, following through on decisions, testing whether routines work, protecting teachers from distraction, and making it easier for good teaching to happen. For years I’ve argued that primary responsibility of school leadership is to strip out all other demands on teachers’ time so that they can focus on teaching the best possible curriculum as well as they possibly can.

This is not to say that the many and various other leadership responsibilities are unimportant. Safeguarding, finance, attendance, staffing and pastoral care all matter enormously. Some are legally and morally non-negotiable. But they are not the purpose of a school. They are the conditions that allow the core purpose to be pursued. If instructional leadership is not given the highest priority, you may be managing a complex organisation, balancing competing pressures and keeping the machinery moving, but you’re not really running a school. You’re running everything around the school while the core work — teaching children to master an ambitious curriculum — is left to chance.

Stage 1: Diagnose the instructional architecture

The coach and head would diagnose the school’s instructional architecture: behaviour, curriculum, assessment, line management, meetings, department routines, staff development, induction, cover, communication and workload. The point would not be to create another grand strategic plan — schools have enough of those — but to identify the constraints that are making good teaching unnecessarily difficult.

Every school, intentionally or otherwise, trains its teachers. The only question is whether it trains them into better habits or teaches them how to survive bad ones. A headteacher coach should help the head see these messages clearly. The first task is not to ask what the head wants to achieve, but twhat the school is already teaching its staff to do.

Stage 2: Rehearse instructional leadership

The process would then move from diagnosis to practice. Heads should rehearse giving feedback after a lesson visit; practise asking better questions in a department review; earn how to look at student work without turning the process into a hunt for defects, and develop enough understanding of each area of the curriculum to recognise expertise, challenge weak thinking and know if they’re being fobbed off.

But they should also deliberately practice the hardest parts of leadership: challenging weak practice without humiliating people, distinguishing evidence of compliance from evidence of change, and training their leadership teams to do the same. It isn’t enough for the head to develop better judgement, hat judgement has to be distributed, practised and held to account across the wider team.

Stage 3: Build implementation discipline

Schools are excellent at producing evidence of activity. There’s always a document, a spreadsheet, a folder, a policy, a tracker, a learning walk, a survey, a set of minutes. Much of it proves very little. Instructional leadership requires the discipline to ask: what has actually changed in what teachers do, what students experience and what students can now do that they couldn’t do before?

A headteacher coach should be irritatingly interested in implementation.

• What will teachers stop doing?

• What will happen when the keen staff have adopted it and everyone else is waiting to see whether it goes away?

• What will this look like in the weakest department?

• How will you know whether people are pretending?

• Where will the workload go?

• Which middle leaders have the expertise to make this work?

• What will you do when the first version fails?

These are not administrative questions. They are the substance of school improvement.

Principal coaching should therefore be organised around a small number of high-leverage practices.

Stage 4: Scale through the system

Once the head’s judgement has been sharpened, they must ensure it moves through the system. Heads don’t simply improve their own practice. They alter the ecology in which everyone else practises. This means building a coherent instructional model. Not a branded checklist or a laminated set of non-negotiables. A shared understanding of how attention, explanation, modelling, practice, feedback and assessment fit together.

It means developing middle leaders. In most secondary schools especially, the head’s influence on teaching operates through heads of department. If middle leaders can’t evaluate curriculum, lead subject-specific development, identify weak practice, analyse student work and give useful feedback, the head’s vision is much less likely to be realised.

This requires a coherent strategy for each of the following:

1. Behaviour: Systems must be utterly reliable. Nothing improves teaching quite like students being able to listen, think and work without constant low-level disruption. This is the precondition for effective school culture.

2. Curriculum: The main thing should be the main thing. If schools don’t give a minimum of one hour per week to departmental priorities, they’re not taking curriculum seriously. Schools are forever adding: more projects, more strategies, more priorities, more awareness days, more forms of evidence. The head’s job is not to feed the machine but to decide what matters enough to displace what’s merely nice.

3. Intelligent accountability: Schools need feedback loops that encourage honesty and expertise in all staff. Performance management theatre undermines professional development. Data drops too often convert uncertainty into decimals. Feedback loops should help leaders know whether the school’s chosen improvements are reaching classrooms and improving student work.5

Coaching heads is not just easier to organise, it has greater leverage. If every head shapes the conditions in which many teachers work, coaching current and aspirant principals to be superb instructional leaders makes it possible for teachers to thrive.

If the supply of future heads is shrinking, we cannot afford to treat headteacher development as an optional extra. Coaching heads cannot continue as a optional extra for those senior leaders who see its value. Whilst not easy or cheap to implement, a mandatory principal coaching system is not only a more effective way of protecting the conditions in which teachers can improve but could also make headship feel less like a lonely act of professional self-sacrifice and more like a difficult though learnable craft.

Of course, headteacher coaching can fail in all sorts of predictable ways. It could drift into motivational chat or be captured by consultants with no serious understanding of teaching or school leadership. Clearly the role of the headteacher coach requires careful thought and should depend on a nationally recognised qualification.

Coaching for heads should be evidence-informed, school-facing and instructionally specific. It should involve visiting the school, looking at real artefacts, watching lessons with the head, analysing student work, observing leadership meetings and reviewing whether previous decisions have actually changed anything. It should be demanding enough to be uncomfortable but supportive enough to be valued and valuable.

My suggestion, then, is not to abandon teacher coaching but to stop develoving to teachers work that belongs to leadership. Of course coaching can help teachers get better, but if we understand and accept that it is the school which determine whether getting better is easy, hard or almost impossible, we’re more likely to implement coaching programmes effectively. A well led school multiplies coaching; a poorly led school absorbs it.

Because we’ve treated instructional coaching as though it were a universal solvent, capable of dissolving every weakness in practice huge amounts of precious time and resources have been frittered away. Teachers are made to look like the locus of underperformance because teaching is where underlying issues become visible.

So by all means coach teachers but if we want coaching to scale, we should start where the leverage is greatest. We need to coach the people who shape the conditions in which teachers work, who decide what schools attend to and whose judgement makes the weather everyone else has to live with.

If we take headteacher coaching seriously, we’re more likely to get schools where teachers can thrive. And if we do that, teachers have a fighting chance of being the best they can be.

Whenever a practice is mandated, there’s an inexorable tendency for it to lethally mutate. Retrieval practice is a perfect example. When I first started writing about the testing effect as we used to call it, many people found the idea surprising. The idea that attempting to dredge something up from memory was a more effective way to learn it than simply restudying it felt deeply counterintuitive. Surely looking again, rereading, highlighting and “going over” material ought to work? The fact that it seemed to contradict teachers’ lived experience made sure people responded in one of two ways. While some dismissed it as obviously wrong, others grappled with the idea in order to make sense of the discrepancy. This meant that those who began using retrieval practice in their classrooms usually did so with a degree of care. If they didn’t get the predicted results straight away, they’d rethink, go back to the source material, tweak the task, adjust the timing and try again. In other words, early adoption tended to be thoughtful because we were finding our way.

Today, retrieval practice is so much a part of the furniture that new teachers are routinely told to ‘do’ it without anyone ever explaining the underlying processes. It’s built into lesson expectations and bandied about as if it were beyond dispute. It’s travelled from interesting research finding to institutional commonplace.

Unsurprisingly, that’s been the cause of thousands of misbegotten versions of the ‘retrieval practice’ which copy the superficial appearance and ignore the meachanisms that make it work; a bit like copying the minty taste of toothpaste but leaving out the fluoride.

A powerful cognitive principle has, in too many classrooms, become an empty ritual. Students write down questions from the board. Some can’t answer them. The teacher reveals the answers. Students copy them in green pen. Everyone has technically ticked a retrieval shaped box but no actual retrieval practice has taken place.

Students learn either that the opening minutes of a lesson are a bit of administrative grit to endure, or — worse — that they routinely experience failed before the lesson has even begun. The problem is the misguided attempt to turn retrieval practice into orthodoxy.

How forgetting works

Forgetting is normal. We often think we know things we’ve actually forgotten. The more familiar something is, the more likely we are to believe we can recall it whenever necessary. Rereading a page, looking over notes or nodding along to an explanation can make students feel as if something has been learned. But then, when they need to recall the knowledge unaided, it isn’t there.

Retrieval punctures the illusion, showing students what they can and can’t bring to mind. More importantly, the act of trying to remember also strengthens the very knowledge it tests. The effort of dredging through our memory makes future remembering more likely.

But this doesn’t mean any old quiz is retrieval practice. Neither does it mean all lessons should begin in the same way, or that disconnected interleaving is always the best use of the opening minutes. And it certainly doesn’t mean retrieval is the whole of learning.

The New Theory of Disuse helps distinguishes between two kinds of strength in memory: storage strength refers to how firmly knowledge is embedded in long-term memory and how likely it is to endure, whereas retrieval strength refers to how accessible that knowledge is right now.

If knowledge vanished entirely, there would be nothing to retrieve. We’d have to relearn everything from scratch every time we forgot. But that’s not the case. When we forget, the route between here and now and what we want to know has weakened. A cue, a first letter, a familiar context or a partial prompt is often all it takes to make the answer available again. This explains why relearning is significantly quicker than first learning; a trace remains, even when he cannot being something to mind unaided.

When students cannot provide an answer, it might mean one of several things:

• They never learned it in the first place.

• They learned it, but not securely enough. Either the cue isn’t strong enough to access it or the retrieval interval was too long, so access has weakened.

• They’ve confused it with something similar

• The question is badly phrased or asks for the knowledge in an unfamiliar way

• They’re anxious, rushed or worried about being wrong.

Obviously, we cannot retrieve what we don’t know. Before students can retrieve anything, it has to be explained, modelled, practised and corrected. Retrieval stabilises knowledge but it can’t conjure it from the air. The important point is that “they don’t know it” isn’t a diagnosis, just the first step in creating more flexible, durable learning.

Performance vs learning

Understanding the difference between retrieval and storage stength helps explains why responses to recent teaching can be so misleading. If students have just been taught something, retrieval strength may be high because the material is still warm. They can recall it at the end of the lesson, answer questions on it, and appear to have made progress. But this immediate performance tells us more about temporary accessibility than durable learning.

Storage strength works differently, growing through successful retrieval event after some forgetting has taken place. When students have to work to bring knowledge back to mind, and succeed, that effort strengthens future access.

The paradox is that attempts to maximise performance now can reduce learning later. If we keep knowledge constantly available, over-prompt students, reteach before they have tried to retrieve, or test only what’s just been explained, we raise retrieval strength in the short term. Students appear successful because the answer is close at hand. But because retrieval is too easy, storage strength doesn’t grow. If a memory hasn’t had to be searched for and reconstructed it hasn’t been strengthened.

This is why ‘getting it in the lesson’ is such weak evidence. It may show only that temporary retrieval strength was high but conceals the likelihood that knowledge is less durable.

Although retrieval practice should be less fluent than simply rereading or being told an answer, the difficulty has to be desirable. If students fail to retrieve an answer, they’re not strengthening access. Often students are guessing, waiting or copying. Likewise, if knowledge is retrieved but never used, instructional gains will be limited.

The crucial point of retrieval practice is not to make students perform well here and now but to improve the chances of performing well elsewhere and later. Retrieval strength is what students can access right now; storage strength is what makes that access more durable tomorrow. Effective retrieval is always oriented to the future.

The following principles are all designed to focus on improving performance in the longer term:

1. Recall beats recognition

2. Success before struggle

3. Retrieval should serve the curriculum

4. Secure then use

5. Make it visible

1. Recall beats recognition

We often mistake familiarity for knowledge. Students recognise the term, remember seeing a diagram, can nod along to an explanation, and can pick the right answer from a list. But the real test is whether they can produce the answer when support is taken away.

Rebecca Lawson’s study, The Science of Cycology, exposes the illusion of knowledge. Lawson asked people to complete drawings of bicycles by adding missing parts such as the pedals and chain. Bicycles are hardly exotic and we’ve all seen thousands of them. Most of us can ride a bike and many even own one. But when asked to reconstruct how a bicycle actually fits together, participants made frequent and serious errors: chains went to the wrong wheel and pedals appeared in impossible places. Lawson’s conclusion was that people’s understanding of familiar everyday objects can be sketchy and shallow, even when the information is often seen and easily perceived.

Familiarity flatters us. Because we know a thing when we see it, we assume we know all about it. But while recognition is cheap recall is costly. And paying that cost is the point of retrieval practice.

The illusion of knowledge is endemic in the classroom, reiforced day after day, lesson after lesson. Students may recognise Pythagoras’s theorem when it appears in a list, but be unable to state it. They may recognise the word “mitochondria”, but be unable to define it. They may be able to spot a quotation from Macbeth, but be unable to bring it to mind when writing an essay. They bobble along with the comforting feeling of knowing stuff without the usable possession of anything.

Retrieval practice works because students have to bring knowledge back from memory and are confronted with unambiguous evidence or what they do and don’t not know. As we’ve seen, it’s the effort of retrieval that strengthens future recall. This means it’s important to structure retrieval questions that provide enough of a cue to prompt a search of memory without actually giving away the answer.

Which of the following versions of the same question is easier to answer?

A Who is Scrooge’s dead business partner in A Christmas Carol?

B Who is Scrooge’s dead business partner in A Christmas Carol?

a) Bob Cratchit
b) Jacob Marley
c) Fezziwig
d) Fred

B is obviously easier because it dangles the answer in front of students, just waiting to be recognised. Students can get it right because they recognise Jacob Marley as the right answer but they haven’t had to dredge up anything from memory in an effort to recall it. Of course, the problem with A is that students might try to recall but come up with nothing. As we’ll discuss further in Principle 2, there’s a sweet spot we should aim at but for now, a better restricted retrieval prompt would be:

C Scrooge’s dead business partner is J________ M________.

This gives enough of a cue to start the search, but not so much that students can simply recognise the answer without having to check through previously learned material. If they’ve previously learned the answer they should be able to bring it to mind. The initials narrow the field but the blanks still require recall. That’s the sweet spot: supported production, not recognition.

Multiple-choice questions are often a poor fit for retrieval practice because they are too focused on recognition. Because the answer is provided, students may choose it either because it looks familiar, because the alternatives look wrong, or because the wording gives the game away. This doesn’t mean MCQs are useless. They can diagnose misconceptions, check fine distinctions and, with good feedback, support learning, but they’re a risky default. They may help students access knowledge, but they also cue recognition and expose students to plausible wrong answers. If the aim is fluent recall, students still need to practise producing answers unaided. MCQs are best used for diagnosis, discrimination and discussion, not as the main vehicle for retrieval.1 If we want students to remember, they need to practise producing answers, not merely spotting them.

A retrieval task should usually ask students to produce something: a definition, quotation, term, cause, method, distinction, example, sentence, diagram, step or explanation. “What is Pythagoras’s theorem?” is retrieval. “Which of these is Pythagoras’s theorem?” is recognition.

The best restricted prompts support recall without doing the remembering for students. “The rightful heir to the Scottish throne is M______” is better than asking students to pick Malcolm from a list. The cue narrows the field, but students still have to retrieve the answer. The same principle applies in science: “The structure responsible for generating energy in both plant and animal cells is m___________” supports recall without devolving into recognition.

This is not dumbing down. It is the careful engineering of successful remembering. The point is to give students enough support to retrieve, but not so much that they only recognise.

Good retrieval punctures the illusion of knowledge while students can still do something about it. It shows them that “I know it when I see it” is a long way from “I can bring it to mind when I need it.”

2. Success before struggle

One of the predictable ways retrieval practice goes wrong is by asking students questions that are just too hard. Retrieval is just one of several ‘desirable difficulties’ identified by Robert and Elizabeth Bjork as being likely to support storage strength whilst reducing retrieval strength.2 The difficulty comes from having to remember something you haven’t just been told, not come from being unable to answer. A desirably difficult task is one that requires effort but — crucially — results in success.

Difficulties are only desirable if they lead to successful remembering. If students can’t retrieve the answer, they are not strengthening memory. They’re guessing, freezing, waiting or copying. There’s no useful struggle and no success. There is just time passing as students fail.

This is where so many attempts at retrieval practice go wrong. A question appears. Students don’t know the answer. The answer is revealed. Students write it down. The teacher moves on. Nothing has been retrieved and — probably — nothing has been learned.

Here are some examples of pointlessly difficult questions I’ve seen recently.

• English: “What is anagnorisis?”

• Maths: “What is the formula for the area of a trapezium?”

• Science: “What is the function of the Golgi apparatus?”

• History: “What was the significance of the Treaty of Brest-Litovsk?”

Conceivably, this questions might be fine except for the fact that in each case, students gazed at the board blankly and clearly had no idea how to answer. I asked a couple of students about anagnorisis and most of them shrugged. One girl thought it was a type of metaphor. These questions all come from classrooms in different schools, set by teachers trying to leverage the effects of retrieval practice in good faith. Needless to say, you have to know something in order to retrieve it.

The remedy is to ensure retrieval is successful. If students studying Macbeth are asked, “Who is Malcolm?” there’s a lot that can go wrong. Maybe they genuinely don’t know Malcolm is. They may remember hearing his name but not the context. Or, maybe they do know who he is but answer in a way that doesn’t connect with the answer the teacher was looking for. If the teacher is expecting “Malcolm is the rightful heir to the Scottish throne,” but students answer, “One of the guys who ran off after Duncan was killed,” there’s an issue. Although they’re right, the problem was caused by the ambiguity of the question. Now, if instead the teacher had asked, “Who is the rightful heir to the Scottish throne?” we get another set of predictable problems. If Malcolm isn’t familiar enough, this might be too weak a prompt. Students might guess the wrong answer. Worse, they might guess the right answer without necessarily matching the retrieved name to the prompt in their memory. The result of this is a comforting illusion of knowledge because they’ll remember they got the answer right, but not what the answer was.

What about if we tweaked the prompt like this:

The rightful heir to the Scottish throne is M _ _ _ _ _ _.

What will go wrong now? Well, as you’re probably aware, there are a number of plausible candidates whose nemas also begin with M. Students might guess Macduff or even Macbeth. The solution is for the teacher to circulate during retrieval practice to see whether students can answer the question. Now, if we see students are struggling we can add more letters: M _ l _ _ _ m. Even if students need every letter revealed except one, they still having to do a minimal amount of recall. The point is that the level of difficulty should adapt to the level of students’ struggle. If they struggle too much, they clearly need reteaching. If they’re struggling at all, they need more practice. Once they can reliably answer this question without additional promption, we can can then start to ramp up the difficultly and start asking, “Who is the rightful heir…?”

Recall starters should aim for universal success. Not because challenge is unimportant, but because failure at the start of a lesson is corrosive. Students who repeatedly begin lessons by failing learn that they’re rubbish at this subject or that school isn’t for them. This is in no one’s interest.

And yet, aiming for 100% success shouldn’t mean retrieval is easy. Instead it should mean effortful success. Students should have to stretch but they should be able to grasp.

Importantly, this isn’t to make students look impressive but to slow forgetting and build accuracy, fluency and retention. Students need to know that effortless success is the goal. Once they no longer have to think, when they ‘just know’ to right answers, they can start to think with the knowledge they’ve acquired. Accuracy must come first: students need to get it right. Fluency follows: they need to get it right without using up too much mental effort. Retention is the long game: they need to get it right later, when the knowledge is needed for something more demanding.

If students cannot retrieve quickly and successfully on mini whiteboards, the task probably isn’t functioning as retrieval practice. Although it may still be a worthwhile question, a good discussion prompt, or a useful problem, we shouldn’t call everything retrieval just because students are being asked to think.

If most students can’t retrieve the knowledge, that’s a vital piece of information. Perhaps the question needs more scaffolding? Perhaps the content needs reteaching? Perhaps the gap between teaching and recall was too long? Perhaps the previous teaching was less secure than we hoped? Whatever we discover, the point is to repsond.

3.Retrieval should serve the curriculum

Read more

So, as is usually the case when a pattern in the data seems to clarify a messy problem, the school sprang into action. Although everyone was already aware of the small group of high profile students, the new information focussed on where, when and with whom the most serious incidents seemed to occur. Attention therefore shifted to the contexts that appeared to be producing the behaviour. The teachers whose lessons appeared most often in the logs were offered support, and middle leaders were asked to look more closely at the relevant lessons, corridors and transition points.

For a while, it seemed to work. The number of high-tariff incidents fell. The behaviour lead’s hunch appeared to have been confirmed. The pastoral team felt encouraged. Senior leaders were relieved. The graphs moved in the right direction, and the school began, cautiously but understandably, to congratulate itself on having found the source of the problem. The only people less convinced were the teachers whose lessons had been placed under scrutiny.

Then, within the space of a week, three of the students at the centre of the original analysis were permanently excluded following serious incidents.

So what had gone wrong?

The evidence, though accurate, was misleading. The students really had been involved in incidents clustering around particular places, times and lessons. And the teachers’ names really did appear more often than others. The trouble is, data sets are unable to tell the full story. In this case, the pattern had been imposed before anyone began to interpret the data.

A behaviour log is not an accurate transcript of school life. Incidents are only recorded after a chain of judgements has already taken place: what was noticed, what was tolerated, what was challenged, what was escalated, what was written down, and what was dealt with in ways that never entered the system at all.

Ask yourself why some incidents might be logged while others might not. What are the factors that could lead to this outcome?

Here a few of the possible variables:

• Some teachers follow the behaviour policy closely and log every incident that meets the threshold.

• Some teachers trust the system and believe logging will lead to useful action.

• Some teachers don’t trust the system and have learnt that logging rarely changes anything.

• Some teachers have different thresholds for what counts as serious enough to record.

• Some teachers avoid logging because they worry (often correctly) frequent incidents will make them look unable to cope.

• Some teachers deal with incidents informally before they reach the official system.

• Some teachers absorb too much disruption because they’ve normalised it or feel there’s no alternative.

• Some teachers log less because they don’t want to damage relationships with students.

• Some incidents are logged because they are highly visible: shouting, refusal, walking out, aggression.

• Some students are logged more readily because they already have a reputation.

• Some incidents are not logged because the system is time consuming, awkward to use or unclear.

• Some incidents are logged because they involve students already under review.

• Some incidents are/are not logged because they fit descriptions in the behaviour policy.

• Some incidents are/are not logged because they trigger an automatic sanction.

If we consider just the variables dealing with teachers, we could assume that incidents are more likely to be logged by teachers who:

a) Follow the behaviour policy closely and believe consistency matters.

b) Trust leaders to respond fairly and use the information properly.

c) Are less confident managing behaviour and want support or escalation.

d) Are new to the school and trying to follow the system exactly.

e) Notice, challenge and record behaviour others may tolerate or absorb.

While there may be some over lap between some of these factors, we can be fairly sure that teachers a and e are probably distinct from theachers who fall under c. A cluster around particular teachers might indicate weak routines, difficult classes or poor relationships. But it might also indicate higher standard and fidelity to the behaviour policy. An absence of logs might suggest calm, well-established classrooms, but might just as easily suggest avoidance, resignation or invisible disorder.

In this case, once attention shifted to the teachers and locations that appeared most strongly in the data, the system changed. Teachers knew they were being watched. Some became more cautious about what they recorded. Some managed more incidents informally. Some may have hesitated before escalating because escalation had begun to look like evidence against them. The behaviour logs improved, but that didn’t necessarily mean behaviour had improved.

The school had mistaken a pattern for a cause. The data had clustered around particular lessons, corridors and transition points, but that didn’t mean those contexts were producing the behaviour. They were simply the places where behaviour was noticed, challenged, escalated and recorded. Just as a map is not the territory, a behaviour log is not behaviour itself, but the paper trail it leaves behind.

The statistical name for this particular problem is collider bias. When two or more factors influence whether something appears in a data set, and we then look only at that selected group, those factors tend to appear to be related in ways that mislead us. The collider is the selection point, the place where different causal paths meet, and the source of the trouble is that selection can create patterns which look meaningful precisely because we have forgotten that selection has occurred.

Imagine a university that admits students partly because they have high test scores and partly because they’re very wealthy. In the general population, ability and wealth may be related, unrelated or only weakly related. Yet among admitted students, a pattern appears: less wealthy students may tend to be especially able, while less able students may tend to be especially wealthy. In a selected group where either advantage can help secure entry, those who lack one advantage are statistically more likely to need the other.

The university, by acting as a filter, has created an apparent relationship between ability and wealth. It may even make the two qualities look as though they trade off against each other, when what we are really seeing is the effect of the admissions process. Collider bias is a mechanism for manufacturing plausible falsehoods.

In 1946, Joseph Berkson showed how hospital samples could produce misleading associations between diseases. Suppose two different conditions both increase the chance that a person is admitted to hospital. If instead of looking at the wider population we study only hospital patients, the two conditions may appear negatively related. Among people already in hospital, someone who doesn’t have one condition must have some other reason for being there.1

The hospital, like the university, is the filter. What became known as Berkson’s paradox was later absorbed into the wider development of causal inference, particularly through the use of causal diagrams, or directed acyclic graphs. In these diagrams, a collider is a variable with two arrows pointing into it: X causes Z, and Y causes Z. If we leave Z alone, X and Y need not be associated; but if we condition on Z, by selecting only cases where Z occurs, controlling for Z, or comparing only within levels of Z, we may open a path between X and Y that wasn’t there before.

Although better analysis usually means controlling for more variables, this depends on understanding the difference between confounders and colliders. Confounders are common causes, whereas colliders are common effects.

Suppose we analyse our data and find students who attend revision sessions get better results. Should we make all students attend revision sessions?

Motivation may be a confounder because it sits upstream of both: motivated students are more likely to attend revision and more likely to revise independently, complete homework and do well in the exam. If we fail to account for motivation, we’ll give the revision session too much credit.

But selection can also create a collider. Suppose we look only at students who attended revision. Some students were there because they were highly motivated; others were there because they were seriously underprepared and had been required to attend. Within that selected group, motivation and prior weakness may appear to trade off against each other: the most motivated students may look less weak, while the weakest students may look less motivated. But that pattern may have been created by the fact that either motivation or weakness could get a student into revision in the first place. The revision session is the collider: a common effect, not a common cause. Controlling for a confounder may reduce bias; controlling for a collider creates it.

Medicine has had to take this seriously because so much medical evidence is observational, partial and filtered through systems of access, diagnosis and recording. During COVID-19, researchers warned that analysing only people who were tested, infected or hospitalised could distort relationships between risk factors and outcomes. Testing and hospitalisation wasn’t random. Age, occupation, symptoms, health behaviour, existing illness and access to healthcare could all affect who appeared in the data, which meant that, once researchers looked only at one selected group, some relationships could appear weaker, stronger or simply stranger than they really were.

The so-called obesity paradox raises a similar problem. Some studies have suggested that, among people already diagnosed with certain conditions, higher body weight can appear to be associated with better outcomes.2 Of course, there may be biological explanations in some cases, and it would be foolish to dismiss that possibility out of hand. But collider bias fits the facts: if body weight and illness severity both affect diagnosis, hospitalisation or inclusion in a study, then studying only diagnosed or hospitalised patients will distort the apparent relationship between weight and survival.

If you’ve been thinking that this all sounds pretty similar to survivorship bias, you’d be right. During the Second World War, analysts examined bullet holes in returning aircraft and, quite reasonably at first glance, concluded that armour should be added where the bullet holes were most common. Fortunately, Abraham Wald saw the flaw. The planes that returned had survived those hits; the missing planes mattered more. Armour was needed where returning aircraft had fewer bullet holes, because planes hit in those places were less likely to make it back.

Survivorship bias is about what sample leave out; collider bias is about the false relationships samples can create. Observed cases have survived a selection process, and that process is part of the evidence. If we ignore this fact, our data may point in exactly the wrong direction.

Once you understand the problem, examples begin to appear everywhere. A dating app can make attractive people seem boring and interesting people seem unattractive, but that’s because you’re only looking at people who have passed through some combination of attractiveness, wit, proximity, availability, algorithmic visibility and your own habits of attention. If someone is less attractive, they may need to be more interesting to catch your attention; if someone is less interesting, they may need to be more attractive. The app doesn’t show you the broad sweep of humanity, just the cross section you’ve selected.

Hollywood creates a similar illusion, making it seem as though beautiful actors can’t act and brilliant actors are less atractive, but the visible sample has already been filtered by beauty, talent, connections, luck, persistence, timing and notoriety. Those with neither beauty nor talent disappear from view long before we start forming theories about the relationship between the two. Among those who remain, beauty and talent may therefore appear to trade off because either can help someone survive the selection process.

Elite sport produces its own versions of the same mistake. A smaller footballer may need exceptional technique, balance and anticipation to survive selection, while a technically ordinary footballer may need unusual speed, strength or stamina. Among professionals, qualities can look as though they substitute for each other, even when they may be positively related in the wider population. The academy, the scout, the coach and the contract have already filtered the sample before the pundit begins to explain it.

Business is full of this sort of nonsense. A company studies its top salespeople and finds they all ignore the script, so it relaxes the script and then wonders why sales fall. The problem is that only the successful rebels were studied. Success selected the sample, and, within that selected sample, rule-breaking looked like genius.

Social media may be the most efficient collider engine ever built. The posts you see are selected by outrage, novelty, status, timing, your previous behaviour and the platform’s appetite for attention. You then infer truths about the world from what survives that tournament. Everyone seems furious, every debate seems polarised, every institution seems to be collapsing, and moderation appears to have vanished. Some of that may be true, but your feed is not a reliable survey.

Now return to the headteacher and the behaviour log. Once we recognise that behaviour incidents appear in the log only after a chain of events has taken place. The reason she was telling me the story because she’d seen through the flawed data pattern and realised they were in danger of penalising teachers for using the very system they relied on for information. The best way to improve a system that relies on teachers being willing to log incidents is to incentivise them to do so by making sure you’re trustworthy. If teachers can’t trust leaders to either interpret the data correctly or help manage incidents when they arise, why would they bother to continue using a broken system?

We may believe the log is giving us information about patterns of student behaviour but what we’re actually seeing is a survey on who uses the system, how often and for what. That doesn’t make the data useless but it does mean that it’s partial. If we want to find out where and how students behaver there is no substitute for boots on the ground.

Schools are constantly looking at selected groups: students removed from lessons, students receiving intervention, students attending revision, parents who complain, departments below target, teachers who request coaching, students entered for foundation tier, persistent absentees and children labelled as causes for concern. These categories can be useful, and often they’re necessary, but they’re never neutral. Here’s a list of some the wasy we both rely on data and routinely fall victim to collider bias:

• Managing risk: removals, repeated sanctions, persistent absence and safeguarding concerns demand attention.

• Rationing scarce resources; Intervention, tutoring, mentoring, pastoral time and coaching can’t be offered to everyone.

• Meeting accountability demands: Disadvantaged students, SEND students, borderline candidates and persistent absentees have to be tracked.

• Making messy problems easier to discuss: “Year 9 boys in science” is easier to handle than a knot of curriculum, status, staffing, routines and prior attainment.

• Studying the cases that seem most relevant: Exclusions are analysed through excluded students, intervention through students in intervention, complaints through parents who complain.

• Starting after the filter has already operated: By the time a student is labelled “persistent absentee” or a teacher “causing concern”, earlier decisions have already shaped the category.

• Turning labels into explanations: “challenging students”, “weak teachers” and “hard-to-reach parents” begin as descriptions and harden into causes.

• Comparing groups: Students who attend revision, receive intervention or get removed aren’t randomly assigned.

Once a category exists, we start treating it as self evident. And what’s easy to record becomes all we know.

The easy availability of data makes it tempting to study spreadsheets instead of doing the hard, dull work running a school: visiting lesson, patrolling corridors, talking to teachers and students, looking for what isn’t logged, asking which students are invisible, examining where policies are being followed and where they’re being worked around, and considering whether the systems we put in place reward accuracy, compliance, avoidance or performance.

Knowing about collider bias doesn’t mean we should ignore evidence but that we should stop mistaking selected evidence for reality. Sometimes where the evidence looks clearest is also where it’s hardest to see the truth.

Feedback is widely claimed to be one of the most powerful things teachers can do. It appears near the top of effect-size league tables, features in almost every school improvement plan, and has acquired the status of professional common sense. If students don’t know how they’re doing, how can they improve?

And yet teachers know from long experience that feedback doesn’t always make the difference it should. If you’ve ever found yourself writing the same comment on the same student’s work for the third time in a term, you’ve probably wondered what the point is. We’ve all felt the frustration of seeing yesterday’s carefully explained correction vanish in the very next lesson. We’ve all experienced the empty ritual of giving out carefully crafted targets which are then ignored the moment students work independently.

But, of course, sometimes things do click. A comment lands, a misconception shifts, and a student suddenly sees what they couldn’t see before and the work they produce is permanently improved. Nothing obvious has changed, and yet everything has. What makes feedback so mysterious is that it can feel powerful, precise and necessary in the moment but disappear without a trace.

The trouble is, we mistake feedback for learning. Following an instruction is not the same as acquiring judgement. Students’ work might look better, but next week the same problem returns wearing a different hat. Or even the exact same hat. More troublingly, feedback can actually reduce performance.

Kluger and DeNisi’s classic 1996 meta-analysis found that in 38% of studies, feedback interventions had a negative effect on performance. In many of the studies they analysed, feedback effects were close to zero, meaning feedback often made little practical difference. So, although some were strongly positive, many were modestly positive. A handful of very large positive effects pull the average upwards which helps explain why feedback effects are often reported as being strong. They may be impressive in the aggregate but, as all teachers know, maddeningly inconsistent in the classroom.

The extreme value of the data point on the far right of the graph (d = 12) should be treated as an outlier rather than as evidence of a normal classroom effect. For clarity, it suggests that in one study the effect of giving feedback was 12 standard deviations which means the feedback group performed so far above the comparison group that the result is almost certainly driven by something unusual in the study design, the measurement, the task, the sample, or the way the effect size was calculated. In real classroom terms, this is not a gold standard but evidence of something very odd. Although the paper shows the distribution of 607 effects, it doesn’t identify which study produced each bar. Its main value is therefore illustrative: a few enormous positive effects help pull the average upwards to the reported average effect of d = .41 and help disguide the sometimes powerfully negative effects of feedback.

Hattie and Timperley make a similar point in The Power of Feedback:

Feedback is one of the most powerful influences on learning and achievement, but this impact can be either positive or negative. Its power is frequently mentioned in articles about learning and teaching, but surprisingly few recent studies have systematically investigated its meaning. This article provides a conceptual analysis of feedback and reviews the evidence related to its impact on learning and achievement. This evidence shows that although feedback is among the major influences, the type of feedback and the way it is given can be differentially effective. [My emphasis]

All this should make us pause. If feedback is so effective, why does it so often fail? If teachers spend so much time giving it, why do students so often ignore it? And if feedback is meant to improve learning, why does it so often merely improve the piece of work in front of us?

The problem is that we tend to treat feedback as a substance. The more of it we pour over students, the more learning we’ll get. But feedback isn’t fertiliser; it’s information. And information doesn’t cause improvement unless students understand it, value it and know how to act on it.

Hattie and Timperley argue that feedback works when it answers three questions: Where am I going? How am I going? Where to next? In other words, effective feedback clarifies the goal, helps students understand the gap between current and desired performance, and gives them some useful sense of their next steps. They also distinguish between feedback directed at the task, the process, self-regulation and the self. The most useful feedback tends to focus on the task, the strategy or the student’s capacity to monitor and improve their own work. The least useful is feedback directed at the self, including much of what passes for praise. Feedback can be powerfully effective when it directs attention to the right thing at the right time.

Sadly, there’s no magic cheat code. One of the few things we can reliably infer from the research on feedback is that there’s no formula for perfect feedback. Anyone who says there is, is either misinformed, selling something, or in urgent need of spending more time in classrooms. Written comments, whole-class feedback, verbal feedback, live marking, codes, rubrics, exemplars, model answers and one-to-one conversations can all work but, equally they can all fail. What matters is whether the feedback fits the student, the task, the subject, the phase of learning and the reason for the gap.

So, rather than asking which feedback method works best, we need to ask what feedback is supposed to do. The answer can, I think, be boiled down to five principles:

1. Feedback should build judgment, not dependence

2. Feedback must respond to current performance

3. Feedback must change with the phase of learning

4. Feedback should avoid ‘ego involvement’

5. Feedback must be used, not just received

1. Feedback should build judgement, not dependence

The purpose of feedback shouldn’t be to improve students’ performance in the here and now but to ensure that improvement shows up elsewhere and later. This may sound obvious, but so much of what passes for feedback is really editing in disguise. We spot an error, explain the problem, supply the improvement and the student makes the desired change without being any the wiser. The work is improved but not the student.

In English, this is particularly easy to see. A teacher writes, “Develop this point,” or “Embed your quotation,” or “Link back to the question.” The student dutifully adds a sentence, drops a quotation into the middle of a paragraph and repeats a word from the question in the final line. The essay may now look better but has the student learned anything? Do they know why one quotation is more effective than another? Can they see the difference between a genuine conceptual link and repetitive waffle?

In his classic account of formative assessment, D Royce Sadler argues that students need a concept of the standard being aimed for, the ability to compare their current performance with that standard, and some understanding of how to close the gap. Crucially, this doesn’t giving more feedback. Sadler’s point is that students must come to share in the teacher’s evaluative expertise. They need to recognise quality while they are working, not merely be told afterwards that quality was absent. As he puts it, “students have to be able to judge the quality of what they are producing and be able to regulate what they are doing during the doing of it.” Feedback, then, shouldn’t just consist of information transferred from teacher to student but be part of the process by which students develop their own judgement. The aim is for them to become less dependent on feedback over time.

It might appear that Hattie and Timperley cut against my claim that feedback should change the student, not the work. After all, they argue that the most useful feedback tends to focus on the task, the process or the student’s capacity for self-regulation, while the least useful feedback is directed at the self. Doesn’t that mean feedback should focus on the work rather than the student?

Only if we confuse the target of attention with the purpose of feedback. Feedback should focus on the work because that’s where improvement can be made visible. But its purpose is to change the student: what they notice, what they understand, what they can do, and how well they can monitor their own performance next time. The more specific teachers are the better students’ attention is focussed where it belongs: “Your second sentence explains the effect of the image; now make the first sentence do the same,” keeps attention on the task, the process and the next action. Although the immediate focus is the sentence, the intended change is in the student’s judgement.

Students need to internalise ways of noticing what matters. When feedback moves students from correction, to strategy, to self-monitoring, they move from seeing only what the teacher points it out to being able to see it for themselves.

With simple tasks, the gap may be obvious. Where there’s a binary condition — the answer is either right or wrong, a capital letter is missing or present, the procedure has broken down at the third step or it hasn’t — students can be shown the error, practise the correction and get better. But not all tasks are like this. Some domains are relatively ‘kind’ with clear goals, stable criteria and easily visible errors. Spelling, arithmetic procedures, grammatical accuracy, scales on a musical instrument and many aspects of pronunciation fit this pattern. In kind domains, feedback can often be precise and direct because the relationship between error and correction is clear.

Other domains are more ‘wicked.’ Goals are harder to define, quality is contested and improvement depends on taste, discrimination, context and accumulated knowledge. Writing an analytical paragraph, interpreting a poem, evaluating a historical argument or composing an original piece of music can’t be improved simply by being told the answer. In wicked domains, feedback has to teach students to see.1

The difficulty is that students often don’t know what they’re looking at. They can see the words on the page, the explanation in the exercise book or the pattern on the map, but they can’t yet see the quality of the move they need to make. They don’t know whether an explanation explains, whether evidence has been integrated or merely dropped in, whether a pattern has been described or just accounted for. In these cases, feedback has to draw attention to the feature of quality the student has not yet learned to notice.

English

• Task: Explain how Shakespeare presents Macbeth as vulnerable to manipulation.

• Students’ response: “Macbeth is ambitious because he wants to be king.”

• Weak feedback: “Develop this point.”

• Better feedback: “This identifies Macbeth’s ambition, but it doesn’t explain why ambition makes him vulnerable. Add a sentence explaining how wanting power makes him easier for the witches and Lady Macbeth to manipulate.”

• The student needs to learn that writing an essay isn’t about simply adding more words but explaining relationships:

  • ambition → vulnerability

  • vulnerability → manipulation

  • manipulation → action.

Science

• Task: Explain why increasing temperature affects the rate of reaction.

• Student response: “The rate of reaction increased because it got hotter.”

• Weak feedback: “Use scientific keywords”

• Better feedback: “You’ve identified temperature, but you haven’t explained what changes when the temperature increases. Add a sentence explaining that the particles move faster, collide more often, and are more likely to collide with enough energy to react.”

• The student needs to learn that a scientific explanation isn’t about naming the condition that changed. It’s about explaining a complete causal sequence:

  • higher temperature → greater particle speed

  • greater particle speed → more frequent collisions

  • more frequent collisions with enough energy → faster rate of reaction

Geography

• Task: Explain why population density is higher near the river.

• Student response: “The population is higher near the river because there is water.”

• Weak feedback: “Explain why water affects settlement.”

• Better feedback: “You’ve identified water as a factor, but you haven’t explained why it attracts settlement. Add a sentence linking the river to farming, transport, trade or industry, then decide which factor is most important in this location.”

• The student needs to learn that geographical explanation isn’t about naming a feature on the map. It’s about explaining how physical and human factors interact in a particular place:

  • river → water supply, fertile land or transport

  • water, land or transport → farming, trade or industry

  • farming, trade or industry → higher population density.

The pattern is the same in all three examples. Better feedback points out both what the student can currently see and what they are missing. The feedback then focuses attention on a feature of quality: relationship, causal sequence, mechanism, interaction, precision. The next step requires the student to act on that feature, not merely comply with an instruction.

This distinction between kind and wicked domains helps explain why generic feedback policies so often disappoint. If we treat all tasks as if they belong to the same kind of domain we shouldn’t be surprised when students struggle to make progress. A missing apostrophe and a weak interpretation of Macbeth are not the same kind of problem, however much a tracking spreadsheet might give the appearance they are. In kind domains, feedback may correct an error. In wicked domains, feedback has to build discrimination.

If feedback improves the artefact but leaves the student no better able to judge quality, it will make students more not less dependent.

2. Feedback must respond to current performance

Feedback doesn’t act directly on learning. Kluger and DeNisi’s Feedback Intervention Theory suggests that feedback works by drawing attention to a discrepancy between current performance and a goal. What happens next depends on how the student responds to that discrepancy. The student might increase effort, change strategy, lower the goal, reject the feedback or disengage altogether. The key variable is attention. Feedback is more likely to be beneficial when it focusses attention on the task and the next action; it’s more likely to backfire when it shifts attention towards the self. This explains why the same piece of feedback can land so differently with two different students.

The table above shows why feedback so often fails to do what we want or expect. If our feedback tells students they are exceeding expectations, they may choose to aim higher, but they may also decide the task is under control and reduce thier effort. They may conclude that the goal was too easy to be worth pursuing, or simply ignore the feedback altogether. Only one of these choices (aim higher) is desirable and so feedback must be consciously titled towards this end.

The real problem comes when students are falling short. Feedback is only useful if it results in them trying harder and changing strategies, but that’s only one possible response. All too often students decided to aim lower, give up entirely, or reject the feedback and disengage.

The feedback we give doesn’t determine the path students take. Feedback creates a discrepancy between current performance and the desired goal. Clear, accurate, precise feedback will always backfire if the discrepancy it reveals is too overwhelming, humiliating or hopeless. We often assume that pointing out the gap will lead students to close it, but students are acting rationally when they protect themselves from the gap instead. The key question is not just whether the feedback is correct, but whether it makes productive action more or less likely.

Saying, “You need to make your explanation more precise” might be exactly what a confident student needs but for a student who barely understands the content, it may simply confirm how lost they are. The same words but a very different effect.

Feedback must therefore be adjusted according to current performance. If students are broadly meeting expectations, feedback can refine, extend and complicate. It can draw attention to nuance, invite comparison and raise aspiration.

• In maths, a student who can reliably solve a linear equation might be asked to compare two methods and explain which is more efficient, or to solve a problem where the equation has to be constructed first. The feedback shifts from “do the next step” to “choose the best route”.

• In PE, a student who can perform a basic lay-up might be asked to adjust the timing under defensive pressure, use the weaker hand, or decide when not to take the shot. The feedback should focus on how they can adapt techniques to make them more useful in game situations.

• In music, a student who can play the notes accurately might be asked to shape the phrase, control dynamics, or compare two interpretations of the same passage. Instead of being asked to play the correct notes they might respond better to being prompted to make different musical choices.

When students are failing to meet expectations, feedback may need to redirect attention, reteach content, simplify tasks, clarify goals or rebuild the conditions for success. There’s little point offering sophisticated feedback to a student who has misunderstood the question, lacks the necessary knowledge or cannot see what success is supposed to look like.

Crucially, feedback also has to make renewed effort feel worth the effort. If students have already tried harder and still got nowhere, why would they listen to someone telling them to try again? “More effort” is only motivating when students can see how that effort might pay off. Otherwise, it just sounds like nagging. So feedback must offer a strategy worth trying. It has to show students what to attend to, what to change, and why this next attempt might be different from the last. The message shouldn’t be, “Try harder,” but, “Try this.”

3. Feedback must change with the phase of learning

Read more

I’ve no idea what I’m doing in the thumbnail! If you’d rather not see, you can ust listen here.

Schools are right to ban phones. They’re less obviously right to pay for a system that requires every student to carry round a locked version of the thing we’re trying to make irrelevant. A phone-free school should begin with culture, not kit.

The phone debate has moved quickly from fringe grumbling to mainstream policy. Jonathan Haidt’s The Anxious Generation helped turn parental unease into a movement, arguing for no smartphones before high school, no social media before 16, phone-free schools, and more real-world independence. In the UK, Smartphone Free Childhood has turned anxieties around the combo of phones plus social meida into collective pressure, while government guidance now encourages schools in England to prohibit phone use throughout the school day, including lessons, breaks, lunch and movement between lessons.

For years, schools were left to manage the consequences of a phone-based childhood while everyone else shrugged and muttered about preparing children for the real world, as if the real world were best understood as an endless queue of notifications, pornography, group-chat malice and algorithmic rubbish. Haidt’s great service has been to make the problem collective. One family denying a smart phone to the pre-teen children seems eccentric; a single school banning phones looks draconian but a whole community moving together starts to look like a possibility.

But is banning phones worth all the time and trouble? A new study from the NBER, The Effects of School Phone Bans: National Evidence from Lockable Pouches, gives an awkward answer. Banning phones definitely appears to dramatically reduce phone use, but that doesn’t seem to produce the broader gains we might expect. In class phone use fell from 61 per cent to 13 per cent in schools adopting phone bans, but the study finds average effects on standardised test scores “consistently close to zero”, with little evidence of improvement in attendance, chronic absenteeism, classroom attention or perceived online bullying. And, maybe more surprisingly, in the first year of a ban, disciplinary incidents rise and student wellbeing dips, although these effects do fade over time.1

So should we abandon phone bans as more trouble than they’re worth? Well, maybe not. It’s important to note that the study focusses on just one way of making schools phone-free: lockable phone pouches. A real problem has attracted a market, and the question of how schools should ban phones is too easily reframed as a choice between permissive chaos and commercial hardware: either students wander around with their own personal attention sink throbbing away in their blazer pockets, or schools buy lockable pouches, magnetic release stations and all the rigmarole that turns policy into procurement.

The researchers evaluated the effects of phone pouches using data from accross the US, including large-scale surveys, GPS pings, standardised test scores, school administrative records and sales records from Yondr, the largest pouch provider, which collaborated on the study. They used a staggered difference-in-differences design and found that while pouch adoption substantially reduced phone use, wider effects are much less impressive.

There are, however, reasons to treat these findings with some caution. Arguably, test scores are a poor measure of whether a phone ban is worth having. Of course academic outcomes matter, but not every worthwhile school decision shows up neatly in short-term attainment data. We don’t require a school to prove that banning vaping raises GCSE results before deciding Year 10 shouldn’t spend lunchtime disappearing inside mango-flavoured fog. Schools ban things for reasons of safety, dignity, order, attention, social trust and adult authority. It may also be that changes in classroom culture, social norms and student wellbeing take longer to emerge than policy evaluations can easily capture.

But there’s a more serious concern. If pouches reduce phone use substantially but produce only modest wider effects, that may tell us something about the limits of pouches rather than the futility of phone bans. It may suggest that containment isn’t the same as absence. Treating evidence about one commercial method as evidence against the whole principle is like trialling one behaviour app, finding it creates extra admin, and concluding that schools shouldn’t have behaviour policies.

A 2017 paper, “Brain Drain: The Mere Presence of One’s Own Smartphone Reduces Available Cognitive Capacity” asks a question pouch policies need to face: does your own phone impose a cost even when you’re not using it? Across two experiments, Ward et al. tested what they called the “brain drain” hypothesis, that the mere presence of one’s own smartphone may occupy limited cognitive resources, leaving fewer available for the task at hand.

In their first experiment, 520 undergraduates were randomly assigned to one of three conditions: 1) phone visible on the desk, 2) phone nearby but hidden in a pocket or bag, or 3) phone left in another room. Participants then completed tasks designed to measure working memory capacity and fluid intelligence. Participants performed best when their phones were outside the room, worst when phones were visible on the desk, and the hidden-but-nearby group fell somewhere between the two. The authors’ conclusion was that the smartphone’s availability may impose a subtle tax on cognitive capacity with subjects habituated to check whether their phone was receiving notifications.

The second experiment sharpened this point by varying both location and power status. If the problem were simply incoming notifications, then switching the phone off should have helped. The researchers found that location mattered a lot more than whether the phone was on or off: participants whose phones were outside the room again performed better on the working memory task than those whose phones were on the desk. The mere physical proximity of the phone appeared to be enough to disrupt attention and reduce cognitive capacity.2

Obviously we should treat all this with as much caution as the new phone pouch study. It’s certainly not supportable to claim that phones turn students into dribbling loons but there’s probably enough evidence, combined with everyday experience, to make the proximity problem worth taking seriously, especially when so much of education depends on attention, working memory and freedom from the tug of the peer network.

Although pouches stop students using phones, the fact that the phone is still physically present means it’s more likely to take up space in student’s minds. The pouch system bans phone ifrom being used but encourages students to bring them to school. The school day is phone-free, except every student carries a locked phone around all day like a tiny imprisoned goblin of distraction. This isn’t so much a ban as a compromise with a zip.

That compromise is better than doing nothing. If the alternative is a toothless “phones must not be seen” policy that everyone knows is barely enforced, pouches may improve things. But “better than chaos” isn’t the same as “best available model”. An egg and cress sandwich found under a car seat is better than starvation, but it’s not lunch at The Ritz.

If the aim is to protect attention, “locked but carried” is an uncomfortable halfway house. It deals with the behavioural problem of visible use more convincingly than it deals with the psychological problem of proximity. Students still have their phone — they know they’re there, pulsing with the lure of their online existence.3

There’s also a procurement issue. Phone pouches turn a cultural problem into a product. Schools face a dilemma: they know that phones undermine attention, behaviour and social life. Along comes a company with a branded solution. It provides the hardware, the implementation model and, crucially, the data. Research becomes easier to conduct on the branded solution than on more Heath Robinson alternatives because the company has records, adoption dates and a coherent package. Then, almost without anyone meaning for it to happen, the question shifts from “How should our school be phone-free?” to “Should we buy this thing?” Corporate interests colonise public policy through convenience, visibility and our weakness for solutions that arrive pre-packaged.

The best behaviour policies don’t come with gamified avatars, just as the best assessment systems aren’t the ones with the slickest dashboards, nor the seductive appeal of glossy resources the best guide to a quality curriculum. And the most effective form of phone ban may not come with magnetic unlocking stations, however pleasingly dystopian they may look.

None of this means Yondr or their competitors is exploitative. IAnd it doesn’t mean the NBER’s research is worthless. Pouches may have a place. However, we should be clear-eyed about incentives. Phone pouch companies are selling one particular answer to the phone problem.

A phone ban isn’t (just) a hardware problem. At heart, it’s a matter of school culture. How schools deal with phones — whether or not they purchase a pouch system — depends on norms, routines, clarity, adult consistency, parental understanding and leadership’s willingness to hold the line. Schools need to decide where phones belong, what happens when rules are broken, how exceptions are handled, and how students are taught the purpose of the policy. A pouch can support that culture, but it can also conceal its absence. It can allow leaders to say, “We’ve dealt with phones,” when what they’ve really done is buy a device to manage the symptom.

The best phone bans are boring. Phones aren’t seen, staff don’t spend the day negotiating, students aren’t invited into endless procedural theatre, and the rule is simple enough to become part of the weather of the school. Pouches, by contrast, make the ban visible every day: bring the phone, pouch the phone, carry the phone, unlock the phone, repeat tomorrow. That’s a lot of attention paid to something we’re supposedly trying to make irrelevant.

There’s also the practical problem. Pouches introduce a new compliance economy. They can be forgotten, broken, swapped, bypassed or tampered with; unlocking stations have to be supervised; exceptions have to be processed; transition points have to be managed. Students who resist the policy are no longer simply breaking a phone rule. They’re breaking a pouch rule, which gives schools a new class of incidents to record, sanction and argue about. The NBER paper’s finding of a first-year increase in disciplinary incidents is therefore unsurprising. Any serious phone ban is likely create friction at first, since very few people applaud when a highly addictive, social and entertaining temptation is removed. We’re unlikely to hear many students say, “Thank you, Miss, for liberating me from the tyranny of online dopamine hits.”

A good policy should reduce the number of daily transactions adults have to police. Pouches risk increasing them. They give everyone something else to inspect, enforce, manipulate, lose, damage, resent and game. That doesn’t make them unworkable, since schools manage awkward systems all the time, but it does make them less elegant than they first appear.

The alternative is not shrugging and hoping for the best. Nor is it a “phones must not be seen” policy that only works when adults have the time, confidence and energy to pursue every bulging pocket and suspiciously rectangular blazer. The alternative is to start with the principle that phones shouldn’t be part of the ordinary school day, then design the least fussy system your context allows.

Here are three cheaper alternatives:

1. Phones left at home

This has the obvious advantage of making phones genuinely absent. There’s no storage problem, no queue at the end of the day, no argument about damaged property, and no daily ritual in which the phone is formally admitted into school life before being ceremonially disabled. It also has the clearest cultural message: school is not a place for phones. The difficulty will be parental anxiety. Some parents want children to have a phone for the journey to and from school, and some students travel complicated routes or have caring responsibilities. This doesn’t make “left at home” impossible, but it does mean the school needs to win the argument with parents rather than simply announce it. The emergency objection also needs answering firmly: in an emergency, parents contact the school and the school contacts the child, as schools somehow managed to do before every Year 7 became a junior air-traffic controller.

2. Switched off and stored in (ideally) lockers or (less good) bags, with immediate confiscation if seen or heard.

This is cheap and simple. It doesn’t require new kit, and it may be enough in schools with strong routines and high adult consistency. Its weakness is obvious: the phone remains close. If the school is serious, the rule has to be brutally clear. Not “don’t use it”. Not “only at break”. Not “I was just checking the time”. Seen or heard means removed until the end of the day, with escalation for repeat offences. The advantage is that it keeps transactions low when culture is strong. The disadvantage is that, where culture is weak, it can become a game of pocket-policing, suspicious bulges and teachers pretending not to notice because they don’t want a twenty-minute row over a rectangle. All too often, this ‘policy’ gives students tacit permission to do whatever they want.

3. Tutor-time collection and end-of-day return

This is probably the best compromise where leaving phones at home isn’t realistic. Students hand phones in during morning registration; they’re stored in named envelopes, trays, numbered slots or tutor-group boxes; they’re returned at the end of the day. The phone is genuinely away from the student, which is the point pouches miss. The downside is admin. Schools need secure storage, clear routines, a system for late arrivals and early leavers, and a sensible process when someone claims their phone has gone missing or been damaged. It also requires a second afternoon tutor session which chips away at curriculum time. It’s not glamorous but it has the great virtue of matching the policy’s purpose: phones are not part of the school day whilst by-passing most parental concerns.

Secure year-group storage may be another option, especially in larger schools where tutor collection becomes unwieldy. Again, it isn’t effortless. Nothing involving hundreds of adolescents and expensive glass rectangles is effortless but it’s at least trying to remove the device rather than ritualise its presence.

Of course there will need to be exceptions but these should be genuinely exceptional. Some students will need access for medical, safeguarding or accessibility reasons. Those exceptions should be specific, documented and controlled. The point of an exception is to protect the rule from disregarding real needs, not to dissolve it into a thousand special cases. Once every inconvenience becomes an exception, the policy becomes a vibe.

None of these alternatives is perfect. But imperfection isn’t a reason to choose the most commercialised solution. Pouches are appealing partly because they look serious and tangible. They allow leaders to say they’ve “done something”, which is one of the most dangerous utterances in education.

A smarter phone ban starts with a simpler principle: school is a phone-free environment. Not “phones may be carried but not used”; not “phones may be used at break”; not “phones may be kept in pouches on students’ bodies”. Just this: during the school day, phones shouldn’t have a claim on students’ attention.

The real case for phone bans was never that exam results would instantly soar, behaviour logs would evaporate and Year 8 would develop the contemplative habits of a Shaolin monk. You wouldn’t expect the absence of wasps in classrooms to result in improvements in all these areas but every teacher is all too aware of havoc one of these winged terrors can wreak in a classroom.

Schools are one of the few remaining places where adults can say to children: for a few hours, you don’t have to be available to everyone else. You don’t have to respond, perform, monitor your status, inspect your humiliations, defend your streaks or keep one eye on the latest tiny drama unfolding in a group chat named after something unprintable. For many students, school may be the only part of the day in which the phone’s claims can be suspended. If we’re serious about attention, we should protect that space. If we’re serious about childhood, we shouldn’t treat permanent connectivity as inevitable. If we’re serious about authority, we shouldn’t allow every lesson to be haunted by a device designed by some of the most successful companies in human history to make itself utterly irresistible.

The wrong lesson from the NBER study is that phone bans don’t matter. They absolutely do.

Schools should ban phones because they understand the damage they do to children’s capacity to attend to the curriculum, not because a vendor’s offering them a great deal on the latests product. The aim shouldn’t be to lock phones up while keeping them close. The aim should be to make the school day a sacred space where the only claim on students’ minds is that of the riches offered by the curriculum.

Phones out of sight is a start but phones out of reach is better and phones out of the room is better still. And, where possible, phones out of school is best.

Their growing popularity has, inevitably, produced a backlash. Several commentators have begun warning teachers about the risks of overusing them, so it seems worth taking those concerns seriously.

First, here’s a post by no less an authority than Doug Lemov:

Here’s a summary of Doug’s downsides:

1. MWB writing is transient. Once they’re wiped away, students’ answers are unavailable for future reference

2. MWBs can encourage haste. Because the format rewards quick responses, students may write less carefully than they would on paper. Writing more slowly supports better word choice, deeper thinking and stronger encoding.

3. MWBs can become a crutch. Because they make participation feel simple they can become a crutch. Teachers still need other tools, such as cold call and stop and jot.

Let’s consider each of these limitations in turn.

1. The transience problem

I’ve always considered transience to be one of the great strengths of MWBs. They provide a midpoint between the complete transience of speech and the durability of writing.

Spoken language’s power stems from its immediacy. It allows us to try out ideas, respond to others, clarify uncertainty and participate before thoughts are fully formed. It’s socially supported: gesture, tone, facial expression, prompting and the surrounding environment all help meaning along. But that’s also its weakness. Spoken language is informal, fragmented and context-dependent. It disappears as soon as it is uttered. A student may say something promising, but unless it is captured, refined or rehearsed, it rarely becomes something they can use later.

Writing has almost the opposite strengths and weaknesses. Because it’s durable, it can be checked, improved and assessed. It forces students to make meaning explicit because the reader cannot rely on tone, gesture or shared context. Writing demands clarity, structure and independence from speaker support. But this durability also raises the stakes. A sentence written in an exercise book can have a daunting permanence. For many students, that makes participation slower, riskier and more exposing. They may write less because they are afraid of being wrong, or because the gap between what they can half-say and what they can properly write feels too great.

MWBs occupy a unique position between these two modes. They turn thought into visible language without making it feel final. They give teachers something more visible, more perceptible, than speech, but less intimidating and quicker to respond to than writing in books. Because errors can be wiped away, students are more willing to take risks, make corrections and improve their thinking. Because answers are written, teachers can see misconceptions, compare responses and shape next steps.

MWBs are, in my view, an essential pivot, helping students move from fleeting talk towards more durable writing. They’re a rehearsal space — a sandbox — where thought can be externalised, tested, corrected and sharpened before it’s either committed to the page and shared with a wider audience. They provide a means of making thought visible while it’s still forming. Much of what’s written on whiteboards should be wiped away. I encourage students to use them as place holders to help them articulate their ideas more fluently. The act of having said something aloud provides an internal durability that rarely comes from simply reading or hearing an idea.

Obviously, if we were to decide that whiteboards were an end point, Doug’s criticism would be more valid. If students only ever jotted on MWBs, much would be lost. As such, they should be seen as a stepping stone. If something’s worth keeping, it should be transferred into an exercise book once it has been checked and improved. The problem isn’t disposability it’s the understandable tendency to confuse rehearsal with recording.

2. The sloppiness problem

The sloppiness problem is real, but it’s a feature, not a bug. There are moments when rough, unconnected jottings are exactly what students need. When I ask an open question and then select a student to answer, it’s perfectly reasonable for that student not to have a polished response ready. Thinking time helps, but it doesn’t solve everything. Under pressure, students can freeze; ideas that were available a moment earlier can disappear under pressure.

MWBs reduce that problem, allowing students to jot down partial thoughts, key words, fragments of evidence or possible sentence openings before they are asked to speak. This gives them something to hold on to while they assemble an answer. Roughness is not always carelessness but a stepping stone to coherence. The board helps students bridge the gap between having a thought and being able to articulate it under pressure.

That said, there are other times — especially if I want students to demonstrate mastery of a process — when accuracy matters. If I’ve asked students to polish an answer, I’ll circulate to pick up errors. If I notice a missing capital, a misplaced comma or a spelling mistake I’ll point it out and ensure it’s corrected. Then, at the point where students have written something they can be proud of, I’d get them to transfer into their exercise books.

3. The crutch problem

Any tool has the potential to become a bad habit when it starts replacing professional judgement but to describe MWBs as a crutch is — I think — to miss the point. Some tools are essential. It would be absurd to expect a surveyor to make measurements without a theodolite or a surgeon to perform an operation without a scalpel. I feel the same way about MWBs in teaching. Could I teach without one? Yes, of course. But could I teach effectively without a MWB? Emphatically, no.

To that end, I’d dispute the idea that MWBs are a crutch. Crutches are only used to compensate for injury or weakness. Thinking of them as a crutch implies that only the unskilled would rely on them. MWBs extend a teacher’s perception and make visible what would otherwise remain hidden. But just because MWBs are an essential and invaluable tool in a teacher’s toolbox, does not mean either that they’re sufficient for every eventuality or that other tools aren’t also required.

The craftsperson analogy cuts both ways. A craftsperson needs many tools, but they also need to know which tools are unusually good for particular jobs. MWBs are no substitute for deep writing, extended thought or permanent notes but (for all their weaknesses) they’re the absolute best way of ensuring 100% participation, rehearsing for oral answers and providing opportunities for correction and refinement before committing to a more permanent response. The idea that MWBs somehow replace ‘cold call’ or targeted questioning seems odd to me. They are mutually reinforcing. If students have first jotted a word, phrase, quotation or sentence stem, their answers to targeted questions are usually more coherent. And, as far as I can see, ‘stop and jot’ is always best done on a MWB; it’s quick, provisional, visible and easy to correct.

All these ‘limitations’ are solved when teachers know why they’re using the tool. It should be clear that MWBS are the wrong tool when we want extended written fluency, make durable notes, sustain an argument or write at length. And, as I’ve written before, MWBs can and should be deployed differently to find out whether students are paying attention, making meaning or developing mastery.

Another potential source of misunderstanding about the limitations of MWBs stems from response on Twitter to a presentation given by Adam Robbins at researchED NYC.

I don’t want to cast aspersions on either Adam or North Landesman, but the tweet above confused me. I responded by asking, “Why would students just look at each other’s whiteboard? Struggling to see how or why this would happen.”

Without being at the talk, I can’t be sure but I think Adam’s point is that unless we have a structured ‘show me’ routine, students will simply copy the answers of those who have already written an answer, potentially misleading teachers into thinking students understand something they don’t. I guess this can and does happen but it doesn’t have to.

4. The copying problem

I try to make it clear to students that the reason I ask questions is to find things out. In general, I’m much more interested in wrong answers than right ones. Right answers may be reassuring, but wrong answers reveal hidden seams of misconceptions that need addressing.

One way I use MWBs is to check students’ perception of whether they understand a word. In a lesson on An Inspector Calls, we were discussing Gerald’s description of prostitutes as “dough-faced”. It became clear that one student didn’t seem to know what dough was, so I wanted to find out how widespread the issue was. I asked students to write 1 on their boards if they were confident they understood the meaning of dough, 2 if they weren’t sure, and 3 if they had no clue. Almost all the students showed a 2 or 3. That made the next instructional decision obvious: I needed to teach the word.

Now, perhaps some students copied a neighbour’s answer. I’ve no way of knowing for certain. But in this case, I can’t see that it would have made much difference. Unless a significant majority of the class secretly understood the word but were too embarrassed to admit it, I would have responded in exactly the same way. MWBs don’t need to provide perfect data to be useful. They just need to provide sufficient information to make better teaching decisions.

The important point is classroom culture. If students believe they’re rewarded for right answers and penalised for wrong ones, then of course they’ll look for ways to game the process. Copying, hiding, waiting and guessing are all rational responses to a classroom culture in which being wrong feels costly.

But I’m regularly surprised by how seldom students copy from their neighbours. I’ll often watch a student read someone else’s answer and then make no change to their own. I take that as a sign that they understand why I’m asking the question. If students are interested in understanding the curriculum (and almost all children are, whatever their outward behaviour might suggest) then they can see that it’s in their interests to show me what they actually think. A copied answer might look good in the moment, but it deprives me of the information I need to help them. Once they understand that, the incentive to ‘cheat’ is removed.

Of course, none of this happens by accident. It’s my responsibility to create the conditions in which honest answers are understood to be more valuable than correct-looking ones. That means rewarding students for giving me accurate information, treating wrong answers as useful, and praising the intellectual courage involved in taking risks.

Despite MWBs being one of, if not the, most useful tools available to teachers, they’re nomagical panacea. Used thoughtlessly, they can encourage haste, superficial participation and the illusion that everyone understands. But that’s true of any routine or strategy: cold call can switch between bullying and jazz hands; book work easily devolves into copying; class discussion can become distration and noise. The question isn’t whether a tool has limitations — all tools do — but whether teachers know why they’re deploying a particular tool.

MWBs aren’t a substitute for writing, thought or professional judgement. They are, however, a way of making thought visible while it’s still shapeable. They occupy a liminal position between speech and writing, uncertainty and commitment, first attempts and improved answers. They allow students risk being wrong without making mistakes permanent. They enable teachers to see what needs teaching before a misconception takes hold. The danger — as with any teaching technique — is forgetting why we’re using them. If they’re used to generate performance or tick a box then it should be no surprise that their use results in unintended consequences. But when they’re used to generate information, they’re hard to beat.

Timetables are created so that time becomes governable. Labels are assigned so that children, teachers and departments can be categorised in ways the institution can handle. Data is collected so that learning, behaviour and attendance can be viewed from a distance. Policies are written so that expectations don’t have to be reinvented every morning. Routines are established so that everyone knows what is supposed to happen next.

This is all eminently practical. A school unable to simplify complexity would be chaotic, exhausting and, very probably, unsafe. Children need order, teachers need shared expectations, and leaders need some way of knowing what’s going on beyond their own line of sight. Trouble begins when these necessary simplifications are treated as truths rather than tools.

Once we have simplified reality, those simplifications begin to acquire force. A label such as “low ability” may begin as a rough description of prior attainment, but it can soon affect the work a child is given, the questions they are asked and the explanations we reach for when progress stalls. A teacher thought of as “weak on behaviour” may be seen through that judgement even when the causes of difficulty sit in the class, the curriculum, the timetable or the school’s wider routines. A department thought to be “underperforming” may find that every piece of evidence is interpreted as confirmation of a settled story. What begins as institutional shorthand becomes an explanation, and eventually a self-fulfilling prophecy.

This is one of the most dangerous features of institutional life. Organisations do more than use categories; they begin to think with them. Mary Douglas, the great anthropologist of classification and institutional life, saw this clearly in How Institutions Think. Institutions, she argued, do more than govern behaviour; they govern attention. They teach us what counts as similar, what counts as different, which distinctions matter and which can be ignored. Over time, these classifications acquire the feel of common sense. They no longer feel like active decisions made by particular people in precise circumstances but like a natural expression of how things are.

Whatever else they are, schools are classification machines. They sort by age, subject, attainment, attendance, behaviour, need, risk, progress and performance. They decide what counts as a lesson, what counts as evidence, what counts as disruption, what counts as support and what counts as improvement. These classifications are unavoidable, but never neutral. Once they are built into registers, seating plans, data systems, interventions, reports and meeting agendas, they shape what we are able to see. They steer our attention, making certain questions feel urgent while pushing others out of view.

A child described as “low ability” is no longer encountered only as someone who did badly on a test or has not yet mastered a particular body of knowledge. The label gathers meaning around it. It may affect the curriculum she is offered, the explanations teachers give for her errors, the pace at which adults expect her to learn and the kind of future imagined for her. We may believe we’re recording factual information, but really we’re creating a shorthand for understanding the world.

Albert Michotte, the Belgian psychologist best known for The Perception of Causality, helps explain why this is so treacherous. In his famous “launching effect” experiments, observers watched simple moving shapes. One object moved towards another, touched it, stopped, and the second object began to move. Under certain conditions, observers did not merely report movement followed by movement. They reported causation: one object had hit the other and made it move. Change the timing, even slightly, and the causal impression weakened.

This is uncomfortably close to what happens in schools. We see low attainment and infer low ability. We see missing homework and infer laziness. We see a noisy class and infer poor teaching. Worse, we see evidence of poor teaching and infer a poor teacher. But in school life, as in Michotte’s experiments, what feels like direct perception of cause is often a story imposed on a pattern. We see effects: scores, behaviours, absences, silences, refusals, messy books and stalled progress. Causes have to be inferred, and those inferences are shaped by the categories we already use.

Douglas’s point is that institutions stabilise thought by embedding categories in routines. This helps because it saves effort. Teachers cannot rethink every distinction from scratch every morning, and leaders cannot run a school without shared names for recurring problems. But stability has a price. As Douglas puts it, “Institutions direct individual memory and channel our perceptions into forms compatible with the relations they authorize.” Once a category becomes institutionally useful, evidence is interpreted in ways that reinforce it. We notice what fits more readily than what troubles the category, and disconfirming evidence is discounted. The labels we use start as tools for coordination and end as artefacts of institutional memory.

Ordinary cognitive biases are recruited into institutional life. The halo effect means that one salient judgement spills over into other judgements. In schools, it often works as a kind of reverse halo. Weak writing is taken as evidence of weak thinking. Poor test performance is read as poor motivation. One difficult class becomes a judgement about a teacher’s general competence. A child’s reputation arrives long before the child. On top of this, the fundamental attribution error deepens the problem. Schools, like individuals, tend to overestimate the role of character and underestimate the role of context. Once a label is attached, behaviour is too easily treated as evidence of disposition. Struggle is read as weakness, resistance as attitude, inconsistency as incompetence, and underperformance as low aspiration. The explanation moves too quickly from observation offers to the inference of causes.

The cost is practical as much as moral. Dispositional explanations allow the school to move quickly from problem to person, but they also hide the mechanisms that might be changed. Rather than asking, “How do we fix this faulty individual?”, the better, though more awkward, question is, “What conditions are making these responses more likely?” This forces greater precision. If the problem is framed as laziness, weakness or low aspiration, the default response tends to be exhortation, surveillance or blame. If the problem lies in missing knowledge, poor task design, weak routines, unstable staffing or badly aligned assessment, the response has to be more exact

Take a child who is unable to punctuate speech accurately. The immediate problem is specific: she does not yet understand how direct speech is marked, how punctuation sits inside inverted commas, or how speech tags attach to what has been said. That problem can be taught. It can be broken down, modelled, practised, checked and revisited. If the child has already been entered into the school’s group mind as “low ability”, the instructional problem is easily swallowed by the category. Teachers stop asking, “What exactly has she failed to secure?” and start behaving as though the label has already answered the question.

Something similar happens with teachers. A colleague struggling with a difficult Year 9 group may need help with entry routines, seating, narration, modelling, practice, curriculum materials or confidence. The problem may sit partly in the class composition, partly in the timetable, partly in the curriculum, partly in the teacher’s habits and partly in the needs of individual students. Once the institution identifies the teachers as being weak on behaviour management, complexity collapses. A calm lesson is treated as an exception and chaotic lessons are treated as further evidence that this is a hopeless case. Although it’s a criminal waste of limited human resources, it’s far easier to jettison the teacher than address the systemic obstacles to effective behaviour management.

Bad ideas survive in schools because they have been built into the architecture. They sit in spreadsheets, meeting agendas, intervention schemes, seating plans and reporting cycles. Once a category becomes part of an institution’s ordinary language, it becomes hard to think without it.

The same danger applies to evidence.

Schools run on evidence. Leaders who ignore data and rely only on instinct will soon drift into fantasy. But school evidence is almost always tells us less than we imagine. A reading age gives us a trace of reading performance under particular conditions. A data drop records how students performed on selected tasks at a certain point in time. A behaviour log tells us which behaviours were noticed, recorded and coded. A book scrutiny gives us a sample of work, shaped by task design, curriculum sequence, student effort, teacher expectations and the distortions caused by knowing that books may be checked. A learning walk offers a brief encounter with teaching under conditions that change the teaching being observed.

Evidence becomes more seductive when it’s stripped of context. Once reality has been converted into numbers, those numbers slide easily through meetings, dashboards and reports, offering an impression of clarity that allows leaders to feel that they’re acting objectively. The danger is that numerical proxies begin to feel more solid than the reality they represent.

James C. Scott, the political scientist and anthropologist of state power, peasant resistance and administrative order, gives us the best account of this temptation in Seeing Like a State. Scott’s interest lies in what happens when governments try to make complex human arrangements easier to understand. “I began to see legibility as a central problem in statecraft,” he writes. Forests, cities, farms, names, occupations and populations are simplified so they can be surveyed, taxed, regulated and improved. The state needs maps, registers, grids, censuses and standard measures because action from the centre depends on simplification. But Scott’s warning is that simplification always comes at a cost. To make reality simple enough to administer, the state has to leave things out. Often, those things are important

Although Scott’s examples are not concerned with education, the pattern is immediately recognisable. A natural forest becomes a managed forest because the state prefers trees of the same species, planted in straight lines, measured in uniform units and harvested on a predictable schedule. A city becomes easier to administer when streets are straightened, addresses standardised and neighbourhoods made visible to planners. The simplification may be useful, but it also destroys forms of local knowledge that made the original arrangement work.

Schools have their own predictable ways to make it simpler for those in charge to pull levers in the hope of making meaningful change. Students’ progress is turned into grades, a term’s teaching into percentages, a department’s work into a spreadsheet, a teacher’s practice into a target, a child’s behaviour into a code entered in a Management Information System, and a curriculum into a sequence of boxes shaded green, amber or red. Although some of this can be useful, the danger comes when we treat proxies as the things they are meant to represent. Scott says, “The more static, standardized, and uniform a population or social space is, the more legible it is.” Legibility improves administrative control by reducing local intelligence.

Read more

Working with our feeder primaries, I collected each child’s best piece of Year 6 writing, turned it into a sticker, and gave one to every secondary subject teacher to place inside the front cover of the student’s new exercise book. The idea was simple: remind students what they could do and show teachers what to expect.

Sadly, it made no difference. As in previous years, standards slipped. I blamed entropy, surging teenage hormones, and the shift from one accountable teacher to many.

I now know I was wrong. It might be that teachers don’t know how well students can write, and might be that some students become less focussed on writing well as they move into secondary school, but neither are the problem. The problem is that the writing students master in Year 6 isn’t the writing they’re asked to produce in Year 7.

Writing transition is about mismatch, not handover

Writing transition is often seen as a baton-passing problem: primary schools run the first leg well, secondary schools take the handover, and somewhere on the bend the baton gets dropped. But, as we’ll see, transition doesn’t fail due to poor transfer (though that might also be an issue) It fails because of a mismatch in what primary and secondary teachers see as good writing.

The image below captures what I’d failed to understand. We were using the same word, writing, but we weren’t talking about the same thing.

In primary school, writing means expression, description, narrative, personal voice and the ability to develop ideas across a body of work. In secondary school, writing increasingly means constructing knowledge, marshalling evidence, sustaining argument, using disciplinary conventions and producing precise responses under tighter constraints. Neither conception is wrong. The danger lies in assuming they’re the same.

It’s not that primary writing is deficient because it doesn’t look like GCSE writing. Nor is secondary writing superior because it’s more abstract or disciplinary. The issue is compatibility. Without an adaptor, students can arrive in Year 7 with real strengths that don’t fit the new demands being placed on them.

That adaptor is effective writing transition: shared agreement about successful writing, sequenced knowledge and practice, explicit teaching of disciplinary writing, careful scaffolding, and assessment that tells teachers what to do next. Without it, transition becomes an exercise in hope. Primary schools hope their work will carry forward. Secondary schools hope students will adjust. Students are left to infer the rules. The result is predictable: different preparation, different expectations, different assessment, and then a sense that standards have slipped.

What is effective writing transition?

The questions on which effective writing transition depends are deceptively simple:

• Do we agree on what successful writing looks like in KS2 and KS4?

• Do we understand how writing is produced and assessed at different Key Stages?

• Do children in Year 6 know how writing demands will change in Year 7?

• Do children in Year 7 know what is still important from Year 6?

Too often, the answer to all these questions is no.

What successful writing looks like

The first transition question is whether primary and secondary colleagues agree on what successful writing looks like at KS2 and KS4. Too often, they don’t. Worse, they may use the same words while meaning different things.

In Year 6, successful writing often means sustained control across a body of work. Students are expected to write clearly, accurately and appropriately for different purposes. They need to organise ideas, select vocabulary, vary sentences, use punctuation accurately and show that grammar serves meaning. The judgement isn’t made from one piece produced under exam conditions. It’s built from evidence gathered over time.

At greater depth, this can look genuinely impressive. A Year 6 student might write a controlled narrative opening like this:

At the end of Marsh Lane, beyond the leaning trees and the rusted gate, stood an ancient, crumbling house which had been empty for years. Although no one had crossed its threshold since old Mr Vale disappeared, everyone in the village still whispered about it as if it were listening to their every word. By dusk, the crooked windows had turned as black as pools of ink, the tangled, overgrown garden had swallowed the narrow path, and even the birds seemed too frightened to land on the broken roof. Mina told herself that she wasn’t afraid; she was only curious. Step by trembling step, she crept towards the gate, clutching the tiny silver key in her hand. Then, before her fingers had even brushed the cold iron, the gate groaned open by itself. Curiosity became something colder, darker and far more dangerous.

This is good writing through a KS2 lens. It’s controlled, atmospheric and consciously crafted. The student varies sentence openings, uses fronted adverbials, selects ambitious vocabulary, creates a clear mood and sustains a coherent narrative perspective. The semi-colon is accurate. The expanded noun phrases do obvious descriptive work. The final sentence builds tension and shows an awareness of effect. It’s easy to see why a primary teacher might judge this as evidence of greater depth.

But this kind of success doesn’t automatically prepare a student to write a Year 7 history paragraph like this:

One important reason William won the Battle of Hastings was that his army was better prepared than Harold’s. Harold had already marched north to defeat Harald Hardrada at Stamford Bridge, before marching his tired army back south to face William. This meant the English soldiers were exhausted and had less time to prepare. Although Harold had the advantage of fighting on home soil, William’s careful planning, use of cavalry and archers, and ability to exploit English mistakes helped him win the battle.

This is also good writing, but for different reasons. The success doesn’t lie in atmosphere, voice or narrative control but in causal reasoning. The paragraph identifies a reason for William’s victory, explains how Harold’s earlier battle and forced march weakened the English army, and links Norman tactics to the outcome at Hastings. It uses “although” to acknowledge Harold’s advantage while maintaining a clear line of argument about why William won.

What goes wrong is when students approach Year 7 tasks without knowing that the goalposts have been moved:

After a long and exhausting march across the cold English countryside, Harold and his weary soldiers finally arrived at Hastings, ready to face William’s powerful Norman army. On the battlefield, beneath the grey autumn sky, the brave English warriors stood shoulder to shoulder with their shields locked tightly together, while William’s fierce men thundered towards them on noble steeds.

As the brutal battle raged on, Harold was struck by an arrow which peirced his eye. Their king slain, the English army crumbled. By the end of the bloody day, William had won and England would never be the same again.

This is a plausible Year 6 response because it has control, sequence, atmosphere and some deliberately varied sentence openings. It uses fronted adverbials, narrative movement and description. What it doesn’t yet do is think like history. It tells the story of the battle rather than explaining why William won.

By GCSE, successful writing has become more specialised and more pressured. In English Language, students need to write for a specified purpose, audience and form, usually in response to an unfamiliar prompt. They need to generate ideas quickly, control tone, structure a whole response, use ambitious but appropriate vocabulary, vary sentences for effect, and maintain accuracy under time pressure. In literature and across subjects, success increasingly depends on disciplinary writing: explaining causation in history, comparing processes in science, evaluating interpretations in English, justifying methods in mathematics.

So ‘good writing’ isn’t enough as a shared standard. Primary and secondary leaders need to ask a more precise question: good writing for what, in which subject, under what conditions, and with what knowledge already secured?

A child may write an effective narrative in Year 6 but struggle explain why some people continue to live near volcanoes in Year 7. That’s not proof the child has gone backwards. It shows that the task now demands different knowledge, vocabulary, sentence structures and relationships between evidence and explanation.

Thinking in terms of a monolithic notion of ‘good writing’ tempts schools to fix writing through bolt-on interventions: literacy weeks, a paragraph frame, an acronym, a checklist. These may look useful, but they rarely change competence.

How writing is produced and assessed

The second transition question is whether teachers understand how writing is produced and assessed at different key stages. Too often, they don’t. Worse, they may think they do because the same language is being used in both phases.

At KS2, writing is teacher assessed against a statutory framework. Teachers draw on a body of independent writing produced across the curriculum and over time. In theory, this sounds sensible. It avoids reducing writing to a single timed performance and allows teachers to consider what students can do across different purposes and forms. But the system also has distortions built into it.

KS2 teacher assessment doesn’t simply assess what students are capable of when they receive substantial support, feedback, redrafting and discussion. It assesses whether they can demonstrate particular features independently, across a body of work. That distinction can create perverse incentives. Schools may feel pressure to ensure that the required features are visible, even when they don’t serve the writing. Technique becomes evidence. Evidence becomes performance. The writing begins to carry the marks of the assessment framework rather than the demands of meaning.

This is how assessment can start to overvalue technique for technique’s sake. A semi-colon appears because a semi-colon is useful evidence. A fronted adverbial is inserted because it’s deemed to show range. Vocabulary is chosen to signal ambition rather than precision. Instead of grammar serving thought it becomes decorative. Children learn that good writing means displaying features, rather than making meaning with control.

There’s a second distortion. Grammar, punctuation and spelling are assessed separately. This separation may make measurement easier, but it risks teaching students that grammar is a body of detachable knowledge rather than a set of resources for shaping meaning. Students can learn to identify a subordinate clause in a test and still struggle to use subordination to qualify an argument. They can spot a modal verb and still be unable to use modality to express uncertainty, probability or judgement in writing.

At GCSE, the assessment conditions are very different. Students write under timed conditions, for an external examiner, without teacher mediation, without a second draft the following week, and without the benefit of a body of evidence. They can plan, draft and edit, but only inside the limits of the timed test.

This contrast changes everything. KS2 assessment rewards evidence of secure control across a range of work, but it can also reward visible compliance with a framework. GCSE rewards fluent, independent performance under pressure, but it can only sample what a student can do in a particular moment. One is cumulative. The other is immediate. One risks confusing feature-spotting with writing quality. The other risks mistaking exam performance for the whole of writing competence.

This is no to argue one way of assessing is better or worse than the other. The problem is that we tend to assume they’re measuring the same thing in the same way.

A Year 6 writing judgement may show that a student can write well when the content is familiar, the purpose has been taught, the class has discussed the task, and there has been time to produce several pieces across a period of weeks. It may also show that the student has learned to include the visible features the framework rewards. A GCSE response asks a different question: can the student do enough of this alone, at speed, in response to a fresh demand?

Good transition therefore distinguishes between supported performance, independent classroom writing and exam-condition writing. A student may write well after discussion but struggle without it. They may write well after modelling but falter when the model is removed. They may write well when they have time to revise but not yet manage the same quality under timed conditions.

If secondary leaders don’t grasp the conditions behind Year 6 success, they may mistake a shift in demand for a loss of ability. If primary leaders don’t understand later demands, they may overprepare students for polished, cumulative writing and underprepare them for time-bound, analytical independence. And if both phases accept the assessment system too uncritically, they may end up confusing evidence of writing with writing itself.

Treating supported competence, framework compliance, independent classroom writing and exam-condition performance as identical leads to false conclusions. It also leaves students stranded between two systems, each convinced the other has misunderstood the problem.

What Year 6 students need to know about Year 7 writing

The third transition question is whether children in Year 6 know how writing demands will change in Year 7. Usually, they don’t. They may know they’re going to have more teachers, more subjects and more homework, but they’re rarely told how writing itself will change.

They need to know that secondary writing will be more subject-specific. A history explanation isn’t a story. A science comparison isn’t a list. An analytical paragraph in English isn’t a personal response with a quotation attached. Each subject has its own ways of using evidence, organising ideas and deciding what counts as a good explanation.

They also need to know that secondary writing will make greater demands on independence. There’ll be less whole-class co-construction, fewer extended drafting cycles, and more expectation that they can plan, compose and check work with increasing autonomy. They’ll still be taught but the support will change shape.

Most importantly, they need to know that the sentence is where the new demands will show up first. A student who can’t connect two ideas in one sentence can’t build a paragraph. A student who can’t turn an adjective into an abstract noun can’t move from description to analysis. A student who can’t use because, although and therefore with precision will struggle to explain, qualify and argue.

That means transition work should give Year 6 students a taste of the kinds of sentences they’ll need in Year 7: sentences that explain causes, compare processes, introduce evidence, qualify claims, define concepts and link ideas. This doesn’t mean turning Year 6 into GCSE preparation. It means helping students see the next step.

Talk has a part to play here. Writing grows out of speech, but speech and writing aren’t the same. Speech is natural; writing isn’t. Writing must carry meaning without tone, gesture or shared context to help it along. So transition doesn’t need more talk in the vague sense. It needs better talk: structured discussion that rehearses the reasoning students will later write. If they can say it in academic language, they’re far more likely to write it well.

What Year 7 students need to remember from Year 6

Secondary school can make prior learning feel obsolete: new subjects, new books, new teachers, new routines. Students can get the impression that they’re starting again.

Year 7 should be a hinge. The job of transition is to make continuity visible and difference explicit. But that absolutely can’t mean telling students to keep writing as they did in primary school.

Much of the writing students do in primary is most directly applicable to creative and transactional writing in English. But in most secondary subjects, those qualities are either irrelevant or actively unhelpful.

Accross the curriculum, clarity, accuracy, explanation and analysis are prized more highly than voice or effect. A vivid sentence about a volcano is less useful than a precise explanation of why people continue to live near one. A dramatic account of the Battle of Hastings isn’t the same as a causal explanation of why William won. A confident paragraph is only valuable if it says something accurate, relevant and properly connected to the subject.

Students may bring real strengths from Year 6, but they don’t always know which of those strengths still apply and which need adapting. Accuracy, rereading, sentence boundaries, paragraphing, vocabulary, tense, punctuation and grammar are all still important. The habits developed in primary school don’t become obsolete because the timetable has changed. They are the foundation on which more specialised writing can be built.

There’s a further problem. Much of the grammatical knowledge introduced in KS2 withers on the vine in secondary school. Children may first meet subordinate clauses in Year 3, but will they still know how to use subordination in Year 10 to qualify an argument, express causation or weigh competing interpretations? They may have learned about modal verbs, expanded noun phrases, conjunctions and adverbials, but unless secondary teachers know how these tools work in their subjects, the knowledge gradually decays.

This is a system problem. Too few secondary teachers share the grammatical knowledge base introduced in primary. Even when they can recognise good writing, they may not be able to explain the choices that produce it. So the knowledge students acquired in primary remains inert: something remembered, if at all, for the GPS test rather than used to make meaning.

A student might know what a subordinate clause is and still not know how to use subordination to write, “Although Harold had the advantage of fighting on home soil, William’s planning and tactics helped him win.” That sentence doesn’t just sound more mature. It thinks differently. It concedes, weighs and argues. If students aren’t shown how KS2 grammar becomes secondary reasoning, the bridge is never built.

This is also why leaders need to distinguish between knowledge problems and habit problems. Not all mistakes are the same kind of mistake. Some mistakes occur because students don’t yet understand what to do. These need explanation, modelling and careful teaching. Other mistakes occur because students know what to do, but don’t yet do it reliably under pressure. These need practice designed to prevent mistakes from being made in the first place.

Capital letters are usually habit problems. Most students know that sentences begin with capitals and proper nouns need them. Ask them the rule and they can usually tell you. Yet in extended writing, they still omit them because the correct habit isn’t secure enough to survive the demands of composition. They’re thinking about ideas, spelling, handwriting, sentence construction and what comes next. If they’ve practised writing inaccurately for years, the wrong habit wins.

Commas, full stops and sentence control are often knowledge problems. Many students don’t know enough about how sentences work. They may not understand clauses, sentence boundaries or how punctuation shows the relationship between ideas. The old advice to “put a comma where you take a breath” doesn’t help. A writer doesn’t need a breathing rule. They need to understand how sentences are built.

So transition can’t rely on marking errors after they’ve happened. What students need is regular practice that stops them making mistakes. That means short tasks, clear expectations before writing begins, immediate feedback, accurate re-performance and repeated success over time. Practice should be narrow enough for students to get it right and frequent enough for the right response to become automatic. In short: we need to teach what students don’t know and practise what they do know until accuracy becomes normal.

Students transitioning from primary to secondary need to see that grammar isn’t a separate subject. It’s the machinery of thought in writing. A fronted adverbial isn’t a decorative opening; it’s one way of positioning information. Subordination isn’t a label to remember; it’s a tool for showing that one idea depends on another. Punctuation isn’t sprinkled on afterwards; it makes structure visible.

A school can have high expectations and still leave too much to chance. It can demand analysis without teaching analytic sentence forms. It can praise independence while withholding the instruction that builds it. It can assume KS2 grammar will somehow survive into KS4 without anyone deliberately using it.

It won’t. Knowledge not used becomes knowledge lost. Habits not secured become errors rehearsed. If Year 7 is to be a hinge rather than a rupture, secondary teachers need to know what students have been taught in primary, and students need to be shown how that knowledge now does different work.

What leaders should do next

If the answer to the four transition questions is no, then writing transition won’t be solved by better handover. It needs a shared model of progression.

Anecdotally, students in secondary often say something along the lines of, “I used to be good at writing but now I’m not.” This is a failure of both phases. Primary schools can leave students believing that lively description, accurate punctuation and visible technique are enough. Secondary schools can assume students will somehow infer the new rules of disciplinary writing. Between the two, students are left with the demoralising sense that what used to be a strength has become a weakness.

The truth is less fatalistic and more useful: they haven’t become bad writers. They’re being asked to do different things with writing. Our job is to make those differences explicit, then teach the knowledge, habits and structures that allow students to succeed under the new conditions.

Primary and secondary colleagues need to look together at real writing. Not to admire it but to work out what it depends on. What knowledge does this writing require? What vocabulary is doing the work? Which sentence structures are carrying the thinking? What has been explicitly taught? What has been assumed? What can the student do with support, without support, and under pressure?

That conversation should lead to curriculum decisions. Which sentence types should be mastered before students are asked to write whole analytical paragraphs? Which forms of explanation recur across subjects? Which habits from Year 6 should be protected in Year 7? Which new demands need to be named, modelled and practised?

The point isn’t to make Year 6 more like Year 11, or Year 7 a rerun of Year 6. It’s to build an adaptor between two systems, so students aren’t left forcing a plug into the wrong socket.

Leaders might begin with five questions;

1. Have you defined what successful writing looks like in each phase and subject?

2. Do teachers know the knowledge, vocabulary and syntax students need first?

3. Do students get enough guided practice before independent writing?

4. Is their practice improving competence, or just keeping them busy?

5. Does assessment tell you what to reteach, or only who can already write well?

Behind them all lies one bigger question: are you mistaking curriculum failure for student inability?

Effective writing transition isn’t a bridging unit, a summer project, or a laminated checklist passed between schools. It’s a shared theory of how writing develops. Primary leaders need to understand what’s coming. Secondary leaders need to understand what’s gone before. Both must stop treating writing as a portable skill and start seeing it as the outcome of curriculum, teaching and practice.

When transition works, students recognise what they already know, understand what’s changed, and are taught how to meet the new demand. When it doesn’t, writing progression is left to chance.

And anything left to chance is left to fail.

Classrooms run on questions. Asking questions and teaching are so closely interwoven that even young children seem to reach for them instinctively when showing someone else how to do something. Sydney Strauss argues that teaching is a natural human cognitive ability rather than a rare professional accomplishment. Children, he suggests, spontaneously teach peers and adults how to play games and carry out tasks, despite rarely being taught how to teach.1 This reminds us that questioning is one of the basic ways human beings direct attention, reveal intention and guide another person’s understanding.

One widely cited estimate suggests that teachers ask between 300 and 400 questions a day.2 Other summaries suggest questioning may occupy roughly 35 to 50 per cent of instructional time.3 By contrast, students ask very few questions in ordinary classrooms. Graesser and Person’s work puts the rate at about 0.11 questions per student per hour, a figure often rendered in later summaries as roughly one question a week.4 The exact numbers should be treated with some caution, since they are drawn from older studies and later syntheses rather than fresh large-scale observation, but the broad pattern is clear enough: questioning in classrooms is still mainly unidirectional. Teachers ask — students answer. Teachers initiate — students respond.

Classroom discourse has long been characterised by what researchers call recitation or IRE/IRF patterns: teacher initiation, student response, teacher evaluation or feedback. The typical student contribution tends to be brief, unevenly distributed, and followed rapidly by teacher judgement.5

If teachers ask hundreds of questions a day, questioning is not some niche technique we can refine at the margins but one of the chief ways lessons are structured, attention is directed, understanding is checked and participation is managed. If questioning is weak, a great deal of teaching goes wrong with it. If it’s skilled, it has the potential to be one of the most reliable mechanisms for leveraging thought and participation. Barrack Rosenshine argues that questioning is not just a means of sampling knowledge but a way to strengthen it: questions help students practise new information and connect it to prior learning. He distinguishes more effective teachers from less effective ones by noting that the former ask more questions, ask more process questions, and check the responses of all students, while the latter ask fewer questions and “almost no process questions”. In practice, a process question is one that asks students to explain how or why they reached an answer, thereby revealing the quality of their thinking rather than just giving the answer.

All teachers are told to ask more questions, ask probing questions, ask open questions, differentiate questions, use cold call, use no hands up, increase wait time. A lot of this advice is sensible but none of it is sufficient. Before deciding how to ask questions, we need to be clearer about what questions are for.

If modelling is the art of making expert performance visible, and scaffolding is the art of making difficulty manageable, then questioning is the art of making thought public. It allows teachers to check attention, secure understanding, develop mastery and prepare students to express ideas with enough precision to support more developed, independent responses. Used badly, questioning becomes a ritual exchange between teacher and student, a way of creating the appearance of engagement. Used well, it is one of, perhaps the most, powerful tools for ensuring students learn.

I’m not claiming the five principles that follow are the only things that matter about questioning, but they are the five that seem most useful to foreground, drawn from both my experience in classrooms and a reasonably thorough reading of the research.

1. Start with purpose.

2. Ensure everybody thinks.

3. Expect everyone to participate.

4. Treat answers as worth shaping.

5. Build sequences of questions that develop thought.

1. Start with purpose

The first principle is — I think — the most important: before asking a question, decide what it is for.

This sounds self-evident, but much poor questioning results from the fact that teachers often ask questions generically, as though all questions served the same function. Some questions are designed to check attention. Did students hear the explanation? Are they following the sequence? Have they noticed the feature the teacher wants them to notice? Other questions are designed to check understanding. Have students made sense of what they have heard? Can they connect the example to the concept? Can they explain the relationship between method and meaning, cause and consequence, claim and evidence? Still others are designed to probe mastery. Can students use what they know? Can they apply it in a new context, discriminate between near misses, explain why an answer is right and why an alternative is wrong?

What should be (but often isn’t) clear is that attention is not understanding, and understanding is not mastery. A class may look attentive and still be confused. A student may be able to repeat an explanation and still be unable to apply it. Rosenshine’s emphasis on frequent checks for understanding is best understood in this diagnostic sense. Effective teachers don’t merely ask questions to keep the lesson moving but to find out specific information.

Attention

When checking for attention, the best questions are usually short, closed and tightly tied to what has just been said. My rule is that I never ask this sort of question unless I’ve already taught the answer. Their purpose is not to explore the depth of students’ thinking but to confirm that they are mentally present and tracking the explanation. Most often, this questions is a variety of, “What did I just say?” In English, that might be, “What word have I just used to describe Macbeth’s state of mind?” In history, “Which did we say came first: the Restoration or the Civil War?” In science, “What term do we use for a substance with a pH below 7?” These are - very deliberately - not profound, but they’re useful because they quickly reveal whether students have noticed and retained what matters.

If a student is asked to repeat what the teacher, or another student, has just said, they have to attend closely enough to reproduce it. That is often enough to sharpen focus and strengthen recall. It also builds accountability into listening. These type of quests swim at the shallow end of the pool: Being able to repeat an answer tells us nothing about understanding, but failure to repeat it is a clear sign that understanding is a distant dream. The value of checks for attention lies in their immediacy; they tell you whether students are with you right now, not whether they might remember something elsewhere or later.

Meaning

Questions designed to check understanding need to go further. Here the aim is to find out whether students have made sense of what’s going on. These questions are more likely to be open or process-based.6 They ask students to explain relationships, causes, meanings or reasons. Often, I will press students to explain their thinking and deliberately test whether they are labouring under misapprehensions.

Students can often produce the right answer for the wrong reason. They may have memorised a phrase, inferred the expected response from context, or hit on something superficially correct without grasping the underlying idea. A student who says Macbeth is “guilty” may be right as far as it goes, but that tells us very little. Do they understand what in the language of the play suggests guilt? Can they explain why Shakespeare wants that guilt to be visible at this point? Can they distinguish guilt from fear, shame or paranoia? Until we press a little further, we may be rewarding verbal approximations rather than genuine understanding.

This is the purpose of process questions. Instead of merely asking what, they ask how and why. How do you know? Why do you think that? What in the text, diagram, example or method makes you say so? What would count as evidence against your view? If the answer changed, what else would have to change with it? These questions slow students down and force them to make the structure of their thinking visible. They’re useful primarily because they make misconceptions easier to detect. Asking these sorts of questions is challenging. Students need time and plenty of prior knowledge to do them justice. They should certainly not be randomly launched as a cold call gotcha.

We’ll return to this script later, but for the moment, consider this dialogue I observed in a maths lesson:

Teacher: What is the answer?
Student: x = 7.
T: How did you get it?
S: I took 3 away from both sides and then divided by 2.
T: Why are you allowed to do that?
S: Because you have to keep the equation balanced.
T: Good. So what does “balanced” mean here?
S: Whatever you do to one side, you do to the other.
T: Exactly. Would the same method work for 2(x + 3) = 14?
S: Yes.
T: Straight away?
S: Oh, no. First you’d have to deal with the bracket.
T: Why?
S: Because the 2 is multiplying everything inside, so you need to simplify that structure before solving in the same way.

I’m not holding this up as a perfect example of classroom questioning, but I do think it’s a great example of a teacher probing to working out a student’s understanding. After this exchange, there can be little doubt that, at this moment in time, the student is really clear on maintaining equivalence. A student who can explain the method, justify the method, and recognise its limits is in a very different position from a student who can merely provide the right answer.

The crucial thing is that these questions are diagnostic. Often I see questions asked as an invitation for students just to say something, to break up explanations and more lessons feel a buit more interactive. That’s not to say these aren’t good things, but they shouldn’t be the reason a question is asked. Good checks for understanding are stress-tests. A student may give an answer that sounds secure, but as soon as you ask them to justify it, connect it to prior knowledge, or distinguish it from something similar but distinct, weaknesses may become apparent. Equally, a hesitant answer may turn out to be more secure than it first appears once the student is given time and prompting to elaborate.

Here’s an example I overheard in a science lesson. The class had been asked to explain why different materials conduct electricity differently, and while students worked through the task independently the teacher moved around the room, stopping to probe individual answers.

T: OK, so why do metals conduct electricity?
S: Because electricity passes through them easily.
T: What’s actually moving?
S: The electricity.
T: What do you mean by “the electricity”?
S: Um… the energy?
T: Are the atoms moving through the metal?
S: No.
T: So what’s moving?
S: The electrons.
T: [pause] Electrons?
S: The free electrons.
T: Good. So is “electricity passes through them easily” really an explanation?
S: Not really. It’s the free electrons moving through the metal.

Without the follow-up questions, the first answer might have sounded good enough. In fact, it concealed a misconception. The student was able to produce a plausible response but needed prompting to explain underlying mechanism.

Mastery

Questions designed to probe mastery ask something more demanding again. Here the issue is whether knowledge is secure enough to be used independently, accurately and in a less familiar context. It’s one thing for a student to follow an explanation, or even to reproduce one they have just heard. It’s another for them to select the right knowledge without prompting, apply it with precision, and recognise when it fits a new problem. Mastery questions are aimed at that difference.

These questions often require transfer, discrimination or detailed elaboration. They ask students to judge, adapt, compare or apply.

• In English, we could ask, “Which earlier moment in the novella would you compare with Fezziwig’s party scene to show how Dickens develops his criticism of selfishness and social neglect?” Here students must choose a relevant comparison for themselves, justify that choice, and then use it to build an interpretation. Answering this question demands judgement, textual control, and a secure grasp of Dickens’ intentions.

• In science, a mastery question might be, “We’ve seen that increasing temperature speeds up the rate of reaction. If we kept the temperature the same but used smaller marble chips instead, would the reaction speed up in the same way? Why?” Now students have to do more than repeat the taught principle. They have to recognise that both changes may increase the rate of reaction, but for different reasons: higher temperature increases particle energy and collision frequency, whereas smaller chips increase surface area. To answer well, they must identify what’s essential in the original example, what’s changed, and whether the same explanatory principle still applies.

• In maths, mastery is often best tested through a carefully chosen task followed by a question about method. For instance: “A rectangle has a perimeter of 34 cm. Its length is 5 cm greater than its width. Find its dimensions.” On its own, this is relatively straightforward. But if the teacher follows it with, “What mathematical tool would be most useful here, and why?” it becomes a probe of mastery. Students must not only solve the problem but recognise that its structure calls for algebraic representation rather than a less efficient strategy such as guess-and-check.

• In history, a mastery question might be, “We’ve argued that nineteenth-century factory reform was driven mainly by fear of unrest rather than humanitarian concern. Does this source on the 1844 Factory Act fit that pattern, or does it challenge it?” This requires students to draw on first-order knowledge of factory reform and its context while also deploying second-order concepts such as causation, evidence and interpretation. They must use the source to judge whether the existing claim is confirmed, qualified or undermined.

Clearly, answering questions like these take time and effort. They can’t be asked on the fly. The need to be built towards across the course of a lesson or a sequence of lessons, so that students possess enough knowledge, and enough control of that knowledge, to answer with something better than a guess or a superficial shrug.

A mastery question tests whether students can recognise the deep structure of a problem rather than being misled by surface features. They reveal the difference between inflexible and flexible knowledge. A student may understand a method, a definition or an interpretation, but still lack the confidence or flexibility to deploy it independently. Or they may apply it too broadly, assuming that because something worked once it must work everywhere. Mastery questions bring those weaknesses to the surface by asking students to make judgements.

They’re also useful because they resist the illusion of knowledge created by fluent repetition. Students can often sound secure when they are reproducing something familiar. The real test comes when the scaffolding is pulled away. Can they still choose the right approach? Can they explain why one example fits and another does not? Can they transfer what they know without distorting it? Can they elaborate in enough detail to show that the knowledge has become precise rather than generic?.

Once you start to think in terms of attention, meaning and mastery, questioning ceases to be a background habit and becomes part of instructional design. At this point in the lesson, what do I need to know? Do I need to know whether students are still with me, whether they understand what has been taught, or whether they can now use that knowledge independently? A good question for one purpose is usually a poor question for another.

2. Ensure everybody thinks

The second principle is that students need time to think, and most classrooms allow far less of it than students need. The research on wait time has been around for decades. Mary Budd Rowe’s classic finding was that teachers typically wait less than a second after asking a question, and that extending wait time to around three seconds or more improves the length and quality of students’ responses. Later studies complicate the picture slightly. For instance, Baysen and Baysen’s study of prospective primary teachers found an average wait time of 2.36 seconds, suggesting some variation by context and culture, but still falling short of the 3 to 5 seconds often recommended in the literature.7

Thinking takes time. To answer a question, a student has to hold the question in mind, interpret what is being asked, search memory for relevant knowledge, decide what matters, organise a response and turn that response into speech. For novices, this is hugely effortful; knowledge is insecure, the language of the subject is unfamiliar, and students are often unsure what a successful answer sounds like. The more time we allow, the more time students have to come up with better responses and the more students will be able to answer the question. The point is to improve the quality of the evidence questions yield. Without a pause, we mostly find out which students are quickest to retrieve, quickest to guess, or quickest to take a risk in public. With a pause, we stand a better chance of finding out what more students actually know, understand, and can explain.

Our enemy is impatience. Silence feels awkward. We ask a question, nobody answers immediately — or a few hands shoot up — and the urge is to rescue the moment by rephrasing, simplifying, nominating too fast or answering the question ourselves. The result is predictable. The swiftest students dominate, while everyone else learns that sustained thought is optional.

The right amount of wait time will vary with the kind of question being asked. A quick check for attention should need only a brief pause. A question probing understanding will need longer. A question designed to test mastery may need longer still, and may require students to write or rehearse an answer before sharing it publicly. The point is not to fetishise a number of seconds, but to give sufficient time for the kind of thinking the question demands.

One useful technique is fill the wait time by repeating the question after a pause. Not because students didn’t hear or weren’t apyaing attention (although that’s certainly possible) but because repetition extends the thinking window and signals that everyone is expected to keep going. Silence, in this sense, is thinking time. If we want students to think, we have to make time.

Even so, wait time is can be opaque. While we’re waiting, or repeating, it can be hard to be sure what’s happening in students’ heads. Some may be thinking hard. Some may be confused. Some may simply be waiting for lesson to be over. One way to make wait time more purposeful is to build a think-pair-share routine around it.

First, while students think, I ask everyone to jot down notes and ideas on a mini whiteboard. That allows me to circulate, monitor what’s being written, and offer prompts or suggestions where needed. The thinking is no longer wholly invisible. MWBs turn hidden thought into something inspectable. Instead of waiting in the dark, I can see who’s on the right track, and who’s taken a detour.

Then, to increase the likelihood that everyone can give a more developed response, I might ask students to turn to a partner and share their ideas. As they talk, I listen in to as many exchanges as I can, paying particular attention to students who might otherwise be reluctant to participate. Finally, when it is time to share, I’ll often ask students to report what they heard from their partner rather than simply repeat what they themselves have just said. That small shift increases accountability, improves the quality of listening, and makes paired discussion feel more like preparation for whole-class thinking than a brief conversational detour.

Of course, think-pair-share is not automatically productive. Left too loose, it descends into a brief, noisy interlude. Its value depends on the tightness of the routine: everyone writes, everyone talks, the teacher listens in, and the whole-class share is structured so that paired discussion feeds back into the main intellectual work of the lesson.

Wait time should never be seen as an awkward gap between question and answer. Considered properly, it becomes part of the lesson structure and shifts questioning from the privileging of the quickest to benefitting the many.

3. Everyone answers

Every lesson should be a promise to students that everyone will be asked and expected to answer at least one question.

Read more

If you’d like the full backstory, you can find the earlier post below:

After running my first half marathon in October 2025 in 1 hour 49 minutes, I immediately decided I needed to double the distance.

The oft-repeated figure is that only 1% of people ever run a marathon. Whether or not that number stands up to scrutiny, it had a predictable effect on me. I wanted to be one of them.1

So I signed up for the Newport Marathon, one of the flattest courses in the UK, and began looking for guidance on what to do next. I decided I didn’t just waqnt to finish, but to aim for something ambitious. I looked up the London Marathon Good for Age qualifying times and found that, for my age group, I would need to run under 3 hours 7 minutes! Obviously, that became the goal.

When I worked out the pace required (4:26 per kilometre, or 7:08 per mile) I nearly backed out on the spot. At that point, I couldn’t run a parkrun that quickly. The goal seemed absurd. But absurd goals force us to stop daydreaming and start planning.

Since first having a crack at Couch to 5k back during COVID, I’ve been struck by the parallels between running and writing. Both require patience, routine, and a tolerance for boredom. Both are built gradually rather than conjured out of natural talent or good intentions. We’d never ask novices to run five kilometres without training, yet we often expect children to write at length with little preparation beyond encouragement. The appeal of the Couch to 5K idea is that it begins with what is manageable and builds, step by step, towards fluency and endurance. That is the case I make in Writing Fitness (due for release on 13 May). And if the writing students need to do at GCSE is equivalent to a 5K, then writing a book has rather more in common with the marathon: more preparation, more hard work, and a great deal more deferred gratification.

Everyone likes to think they have a book in them but far fewer are prepared to undergo the reality of writing one. Much the same is true of marathons. You might like the cachet and the suggestion of grit and nobility but the actual business of training is mostly repetitive, uncomfortable, and — often — a bit ridiculous. It consists of gettting up in the pre-dawn dark and going out day after day, whatever the weather. That, more or less, is also how books get written.

To try to make my audacious marathon goal even remotely plausible, I took two steps. First, I joined Running School at Bristol Physiotherapy. Second, I signed up for online coaching with Lee Grantham, whose videos I’d watched on YouTube. His coaching is not cheap and certainly not for everyone, but it seemed a better bet than trusting a generic training plan and hoping for the best. What I found most compelling about his approach was its scientific insistence on accurate diagnosis. Rather than guessing, he tests and measures in order to establish exactly what we’re working with and then breaks down a training plan to work on specific weaknesses.

That, too, seemed relevant to writing. The crux of my argument is that students are too often expected to produce extended performances before they have mastered the component parts, and that writing improves when it’s chunked into manageable, sequenced routines that prioritise clarity, control and confidence. The same principle applies to running. If you want to get better, you need to know specifically what to work on.

Running School involved gait analysis and a VO2 max test which, thanks to a measurement error, briefly had me convinced I was superman. A second round of testing at Bath University delivered a rather more sobering verdict. I also had a blood lactate test, which turned out to be more useful than any flattering fantasy and showed the points at which running stopped being comfortably sustainable.

The tests showed that below 5:15 - 4:37 mins per kilometre, or 153 bpm, I was running easily enough to recover well and build fitness without undue fatigue. At above 4:36 per kilometre, or 154 bpm, things become much harder to sustain, and above that fatigue would accumulate quickly. My VO2 max of 47.3 suggested a decent aerobic engine, while the more encouraging finding was that I ran with surprising efficiency. In simple terms, I use relatively little energy for the pace I run. Obviously, that would be important in the marathon. The practical lesson was that most of my training needed to stay genuinely easy, with harder efforts used sparingly and deliberately.

My training plan was a lesson in humility. I had to learn that I wasn’t going to improve hammering every run. I ran six times a week: Monday recovery (zone 1), Tuesday easy (zone 2), Wednesday intervals (zones 3-4), Thursday rest, Friday recovery, Saturday easy, Sunday long run (zones 3-4). If you’ve never followed a structured training plan these terms may be unfamiliar. The table below sets out what all this looks like for me.

Running in zone 1 was what I initially found hardest. I got a heart rate monitor to track my progress and if it went over 136 bpm I had to walk. Counterintuitively, the fitter I became, the easier it was to run at a heart rate of below 136 bpm. I can now, literally, keep going for hours!

I followed the plan with iritating earnestness. In fact, owing to some confusion on my part, I often did more than was asked because I added 20-minute warm-ups and cool-downs to everything.2

Before long I got used to the permanent sensation of being slightly ruined. Every morning I would stagger out of bed, hobble outdoors, and then discover, a few minutes into the run, that my body had once again decided to co-operate. Aches that made walking unpleasant would vanish when I started running.

For me, this marks a divergence between running and writing, although I know, not for others. I should say that unlike my approach to running, I’m an undisciplined writer. When I don’t feel like writing, I don’t write. When I’m moved to write it feels effortless. However, there are days when the simplest sentence feels as though it has to be dragged uphill. There are mornings when I suspect I’ve finally exhausted whatever limited talent I may once have possessed. Some writers have told me that for them, writing is the result of a disciplined process of sitting down at a desk and smashing out a word count. This is a useful corrective against the tendency to universalise from my own experience.

My approach to running embodies the argument behind Couch to 5K Writing. Most people don’t become better writers by just writing. They improve through gradual progression, deliberate practice, and repeated attention to the things that actually make writing work. Training for a marathon has — to my surprise — made me more sympathetic to those students for whom writing is like running up a steep hill on a cold winter morning.

As part of my race build-up, I ran the Longleat 10k, a distinctly hilly course, in 45:583, and then the Two Tunnels Half Marathon which I completed in a slightly disappointing 1:39:34. My watch, with its customary spirit of cavalier cruelty, predicted a marathon time of 3:30.4 Nowhere near my target. But the half-marathon had been run to a specific plan. The first 5k were conservative, then I was to keep my heart rate under 153 bpm so as not to drift over threshold too early. It felt controlled, perhaps over-controlled. I came away thinking there was a fair bit more in the tank. On balance, I was starting to doubt I could hit my goal time but thought I had a shot of coming close.

Then the whole enterprise fell apart. After a persisitant niggle over the week, the following Sunday I abandoned my long run after 5k because of sharp pain in my shin that wouldn’t ease off. I’d grown so accustomed to being able to run through niggling injuries, but this felt different. I consulted ChatGPT and it suggested medial tibial stress syndrome and that I shouldn’t run until I’d had it checked out.. A physio later confirmed the diagnosis using the more familiar term: shin splints.

This was six weeks before Newport. The physio told me not to run until it settled down and advised withdrawing from the race. He also gave me exercises and stretches, along with some foam rolling so brutal it seemed less like treatment than punishment for my hubris.

I was about to leave for Australia for three weeks, and the thought of not being able to run while I was away felt frankly miserable. I was told about the hop test: if I could hop five times on the injured leg without pain, I could run. In theory, this was reassuring. In practice, I never once passed it. My weekly mileage collapsed from 80-100 kilometres to less than 20.

At the start of any long project, whether it’s a marathon block or a book, the plan looks neat enough. There’s a schedule, a set of milestones, a pleasing fantasy of linear progress. Then life gets in the way: injury, fatigue, interruption, doubt, competing demands, plain bad luck. Suddenly, your elegant planning no longer fits the conditions you actually have to work in. When I was working on Making Kids Cleverer back in 2018 I abandoned it for almost a year. I just couldn’t face continuing with it in the face of some pretty grim life experiences. But what matters isn’t the elegance of a plan but whether it can be adapted to fit the circumstances in which we find ourselves. Eventually I did finish writing that book. Although I’d clearly have to change my plans, I was determined to do whatever I could to run this race.

To preserve some fitness, and avoid battering my leg further, I bought a second-hand bike and started cycling. Cyclists, I’m sorry to offend you but, oh my god! I’d have to be out for twice the time for half the benefit! But, after a couple of weeks it did seem to help. I managed a fairly brisk parkrun, then, a week before the marathon, I ran 37km slowly through the Mendips with a friend as a proof of concept. If I could do this then the marathon was on. Amazingly, my leg held up. Hope, against all sense, returned.

Like the Furies pursuing Orestes, the next day the shin pain returned, swift, punitive, and impossible to mistake.5 And it brought friends. A sharp pain appeared in the calf of my other leg, while the injured one added knee pain to the tally, as though my body had decided that if it was going to punish me for my hubris, it might as well be comprehensive. A sports massage pacified the calf, but the rest remained mutinous.

By the time I arrived in Caerleon for race weekend, I still didn’t know if I could run a marathon. In my heart I knew it was probably a mistake. On the Saturday I ran Cwmbran parkrun and it felt OK. Not great but bearable. Which, now I think about it, is a fair description of the final stages of editing.

Outwardly, I dithered, but really my mind was already made up. I would stand on the start line and see what happened.

On race morning I limped inauspiciously to my taxi, but after going through my pre-race preparations (coffee, bagels, panic poo, bag drop, etc.) I felt oddly… all right. Unfortunately I had somehow placed myself in the slowest wave, meaning most of the field went off before I was allowed to start. The 3:45 pacer was in the wave ahead. The marshals were utterly inflexible about letting me move up. I stood there watching the clock tick while my anxiety mounted.

Once released, I spent the first part of the race weaving through slower runners trying to find clear road. Despite abandoning the Good for Age fantasy, I still thought 3:30 might be a realistic target, which meant running at exactly 5:00 per kilometre. This felt well within my range. By around 15k I had passed the 3:45 pacer and felt superb. My shin pain and knee twinge had both vanished. At halfway I’d actually managed to run my second-fastest half marathon. I was taking a gel every 25 minutes and was full of bounce. I was even thinking about negative splits in the second half.6

For a while, it felt as though sheer exuberance and bloody-mindedness might be enough. Of course it wasn’t. Six weeks of not being able to train began to catch up. As for Orestes, my reckoning had only been deferred. At around 30km my quads and calves began to rebel. My pace began to ebb away. Runners I’d passed earlier came back past me. Then, at mile 24, crossing the bridge back into the city, my quad cramped so violently I had to walk to the top before I could start running again. Up to that point, I had still been running the race I’d hoped to run. Now I was clinging on.

The same thing sometimes happens to me when I write. I begin by imagining the beautiful confection I mean to produce only to end up with something forged under pressure, under constraint, under the everyday indignities of time and fatigue. I cut, I redraft, I persist. In the end, I finish what can be finished, not the shining idea that existed only in the glow of anticipation.

Once the cramp eased, I ran the final two miles as strongly as I could and even managed a sprint finish, passing other runners in the closing few hundfred meters. My final time was 3:38:44.

All things considered, I’m very proud of it.

Of course it feels bittersweet. It’s impossible not to wonder what might have happened had I been able to train properly through the final six weeks. But marathons aren’t run in the conditional tense, and books aren’t written there either.

We do what we can. We’re limited by the mind and body we have. We write with the time, energy and attention we can muster. We aim high because, sometimes, we exceed what might have otherwise been possible. A year ago, completing a marthaon would have been unthinkable. In the end the only thing that counts is putting one foot in front of the other, one sentence at a time.

Writing a book is the closest I’ve come to marathon running. Both demand the same unglamorous virtues: patience, repetition, self-discipline, tolerance for boredom, and the ability to continue when novelty has faded and only doggedness is left.

That is also why Writing Fitness feels, to me, like an apt title. The book, grew out of the premise that writing competence is built rather than bestowed, and that the route to confidence is through structured practice. The same turns out to be true of running 26.2 miles. With training and hard work you can rise to the occasion, or on whatever scraps of it injury allows you to retain.

I set out hoping to prove something impressive. What I learned instead was — possibly — more useful. Big things are done the same way as small things. One session at a time. One chapter at a time. One step at a time. One sentence at a time.

I wanted to run a marathon, and I have. Not according to plan, and not in the time I hoped, but I got to the start and then I got to the finish.

I’m moving to Phnom Penh in the summer and have just booked a slot at the Hanoi International Marathon on the 4th October. I’ll let you know how I get on.

The charge is by now familiar: GCSE English is too white, too male, and too dependent on a small canon of overused texts. Most GCSE English literature students still study the same few books: roughly three in four read Macbeth, more than four in five read An Inspector Calls, and A Christmas Carol remains one of the most widely taught nineteenth-century texts with about three quarters of students studying it.1 If teachers really cared about diversity, we are told, they would stop falling back on the usual titles and start teaching books that better reflect contemporary Britain.

At first glance, this seems hard to resist. Rachel Fenn of End Sexism in Schools has argued that optional attempts to diversify exam board text lists have changed very little because teachers keep choosing the same books year after year. If the menu expands but departments continue ordering the same meal, then the curriculum remains much as it was. And if only a tiny proportion of students study a whole text by a female author, and fewer still one by an author of colour, then clearly something is going on.2

But almost as soon as the argument is made, it starts to become incoherent. The mistake is to assume that a curriculum should be judged by how well it mirrors population level demographics. A curriculum is not a census or a diversity audit. Nor is it a display cabinet in which schools arrange texts to advertise their culture-war credentials. If the curriculum is a window rather than a mirror, its purpose is not to reflect the world back to children in a comforting likeness of themselves, but to induct them into forms of knowledge, culture and thought they would be unlikely to encounter otherwise.

When deciding which texts should be studied, the question is not whether a work improves representation, but whether it is worth teaching: whether it has literary merit, cultural significance, interpretative richness, and the power to enlarge students’ understanding.

It’s much easier to demand representation than to argue for significance. Easier to say a text belongs on the curriculum because of who wrote it than because it is so powerful, so illuminating, so influential that students would be intellectually poorer without it. But that is the standard that matters. If a work is to be added, the case for its inclusion should be made on its merits. If it’s to displace something already there, then the case must be stronger still.

This is not an argument for preserving the canon in formaldehyde. The canon should not sacred. Any serious culture argues about what should be handed on, and schools should too. It is perfectly reasonable to ask whether some writers have been neglected for reasons that have nothing to do with the quality of their work. It’s perfectly reasonable to challenge stale habits and inherited laziness. But if the curriculum is to broaden, it must broaden for intellectually serious reasons. And some texts matter in ways that are very hard to replace.

Shakespeare is not just another writer on a long list of possible options. His work is woven into the language, but the influence of the King James Bible may be deeper still. Its rhythms, cadences, imagery and phrasing have shaped English prose for centuries. Alongside the Book of Common Prayer, Homer, Beowulf, Chaucer, Milton, Dickens, Eliot, Austen and key political texts such as Magna Carta, Hobbes and Adam Smith, it forms part of the foundation on which so much later literature, public rhetoric and cultural reference depends. Without these works, much of what comes after becomes harder to understand. Remove them and children’s inheritance is diminished.

Observant readers may note that very few of these works are now routinely taught in schools. If we really believe the curriculum should introduce children to the best that has been thought and said, then where is Homer? Where is Sophocles? Where is Virgil, Dante, Cervantes, Goethe, Tolstoy? Where are the Upanishads, the Analects, The Tale of Genji, the Arabian Nights? The difficulty is not merely that the canon is too narrow, but that our actual curriculum is narrower still.

In practice, texts are rarely selected for the curriculum solely because they represent the highest achievements of human thought and art. More often, they are chosen because they are manageable: short enough to teach in the available time, accessible enough for large numbers of students to grasp, and familiar enough to support predictable exam performance. That goes some way to explaining the dominance of Macbeth, An Inspector Calls and A Christmas Carol. All three are relatively brief, heavily resourced, and well adapted to the practical constraints of the system. And, of course, individual teachers rarely get to choose which texts they would most like to teach. These decisions are usually made at school or trust level, where considerations of consistency, resourcing and results tend to matter more than literary ambition. Their popularity reflects institutional convenience.

You may admire J. B. Priestley’s An Inspector Calls, or you may not. That is beside the point. Set before the First World War and first staged in the aftermath of the Second, it is a perfectly serviceable play, and in some respects more than that. But no one seriously believes it stands among the very highest achievements of human thought and art. Its place on the curriculum owes less to indisputable greatness than to a more practical combination of virtues: it is good enough, short enough, and well enough resourced to fit the demands of a system in which schools are held to account for students’ exam performance.

This is the point both sides of the argument tend to miss. Progressives are often reluctant to admit that not all texts are of equal significance. Some works have exercised greater influence, achieved more, shaped more, and so make a stronger claim on our attention. A serious curriculum cannot avoid such distinctions. But conservatives are prone to an equal and opposite error. They treat the existing curriculum as though it were simply the natural expression of merit, when in reality it is also the product of habit, convenience and institutional constraint. Texts survive not only because they are good, but because they are known, resourced and safe.

So what would it take to displace Priestly from the curriculum? What texts could conceivably compete and still satisfy demands for greater diversity? There are plausible options already offered by each exam board but vanishingly few schools take them up. To encourage them to do so, An Inspector Calls would need to be removed from specifications. Here are some options which, when compared against AIC, might look tempting.

If exam boards and publishers put work into making these texts affordable and well resourced and AIC was dumped from the syllabus, it’s hard to see that students would be worse off in terms of cultural richness. My personal pick would be Zadie Smith’s The Wife of Willesden. I’ve used it before in Year 9 and it’s cracking fun and provides a splendid launchpad into Chaucer.

We’re all in favour of high standards, but standards of what? A text is not rigorous merely because it is old, nor lacking in rigour because it is modern. A school is not ambitious merely because it teaches familiar canonical works. Sometimes what passes for rigour is simply the comfort of the already known. Teachers know the familiar texts. They know the contexts, the critical commonplaces, the likely questions, the revision materials, the tried and trusted routines for turning reading into examinable prose. The path has been worn smooth. To teach a less familiar but equally worthwhile work well might demand more scholarship, more preparation and more confidence. It may be harder for the institution even if it is no easier for the child.

That is why so much talk of standards rings hollow. It confuses familiarity with seriousness. Schools should be ambitious and challenge does matters but this does not settle the curriculum question in advance. I for one cannot imagine a reality in which anyone can serious argue that Zadie Smith’s work is less rigorous than Priestley’s. .

So where does this leave us? With a harder argument than the one usually offered. Yes, the curriculum may be too narrow. Yes, schools may rely too heavily on a small group of overfamiliar texts. Yes, there may be writers and traditions that deserve much more serious consideration than they currently receive. But the answer isn’t to make representation the governing principle of selection. The real question is not whether the curriculum looks diverse enough. It is whether it is good enough.

That demands judgement. It demands criticism. It demands a willingness to say that some texts matter more than others, while also recognising that some of what we currently teach survives through custom rather than merit. A curriculum should not be designed to advertise virtue. It should be designed to expand inheritance.

It’s not possible for all children to see themselves reflected everywhere. To make this an entitlement is to reduce the curriculum to a checklist. They are, however, entitled to be taken beyond themselves, to experience worlds and ideas they might never otherwise encounter. Believeing the curriculum should be a mirror may only reflect our own preferences and prejudices. A truly liberating curriculum is a window into a wide world.

For other posts in this series, see below…

Imagine a city where the scaffolding is never taken down.

Years after construction has ended, every building is still wrapped in poles and boards. The boarded walkways remain. The warning signs remain. At first, this might seem prudent. Better safe than sorry, right? But it soon starts to look ugly and unfinished. Worse, it becomes clear that the scaffolds are not helping the buildings to stand. The scaffolding conceals the fact that no one quite trusts the buildings to stand on their own.

This, I think, is where much classroom talk about scaffolding has ended up.

Scaffolding is one of those ideas that enjoys such universal approval that it’s hard to use with precision. We all claim to scaffold but we don’t necessarily all mean the same thing. A writing frame is called scaffolding. So is a checklist, a model answer, a sentence starter, a prompt, a hint, a success criterion, a chunked task, or simply a teacher hovering nearby at the right moment. When a term covers this much ground, it starts to lose its meaning.

Once scaffolding comes to mean almost any form of help, we lose the distinction between support that leads towards independence and support that merely enables task completion. The latter may produce neat work, but it also conceals dependence and fragility. Wood, Bruner and Ross (1976) define scaffolding as a process which “enables a child or novice to solve a problem, carry out a task or achieve a goal which would be beyond his unassisted efforts”.1 Van de Pol and colleagues have sharpened this definition by identifying three core features: contingency, fading, and the transfer of responsibility.2 That gives us a useful starting point: scaffolding is not simply any form of help. It’s temporary, responsive support that makes possible what students cannot yet do alone and is gradually withdrawn as responsibility shifts to them.

Builders don’t erect scaffolding to make a simple job a little more comfortable. They put it up because the work cannot be done safely or successfully without it. And as soon as possible, the scaffolding is taken down and packed away.

We often talk as if scaffolding were the opposite of struggle, as if support and difficulty occupied opposite ends of a continuum and good teaching consisted in finding the most comfortable route between them. But that is not right either. Struggle is only useful once students have first experienced enough success to believe improvement is possible. Too much struggle backfires; too little leaves students fragile and dependent.

We also need to address the zone of proximal development (ZDP) that time-worn incantation wheeled out whenever someone wants to sound theoretically grounded. The problem, as Seth Chaiklin’s analysis make clear, is that ZPD is too often reduced to a Goldilocks fantasy in which every task has a perfectly calibrated level of challenge and every good teacher somehow finds it. Chaiklin argues that Vygotsky’s concept was developmental, not a general recipe for making all classroom tasks feel “just right”. That ought to make us more cautious about using ZPD as a catch-all justification for any form of support.3

For scaffolding to be a useful concept, it needs to be given shape and structure. I think adhering to five straightforward principles can help.

These principles are:

1. Scaffold only where success would otherwise be out of reach

2. Secure early success before increasing struggle

3. Maintain challenge while reducing needless complexity

4. Build every scaffold with a plan for taking it down

5. Adjust support in response to evidence

The first of these two principles are discussed before the paywall, the final three can be read using either a free trial or for the very reasonable porice of just £3.50 per month or £30 per year.

Subscribe now

1. Secure early success before increasing struggle

Students are unlikely to persist with difficult work unless they have some reason to believe success is possible. Repeated failure doesn’t usually build resilience. More often, it teaches students that the task is beyond them. That is one of the clearest implications of Bandura’s work on self-efficacy. Our beliefs about what we are capable of shape the choices we make, the effort we put in, how long we persist, and how we interpret setbacks. Most importantly, As Bandura puts it, “The most effective way of creating a strong sense of efficacy is through mastery experiences. Successes build a robust belief in one’s personal efficacy.” If students repeatedly encounter failure they cannot interpret or escape, they are more likely to infer incapacity than possibility.

According to Rosenshine’s Principles of Instruction, when material is new, teachers should present it in small steps, guide practice, check for understanding, and aim for a high success rate in the early stages. This is regularly misinterpreted as making sure that 80% of students are successful. This is absolutely not what Rosenshine says. To do so would write of the fifth of students who struggle most. What he actually says is this:

Unless all students have mastered the first set of lessons, there is a danger that the slower students will fall further behind when the next set of lessons is taught. So there is a need for a high success rate for all students.

So, while it’s true that in the the most successful teachers’ classrooms, 82% of students answer are correct, I’d interpret the success rule as, 100% of students need to get 80% of answers right.

But again, the point isn’t to make work easy. We need to establish the conditions in which later difficulty can become productive rather than merely demoralising. Early success gives students something to build from: a mental model of what good performance looks like and a reason to believe that effort may pay off. Struggle should never be abolished, but if does require careful sequencing.

There is a broader cognitive case for this too. Kirschner, Sweller and Clark argue that minimally guided instruction is less effective and less efficient for novices because beginners lack the stored knowledge needed to guide their own performance. The advantage of guidance, they say, “begins to recede only when learners have sufficiently high prior knowledge to provide ‘internal’ guidance” (2006, p. 75).

What this means for scaffolding is simple. Novices are rarely helped by being left to struggle without enough knowledge or guidance. Until they know enough to steer their own thinking, what looks like independence is often just confusion. Scaffolding is not the removal of challenge, but the temporary support that makes challenge possible.

So before we demand independence, we need to help students form a clear mental representation of what successful performance looks like and how it is achieved. That may mean explanation, modelling, guided practice, and temporary support sufficient to get them over the line at least once. This isn’t hand-holding. It is the necessary precondition for later autonomy. Students are far more likely to persist when they can see the shape of success and have some experience of attaining it. Without that, exhortations to be resilient are often little more than a demand to endure bafflement politely.

Only then does it make sense to withdraw help and require the kind of effort that strengthens learning. This is where Bjork’s idea of desirable difficulties becomes useful. Some forms of difficulty improve long-term retention and transfer, but only when they are desirable rather than merely destructive. Bjork’s own formulation is blunt: the learner must be equipped, by virtue of prior knowledge and current cues, to succeed, otherwise the difficulty becomes undesirable. So difficulty is not automatically a good. It becomes good when it follows enough success to make further effort fruitful.

That helps clarify the role of scaffolding. Scaffolding is not there to abolish struggle, nor to keep students comfortable. It is there to ensure that struggle has somewhere useful to go. First, students need enough support to experience success and to grasp how success is produced. Then support can be reduced, so that remembering, selecting, sequencing and explaining begin to demand real effort. The aim is not smooth performance but growing competence. Effective scaffolding protects students not from difficulty itself, but from the kind of difficulty that teaches them nothing except that they are failing.

A simple example might come from science. Suppose a Year 8 class is learning to write a method for separating salt from sand. If students are asked straight away to produce a full written method, some will fail not because the task is inherently beyond them, but because they haven’t yet secured the sequence, the vocabulary, or the conventions of scientific writing. In that case, the right move is further explanation, a model, perhaps a partially completed method, and guided practice until students can succeed. Only after that should support be reduced and students expected to produce the method unaided. If they struggle then, that struggle is far more likely to strengthen learning because it rests on something already understood.

The principle, then, is simple enough: secure success before expecting students to work more independently. No one becomes more capable by being left to fail repeatedly. They become more capable when early success gives them the confidence, knowledge and mental model needed to profit from more challenging independent work.

2. Scaffold only where success would otherwise be out of reach

Scaffolding should not be used to help students complete task that are merely challenging. Negotiating challenge is a valuable learning experience. Instead, scaffolding should be reserved for work that is currently beyond students’ independent reach.

This sounds straightforward, but it requires a judgement call. We have to distinguish between students being unable to do something and them finding it hard. A student who hesitates doesn’t necessarily need a scaffold. A student who struggles to begin may not need a sentence starter. A student who finds the work taxing may not need the task simplified. Sometimes what’s needed isn’t support but insistence, practice and time. The existence of difficulty is not, by itself, an argument for scaffolding. We should resist the popular belief that all successful teaching must hover in some perfectly moderated zone of manageable challenge.

The practical question is not, “How can I help?” but, “What is stopping students from succeeding?” Is the problem missing knowledge? Too many simultaneous demands? Unfamiliar vocabulary? Missing prior knowledge? Difficulty sequencing thoughts? Or does the task just require effort that should are disinclined to make? If it’s only the latter, then support does more harm than good. Scaffolding should make the impossible possible, not the easy easier.

A Year 10 English class is given the task of writing a response to the essay title, How does Shakespeare present Lady Macbeth’s changing relationship with power? Some students start churning out word, other’s twiddle their pens. Our first job is to find out why they haven’t started wrting. Do they understand the question? Can they explain an answer aloud? Do they know which quotations they might use? Can they organise an argument into a sentence? Or are they simply reluctant to make the effort of beginning?

Those distinctions matter. If students can talk through a sensible answer but can’t yet shape it into analytical prose, then a scaffold may be useful: a partially completed paragraph, a sentence stem for embedding evidence, or a paragraph frame with one section missing. But if they already know what to say and how to structure it, then more support is the wrong response. What they need is to begin, and to understand that choosing not to do so will have consequences.

Obviously, these are questions we ought to be able to answer in advance of setting the task. If we don’t know the answers, the likelihood is we’ve moved too quickly to independent writing and need to spend more time making sure all are competent with the prerequisites. If we're at all unsure, we should return to Principle 1: Secure early success before increasing struggle.

Scaffolding must be used sparingly and diagnostically. It should be reserved for the point at which students cannot yet proceed without help, not handed out whenever work feels difficult or effort falters. Used well, it makes success possible. Used carelessly, it merely insulates students from exeriencing the very struggle through which success is built.

Read more