Nothing was wrong. The lesson was going to plan. I’d taught the key points, the students had done a mini-whiteboard exercise and most of them seemed to get it. They were now waiting for me to explain something else.

But something was missing. It was the spark that I’d felt the first time I stood up in front of a bunch of thirteen-year-olds without knowing what was going to happen. The excitement that comes not simply from a properly run lesson but connecting. Helping a student to understand something they never knew before, hearing that creative response that comes out of the material you’ve taught but needed that specific student’s input to bring it to life.

When people talk about teaching being a vocation – the bit that makes up for half your salary – that was what had gone.

I noticed the same thing happening again and again: the lessons were fine but something was missing.

I could do the job; that wasn’t the problem. It was something deeper. Then I remembered I was doing a PhD about the role of curiosity in education. Why not try to use some of what I’d learnt in my own classroom?

Last week, I published a piece for Edutopia about evidence-based strategies for bringing curiosity to the classroom. But that was only the beginning.

Tiny Experiments in the Classroom

Around the same time as my existential crisis in the classroom, I came across author and curiosity researcher Anne-Laure Le Cunff’s book Tiny Experiments. Le Cunff argues for approaching life experimentally: prioritising process and learning over an obsession with outcomes.

I thought I could apply it to teaching. We often start out with the end in mind – raising grades or improving attendance or reducing exclusions. Things we can put a number to. That’s understandable; it’s how schools themselves are judged. But as a classroom teacher, what matters most to me is the minute-to-minute feedback. That feeling I had – that the spark had gone out of my teaching – was the problem I wanted to address.

The point of the Tiny Experiments in the Classroom isn’t to raise grades – although there are good reasons to think that when students are curious, they’re likely to learn more and retain that information better. It isn’t to improve behaviour, although if students are more curious, in my experience, they’re more likely to be focused. Every experiment is designed to make the kids more curious, but that isn’t the point.

For me, at least, the purpose is to help me enjoy lessons more. I enjoy lessons when the students are curious. So, their enjoyment is a means to my own satisfaction and vice versa. By enjoyment, I don’t mean going back to making posters and doing wordsearches – I mean deep and meaningful intellectual engagement with the subject – both on my part and my students’.

It might seem a bit selfish, putting my own feelings at the heart of the changes. But, as I know from my own experience, teaching isn’t something you can do half-heartedly. Apart from anything else, the kids can sense it.

This wasn’t the first time I’d lost my love of teaching. I quit altogether after the pandemic. Both times, curiosity helped me to rediscover it – first, beginning a doctorate to work out what curiosity in the classroom is; now, trying to bring the curiosity back into my teaching.

My Edutopia piece took three situations where I’ve applied curiosity research in my own classroom. I’m going to do things a little differently here. One situation – delayed closure – I’ve discussed here before. We’re going to explore the evidence behind the other two, and I’ll show you them in action.

The Curiosity Triggers

Just before the summer holidays, I had to teach about electrical safety to my Year 9s (age 13-14). A lot of the material focuses on the function of the earth wire – you can use a lot of nice cartoons of someone being shocked, but the slide deck I pulled off the system didn’t seem to have much to trigger curiosity. So I turned to some of the foundational, and still the most powerful, ideas in the psychology of curiosity.

Daniel Berlyne – a pioneer of curiosity research – identified facets of a stimulus that can cause curiosity. He called them ‘collative variables’ but I’m calling them ‘curiosity triggers’ when applied to classroom context. Here are the four most commonly cited:

• Novelty – something new or unfamiliar

• Complexity – something with multiple elements or relationships to work out

• Uncertainty – where the answer or solution isn’t immediately clear

• Conflict – clashes with what we expect or previously believed

I came up with a quiz for my Year 9s. Which of these would you least want to get a shock from?

• Lightning

• Toaster

• Electric fence

• Phone charger

This technique draws too on ‘delayed closure’, in which the active ingredients are getting the students to make a prediction and write it down – but avoid giving them the answer straightaway.

I taught students that potential difference is related to the energy each electron in a circuit is given (technically, the energy per unit charge). The bigger the potential difference, the more energy can be carried around a circuit. Then I revealed the size of the potential differences in the four situations:

• Lightning: 100 million–1 billion volts

• Toaster: 230 V (UK mains)

• Electric fence: 2,000–10,000 V

• Phone charger: Typically 5 V

There were looks of surprise about the size of the electric fence’s voltage. And lots of questions. Some of which I deferred answering – I didn’t want to spoil the surprise.

‘Go back to your predictions,’ I said. ‘Now which do you think is most dangerous?’

The key is that voltage is only part of it. You need to know the current – which depends on your body’s resistance – and how long it’s sustained for, to work out the danger. Roughly, 1mA is the point where you can feel a current. Over 100mA is where it can be fatal.

Lightning is, of course, the most powerful. Around 90% of those struck by lightning survive, though it can still cause neurological damage in survivors.

Mains electricity can be more dangerous than lightning, especially if it goes from hand to hand through your chest. If you’re gripping something, the shock causes your muscles to constrict, meaning you can’t release the source of the electricity, and the current can interfere with the heart’s normal electrical activity. People often survive a brief shock; sustained contact can be fatal.

An electric fence causes a painful but brief and low energy shock. It’s virtually never fatal although can be dangerous for those with a pre-existing heart condition.

Finally, a phone charger is too low a voltage to be dangerous – unless you grab the mains side, in which case it becomes the toaster problem.

The extra complexity could be a problem for a class which struggles with new information, but this was a high ability group, and they had no difficulty making links to ideas we’d studied previously, like V=IR: a formula for calculating current from voltage and resistance.

As a Tiny Experiment in the Classroom, it felt like a success. Hearing their ideas about the dangers and seeing the shock on their faces at each stage reminded me why I love doing the job. It wasn’t a full lesson redesign but it took work and a bit of research (which was part of the fun). But you can add novel or surprising facts into your lesson with a lot less effort.

Question Generation

When I saw the lesson plan, my heart sank: another literacy exercise – this one about ecology. I checked the reading ages of my Year 8 (age 12-13) class and saw quite a few were working at an age several years different from their own – below as well as above.

We’ve started doing more literacy exercises in school, but keeping a group with such different abilities cognitively engaged is a challenge. So I drew on one of the few trials into curiosity that has been carried out in the classroom.

Harvard’s Shelby Clark, together with colleagues Allen G. Harbaugh and Scott Seider, conducted a study with over three thousand students to investigate whether directly teaching students how to formulate questions could increase curiosity. For students in the intervention group, their teachers used the Question Formulation Technique:

1. Present students with a prompt.

2. Students produce as many questions as they can, without judgement as to quality or discussion over them.

3. Students can then discuss and change open into closed questions and vice versa.

4. Students select three ‘priority questions.’

5. The class reflect on what they’ve learned during the process and decide which questions to take forwards.

Teachers in the intervention group could use this as often as they liked. The more teachers used the technique, on average, the higher their students’ self-reported curiosity.

With my group, I didn’t use the full technique but borrowed the basic mechanism, using the following prompt:

In pairs, write three questions you’d want answered about how losing whales or birds, or snow leopards would change the world. Three minutes. Any question counts. Daft, weird, obvious, all welcome.

Here’s some of what they came up with (the questions are real; the whiteboard image is an AI recreation):

The questions weren’t all relevant to ecology, but the note I wrote to myself immediately after the lesson read: ‘took some a while to get going but better ownership of literacy task and engagement after’. Overall judgement? ‘Was actually much more enjoyable than anticipated.’

Breaking the Script

Do we have an evidence-informed profession, or one in which our actions are dictated by the evidence? Does the evidence shape what we do or constrain it?

I’ve argued that some schools are moving into an era of mandated pedagogy which I’ve called ‘Total Instruction’. Teachers’ autonomy is restricted under the auspices of ‘evidence-based practice’, which often comes down to techniques – elements of which often do have some research backing – that are mandated as a whole script that teachers are required to conform to.

Despite that, most teachers still have agency. Teachers can plan and deliver lessons tailored to the students in front of them. We need to make the most of this freedom – so that teachers can retain their love of the profession while improving their practice for the students in front of them.

Tiny experiments in the classroom are not academic or scientific studies. We’re not trying to ‘prove’ anything that can be peer reviewed and published.

Nor are they action research. I love action research, but it’s often more systematic, drawing on relevant literature and producing evidence to answer a specific question. Tiny Experiments in the Classroom are simply a way to improve today’s lesson.

We want the experiments to be evidence informed. Begin with an idea grounded in research and adapt it for your own students. Keep ‘field notes’, like an anthropologist would do, Le Cunff suggests in Tiny Experiments. Write a couple of lines on a note on your phone after the lesson. How did it go? Then onto the next thing.

At some point, you can check back and see what’s working. Did I enjoy the lessons more? How did the students respond?

This can help shake us out of our cognitive scripts: we have a sense of how things are going to play out in the classroom. We have good classes, bad classes, favourite lessons and ones that bore us.

The Tiny Experiments help us question our assumptions, our relationships with students and even our subjects.

Maybe they can even help us rescript our relationship with the profession.

Many think they might have found it.

In 2019, England’s inspectorate Ofsted released an ‘overview of research’ which introduced the ‘Learning Sciences’, ‘a relatively new interdisciplinary field that seeks to apply understanding generated by cognitive science to classroom practice’. This ‘understanding’ included spaced practice, retrieval, interleaving, dual coding and cognitive load theory.

The same year, the government introduced a ‘minimum entitlement’ (latest version here) for initial teacher training – effectively, a curriculum for trainee teachers. It became mandatory for all teacher training providers to cover key aspects of the learning sciences.

Then there is the Education Endowment Foundation, which, like its US counterpart the Institute for Education Sciences, funds large-scale trials in an effort to discover ‘what works’ in schools. If you go onto the EEF’s webpage, you can find out exactly how much an individual intervention will help your students via the Teaching and Learning Toolkit. It’s a shopping list of improvements for school leaders.

A diverse range of developments – evidence backing up an inspection framework, a mandatory teacher training entitlement and a summary of the evidence which doesn’t force anyone to do anything. And I’m going to widen the net further as we go.

What they have in common is the fact that they are all converted into policy by schools. Some things schools buy in, some they have to follow and some they choose to implement, but when they appear in Monday morning briefing, they have the same effect and the origin is often irrelevant.

Beyond that, they share the idea that rigorous experimentation can tell us what must be done in the classroom. That through this ‘relatively new interdisciplinary field’ we are getting closer to a single ‘right way’ of doing things in the classroom. Deviation means poor practice.

There are right and wrong questions. There are good and bad methods. There are relevant and irrelevant outcomes. These can be communicated clearly and simply as a holistic and coherent model of ‘good teaching’.

They also imply that we can make predictions about the exact effects of particular instructional methods.

This is the science of learning. And note that it isn’t really about learning at all; it’s about teaching.

Why we need evidence

The move to make teaching accountable to some standards beyond simply a feeling of what works is a good thing. If we claim we want to make teaching more effective, we need evidence to support this. What about learning styles? Brain gym? The thing the deputy head read about at the weekend and now expects us all to implement in our classrooms? We should be subjecting methods and claims to critical, rigorous scrutiny. The key question, though, is what do our conclusions warrant? Do we have a science of learning? Or do we have rules of thumb about what works in the classroom?

A few weeks ago, something strange happened. With very little fanfare, an academic paper was published declaring ‘education researchers have tried but largely failed to conduct experiments capable of directly informing decisions’.1 The authors were not disgruntled Bourdieu scholars but included academics who had worked on EEF-funded projects.

Randomised controlled trials are not, on their own, the solution, these academics pointed out. Effect sizes of interventions are small and often shrink when interventions are scaled up and variation is sizeable – what increases scores for one student at one school might have the opposite effect for a student at another school (and we don’t really know why this is). For many interventions, the confidence intervals are so wide the true effect could be positive. Or negative. Or zero.

Their diagnosis is right. The ‘what works’ project has largely failed to show ‘what works’. The average EEF trial, they point out, costs around half a million pounds.2 As they point out in the pre-publication version of the paper, this has contributed significantly to the fact that ‘Rich nations have spent billions of pounds on experimental education research in the last twenty years, however much of it has proven uninformative’.

But what does it mean for the findings to be informative or uninformative? We could take the results to be a very expensive but meaningful lesson – education isn’t a science. The results of an RCT alone – without considering what we want from our schools – cannot decide policy.

Their solution is a different set of methods and a different role for experiments, but I’m not convinced that substituting a set of untested assumptions in place of randomisation is the answer. Still, they do suggest a more modest goal for the evidence generated – to inform teachers’ mental models rather than telling anyone what to do.

But what about the ‘Learning Sciences’?

The evidence for Cognitive Load Theory was initially developed through small-N lab experiments, not large-scale in-school trials. Do the criticisms still apply here?

I think that those things lumped under the umbrella term ‘Science of Learning’ can be useful, but I do not think that gives us a single ‘correct way’ to teach. As the EEF concluded after conducting a systematic review, ‘The evidence for the application of cognitive science principles in everyday classroom conditions (applied cognitive science) is limited, with uncertainties and gaps about the applicability of specific principles across subjects and age ranges’. They also argue, ‘Theories from basic cognitive science imply principles for effective teaching and learning.’ The key term, here, being principles rather than laws.

Dylan Kane embeds this discussion in his own practice and experience, pointing out that while we have some useful techniques, the classroom is a complex place and we’re still some distance from a comprehensive set of procedures that work all the time for all students.

Cognitive Load Theory in particular deserves extensive consideration, not offhand dismissal, so I’ll return to it in detail in a later post. Three things are relevant here, though. First, CLT is a theory of how information is processed during learning, not a holistic explanation of learning or teaching. It can guide a teacher on how to explain a complex idea while avoiding overloading working memory, but teaching involves considerations beyond task design, prior knowledge and working memory. Children’s learning is also affected by interest in a subject, what they had for breakfast, whether they’re suffering from hay fever, whether they think you ‘looked at them funny’ on the way in, and whether they’re wearing their new prescription glasses today or have ‘left them at home’ and so cannot see the board from where they’re sitting. These are not incidental questions; they can derail a whole lesson for a student. The only theory or algorithm that can weigh them all up simultaneously lives in the teacher’s brain.

Second, in practice, cognitive load theory is a rule of thumb, not a scientific law. An informed classroom observer judges qualitatively how well a teacher is adhering to it. We cannot make quantitative predictions about the exact cognitive load a student will experience under certain conditions.

Third, before we tackle the question ‘how best can we do it?’ we need to address another: ‘what’s it for?’ Is school as an institution a form of incarceration or a vital tool in addressing societal inequality? The debate goes deep, and CLT appears along one branch at a late stage. Before that come other questions – What counts as ‘society’? Who defines inequality? What’s a desirable outcome? What should we measure and how can we measure it, if we can even measure anything at all?

The problem with turning teaching into a science is that you get seemingly objective standards that you can hold a teacher to. You can always refer to CLT as if it tells you exactly what to do under any situation. You can tell all teachers to do the same thing as if it absolves you of the burden of explaining why. Is it because you want to control teachers and hold them to account? Because it makes inspections easier? Or do you think it leads to better outcomes for students? If you think it’ll lead to the best outcomes for students, you’ll need to do better than ‘Doug Lemov told me to do it.’

Should we all be Teaching Like a Champion?

Many of the things Doug Lemov highlights in Teach Like a Champion are effective – and the evidence points that way. But there is no evidence to say that requiring every teacher in a school to follow its prescriptions to the letter leads to better outcomes. As discussed, which outcomes are we talking about? Who for? Even if we fall back narrowly on test outcomes for a large number of students and average them for whole groups, we still can’t claim it works all the time for every student. There is no such evidence for Teach Like a Champion as a package.

Lemov derived these ideas from watching effective teachers, but not every teacher used every technique. Imagine they had. Imagine Lemov had approached each teacher and told them to use every technique with perfect fidelity – would that have improved what they were doing?

Mandating certain things in the classroom is necessary. Some things – don’t assault your students – are so obvious we don’t need to spell them out. My concern, as I’ve pointed out before, is around mandating whole packages as if they’re evidence based when the evidence is for individual elements.

We still judge effective teaching by walking into a classroom and seeing if we like what’s going on. Teachers work from rules of thumb, not algorithms. ‘Good teaching’ is based on professional judgement, how political trends affect what happens in schools, noisy exam grades and student feedback. All of these things are evidence, and we can debate the relative importance of each form, but it doesn’t make education into a science. Insisting it is one narrows the debate around effective teaching rather than opening it up.

Use this secret trick to turn your five-year-olds into thirteen-year-olds

Visit the EEF’s Teaching and Learning Toolkit and you get the impression a school can simply choose an intervention and their students’ results will shoot up by X many months. Now, the EEF do not advocate working through the strands like an eager toddler at the pick ‘n’ mix counter, adding up interventions as you go. But nor do they say you can’t. In fact, the way they’re presented invites it – so let’s do it.

If every school in England chose just one of these interventions, which have an average impact of around 3.7 months,3 we would expect students in a secondary school to make about 18 months’ extra progress over the five years of secondary school. But there is no indication from large-scale tests like PISA or TIMSS that English students are now a year and a half ahead of their international peers.

As Terry Wrigley – then visiting professor at Northumbria – pointed out back in 2017, add the strands together and your students make more than eight years of extra progress. ‘It would seem quite a ludicrous claim that, if you were to follow all these things, suddenly you would have five-year-olds behaving like 13-year-olds.’ In the same article, philosopher of education Gert Biesta was equally critical of the idea that education could work as an input-output system that can be measured through test scores: ‘This may be a logic that works with pig farming, but not with the complex endeavour of education’.

These conclusions are clearly ridiculous. The toolkit is otherwise helpful.4 The summaries are clear and the most serious methodological weaknesses are highlighted. The EEF have done, at scale, what has only been done piecemeal elsewhere.

But the ‘intervention as months of progress’ model implies that if we take two schools and do the intervention in one then we can predict with some exactitude the outcomes for their students. We cannot.5 Pretending we can undermines the nuance and reality of the application of evidence within schools – and applying evidence in the classroom is central to the EEF’s mission. Yes, teachers are busy. But giving them a headline figure rather than the messy reality is not the right solution.

Disagreement is the point

Treating teaching like there’s one form of science backing it up means there’s a danger of disappearing down a rabbit hole searching for a Grand Unified Theory of Teaching. Find it and we’re justified in insisting everyone must follow it. Science, we imagine, converges on the truth. Aristotle’s ideas were superseded by Newton’s which gave way to Einstein’s. Education, learning, teaching – these things are different. There is no single truth they can converge on which tells us how to teach; homogenising them as if we have agreement glosses over this complexity.

Instead, I think there are many useful theories, of which Cognitive Load Theory is one. The disagreement is not a sign that something’s broken but, like politics, it’s the point – it’s what strengthens the field. It points to more educational research, different kinds of educational research, better teacher understanding of the nuance, broader and more sustained teacher education and supported autonomy in the classroom.

Rather than unification, what we need is more curiosity. Why do we have disagreement? Why does someone follow this theory of learning when I subscribe to another? How can one informed commentator claim a classroom is a form of incarceration when another claims it leads to freedom?

This curiosity is what we expect of our students. These are the standards we should hold ourselves to. How else will we improve things for our students?

Here’s what he concluded:

yes, it is worth using strategies that improve motivation in the classroom, but it is more important to use strategies that enhance attainment, learning and competence in students.

I read it with interest. As a teacher, some days it feels like my whole job consists of finding ways to motivate students. I was intrigued, therefore, to find out that motivation might not be as central to education as I thought.

I’ve also spent a fair bit of time reading about and researching motivation. My PhD is on the role of curiosity in science education. Neuroscientists define curiosity as the intrinsic motivation to learn, so motivation is central to my thesis. Muijs is right that the research is complicated and there’s a lot of it. But I think there’s a story he didn’t quite get across.

Motivation comes from the Latin motivus: movement towards something. The key question for me as a teacher was always whether students were moving themselves through their learning – intrinsic motivation – or whether it was down to me to move them – extrinsic motivation – with carrots like praise and house points, or sticks like detentions and phone calls home.

I’m a realist. I still use the carrots and sticks. But I always thought that if I could tip the balance towards students wanting to learn the material for its own sake, the whole business might become more enjoyable for everyone. Which is why I was wary of Muijs’s conclusion. I worry some readers will come away with the impression that motivation is a minor character in the story of learning. A nice-to-have. An optional extra, once the serious business of attainment is sorted.

Nothing could be further from the truth.

Where he’s right

The literature on motivation is messy. Muijs lists some of the ways in which ‘motivation’ has been interpreted: inner drive, self-efficacy, mastery orientation, interest, mindset beliefs. This does make motivation’s effects hard to aggregate. And he’s right to be wary of anyone selling motivation as the key to learning. The ‘90% of learning’ claim he opens with appears to come from Alpha School, the Austin ‘2-Hour Learning’ AI school, whose blog announces that ‘ed-tech is only 10 percent of the solution; motivated students are the other 90 percent’. I’m also sceptical of these claims – it has the ring of a headline in search of a citation.

But having (rightly) pointed out that motivation isn’t one thing, Muijs’s post goes on to draw a single conclusion about it – ‘the effect is modest’. That’s where things start to go wrong.

Problem one: ‘focus on competence instead’ is a motivational strategy

Here’s the strange thing about telling teachers to prioritise ‘attainment, learning and competence’ over motivation. The most influential contemporary theory of intrinsic motivation is self-determination theory, and it’s right there in the studies Muijs draws on. It holds that intrinsic motivation grows out of three basic psychological needs: autonomy, relatedness and competence.

Autonomy is often taken to mean the amount of choice a student has, but it’s subtler than that – it’s about perceived volition and ownership. A student given three tasks to choose from, with no idea what’s going on and no interest in the larger objective, won’t feel autonomous. A student with only one option, who endorses the goal and understands what they’re doing, will.

Relatedness is a feeling of connection: being respected and valued by teachers and peers in the learning environment. But if SDT is a three-legged stool, England’s is wobbling. For far too many students, the relatedness leg is shorter than the others. The proportion of students feeling supported by classmates and teachers has plummeted over the past decade, especially among girls, as I showed recently using the WHO’s HBSC figures.

Competence, finally, is not a test score. A student with near-full marks who obsesses over the ones they missed can feel less competent than a lower-scoring student who senses they’re mastering the basics. Competence is related to attainment, but it’s a subjective sense of getting somewhere – not the same thing.

So competence isn’t an alternative to motivation. In the dominant theory of the field, it’s one of the psychological needs from which intrinsic motivation grows. When a student experiences themselves genuinely improving – well-judged challenge, success that feels earned, feedback that shows the gap closing – their motivation rises. And note that the principal evidence Muijs cites, the Howard et al. (2021) meta-analysis, is itself a meta-analysis of self-determination theory. His own evidence base is organised around a framework in which competence feeds motivation.

The advice ‘build competence rather than motivation’, then, isn’t ranking two alternatives. It’s setting motivation against one of its own ingredients. A teacher who engineers early success for a struggling reader is doing attainment-building and motivation-building in the same act. The dichotomy the conclusion rests on doesn’t exist.

Problem two: ‘modest’ doesn’t survive disaggregation

Here’s what’s odd about the ‘modest’ summary: the stronger numbers are in Muijs’s own post.

He reports, from the Toste et al. meta-analysis, that intrinsic motivation correlates with reading achievement at .32 and competency beliefs at .28 – and he correctly labels these ‘moderate’. It’s the loosely defined constructs – attitudes, generic ‘interest’ – that drag the pooled average down.

And is .32 modest anyway? By Cohen’s old rules of thumb it’s moderate; by the benchmarks now used for educational field research – where an intervention shifting outcomes by a couple of months’ progress is a success worth funding – it’s a substantial relationship.1

Two reasons the true relationship is probably understated even by these figures.

First, meta-analyses of this literature inherit its definitional chaos: pool forty studies measuring forty subtly different things and you attenuate everything. (Muijs flags this limitation himself, to be fair – though he then leans exclusively on meta-analyses anyway.) Look at studies that isolate a well-defined construct and the pattern becomes more coherent. Taylor and colleagues (2014), across multiple school and college samples, found intrinsic motivation was the only motivation type to consistently predict achievement over time, controlling for prior achievement – exactly the construct that tops the meta-analytic tables, behaving exactly as SDT predicts.

Second, we’re only ever studying students who already have some motivation. Everyone in these datasets turned up, opened the booklet and answered the questions. The genuinely unmotivated student is the empty chair – and many of the students for whom motivation is most consequential are partly missing from the data altogether: persistent absentees, school refusers, early leavers. When you compute a correlation on a variable with its bottom end sliced off, the number you get is smaller than the true relationship. We are, in effect, measuring how much extra motivation matters among the already-somewhat-motivated. And still finding .3.

Problem three: a loop is not a league table

The most interesting evidence in Muijs’s post is the longitudinal work – Vu et al.’s finding that attainment influences motivation roughly twice as strongly as the reverse, with both directions significant. A reciprocal relationship: success breeds motivation breeds success. But it doesn’t necessarily follow that because the average arrow is fatter in one direction, teachers should prioritise that direction.

For one thing, look inside Vu’s results: self-belief – the construct closest to SDT’s competence – was the one form of motivation that significantly drove later achievement. The loop’s motivational arrow runs through exactly the territory SDT predicts.

For another, a feedback loop doesn’t come with a ranking of its own components. The practical question for a teacher is never ‘which arrow is fatter on average across 60,000 students?’ It’s ‘what’s the best way to help the student in front of me learn?’ And for the disengaged student – the one this debate is really about – ‘raise their attainment first’ is circular advice. They won’t attain, because they won’t engage. The accessible entry point into the loop is motivational: make the next success reachable, make it feel earned, make the subject worth a second look. Or check the other legs of the stool – is it ownership of the learning goals that’s missing? A lack of connection to classmates or teacher?

That this route then works through attainment isn’t an argument against it. That’s just what a virtuous cycle is.

Where I’m more sceptical than Muijs

There is one point on which I’m more pessimistic than he is: what we can do for students who lack intrinsic motivation.

Showing that motivation matters is not the same as showing we can reliably boost it. IQ is among the strongest correlates of test success ever measured, but nobody runs IQ-boosting assemblies. The last two decades of motivation-adjacent interventions should make us humble: growth mindset went from ministerial speeches to a 100-school EEF trial that found no average extra progress, and grit is currently making the same journey through the Department for Education’s press office. In my own rough analysis of PISA 2022, curiosity had a stronger relationship with attainment than either.2

In my view, we have better evidence that motivation matters than that we can manufacture it on demand. Anyone reading this post as a licence for motivational interventions has read it backwards.

What to do when motivation is everywhere

Muijs and I agree that motivation predicts attainment and vice versa, and even agree on the numbers. Where we differ is over motivation’s role in learning. He treats it as one variable in the model, to be weighed against the others and ranked. I think it’s closer to the lens through which almost everything teachers do takes effect. Behaviour policy affects motivation. Curriculum affects motivation. Assessment, feedback, explanations, relationships, the level of challenge – all of it shapes whether students keep moving through their learning, whether we track it or not. And motivation, in turn, shapes their learning: it’s at work wherever effortful learning – that is, the kind schools deal in – happens. It’s there every lesson, every day.

And rather than concluding that a messy field yields a ‘modest’ effect, I’d say the messiness is the finding. Tease the constructs apart and the well-defined ones – intrinsic motivation, beliefs about one’s own competence – turn out to be among the more substantial relationships in education research, sitting in Muijs’s own sources, correctly labelled, waiting to be taken seriously.

So rather than telling teachers that attainment matters more than motivation, I’d stop treating them as rivals. The teacher engineering early success, calibrating challenge and showing students their own progress isn’t choosing attainment over motivation – they’re doing both at once, because in a loop there is no ‘instead’.

That excitement continues through their first weeks in the lab, especially as we let them loose with the Bunsen burners.1 But they soon divide into two groups. There are those few who, like my top set Year 9s, will do an electricity practical on series and parallel circuits. I’ll show them how to set it up and they’ll race through it. They might not derive all the laws themselves, but they understand the trends when we discuss them.

Those with lower prior knowledge, or who pick up new ideas more slowly, too often remain eternal novices. In that same lesson they may only learn how to set up a circuit. Yet they’re still expected to collect the readings, understand the rules for series and parallel circuits and remember them for future lessons.

The curriculum won’t wait. We talk about equity, but I’m constantly denying knowledge to my bottom set class. Setting up the circuit is a genuine achievement, but because the lesson demands so much at once, too many spend the practical copying the results over their partner’s shoulder – no chance of understanding what they’re actually doing, their working memory always in danger of being overwhelmed by the demands the lesson places on them.

Practical has always been an awkward aspect of science education. It’s difficult to manage well, to assess and we’ve never agreed on what the point of it is. But science itself without practical work is unthinkable. Theory alone would be meaningless. Hypotheses are contested in the real world, with real equipment. Without falsifiable theories – and the experiments to test them – science collapses into story.

Little scientists and big scientists

Children cannot do the same things as real scientists, we are told, because they are not experts. They must have the fundamentals in place to do real science. But what do real scientists do?

Ideally, real scientists fully understand the experiments they are working on – the set-up, the assumptions, what each reading means, when a result is too good to be true. And they are driven by a curiosity that carries on burning: even after a long day in the lab, the next reading might be the one that confirms or refutes their hypothesis. That curiosity is never truly satisfied, because greater understanding leads to further questions.

This is where the novice-expert divide comes in. If you use practical work in school with students who don’t have the requisite background level of knowledge, you end up with one of two things: science as magic tricks or science as box ticking.

Magic tricks

Science as magic tricks – that’s the promise of all the open day exhibits. Science will be fun, because you get to do these cool new things, schools are telling those over-excited Year 6s. And it does generate curiosity – you can see it in their faces, hear it in their questions. But it’s a shallow kind of curiosity. While the professional scientist might be spurred on after a discovery, once that ‘ooh-ah’ school demo is over, you’re left with the same disappointment you get after finding out how a card trick’s been done.

Take the elephant’s toothpaste demo in chemistry:

I take the kids outside to watch me and there’s one question I get again and again – ‘can I film it?’ Occasionally someone might ask what just happened but most simply remember a vaguely chaotic demo.

But maybe these experiments help to promote a long-term love of the subject? The evidence says otherwise.

Less boring than writing

Ian Abrahams, of the University of Roehampton, interviewed 96 students about their views on practical work. Most liked it. By far the most common reason? ‘Because it is less boring than writing’.

Here’s an example. Abrahams asked one student if they were going to continue with science after it became optional:

Student: No not really I’m not really into it all.

Researcher: But you did say you liked practical.

Student: Yeah but, ’cause sometimes it’s fun, and practical’s easier than, well, writing.

Abrahams split students’ responses into two categories: ‘absolute’ and ‘relative’ liking of practical work. Those who liked practical work absolutely said things like ‘it’s fun’. Those who liked it relatively only preferred it to something, e.g. ‘because it is better than reading the textbook.’

Here’s how those students’ preferences changed between Year 7 (age 11-12) and Year 11 (age 15-16):

Obviously this is based on a relatively small sample, but there’s still a clear trend – liking practical for its own sake declines steeply after Year 7.

Abrahams concluded that students’ interest was ‘situational’: ‘unlikely to endure beyond the end of a particular lesson’. The result was that ‘pupils will perceive science as boring despite their having used practical work on numerous previous occasions.’

Teachers, Abrahams found, used practical work primarily for two reasons:

1. To help in the behavioural management of the class—particularly with low to low/middle academic ability pupils.

2. To help off-set the image of science as difficult, dull, and boring by presenting an alternative, arguably misleading, image of science in which the emphasis is primarily on ‘doing’ fun and enjoyable ‘hands-on’ work rather than on learning about ideas.

One head of department put students’ enjoyment of practicals down to the fact they didn’t have to think hard during them.

Practical doesn’t have to be this way, of course. Drawing on the work of scholars like Abrahams, England’s Education Endowment Foundation’s review of best practice is an excellent starting point for making practical work meaningful. But the EEF note:

Pupils should know why they are doing an experiment, but young people often report ‘just following the instructions’ without understanding the purpose of practical work.

This ‘just following the instructions’ is the second category that practical work falls into all too often.

Box ticking

Box ticking happens when we make students do practicals they don’t really get or care about. It’s like dragging your kids around the art galleries and big houses of Rome, pointing – That’s a Titian! That’s a Caravaggio! – and they yawn and ask when you can get a pizza.

Newton revolutionised our understanding of motion by overturning centuries-old ideas, like those of Aristotle. You can demonstrate his laws of motion quite easily in a school science lab. Take a little cart, tie some masses onto some string and let them fall off a bench to accelerate the cart. Measure the acceleration and plot it against the size of the force created by the falling mass and you’ve demonstrated Newton’s Second Law. What better gift to students than allowing them to see this?

We do, of course. In fact, at GCSE, it’s what’s called a required practical. The issue is that very few students appreciate there’s any magic here, because it would take weeks rather than a lesson or two for most to understand the significance of that one experiment. We rush through it and they end up with only the faintest idea about why they’re hanging masses off a cart.

We outsource the importance from the material itself to the qualification that enables students to progress in life. We make students care by putting the practicals on the theory exam. They must memorise a recipe to get top marks, according to the exam board:

The method would lead to the production of a valid outcome. The key steps are identified and logically sequenced.

This can be done with absolutely no knowledge of what the practical itself is for. I spend one lesson doing the accelerating cart practical, then revise it again and again – always in theory. The system doesn’t prize understanding why that simple demonstration of Newton’s second law is revolutionary. It doesn’t know how to. You can recall a method? You can plot points? You know the name for the shape of the graph? That’s what success in school science looks like to students.

Impotent Knowledge

Let’s imagine, for a moment, our curriculum was a little different. Instead of the sciences, we had subjects like astrology and alchemy. These subjects were detailed and investigated rigorously.2 Let’s imagine we codified the knowledge and put it into textbooks and examined students on star charts and the best recipe to turn mercury into gold. Sure, the practicals didn’t always work out, but that’s just what happens in school experiments, isn’t it?

The difference, of course, as Michael F.D. Young explained, is that powerful knowledge allows us to make predictions. It tells us something about reality. Alchemy cannot turn base metals into precious ones, but scientists really can transmute elements – we have created atoms in the laboratory that, as far as we know, have never been seen in nature. Astrology cannot foretell when the end may come for humankind, but physics allows us to track that asteroid which might be heading towards earth.

Except that students can’t do these things. What most actually take away from school science is either banal, like the fire triangle, which they could probably put into action if the bin went on fire in front of them without having been taught it. Or something they couldn’t apply in the real world. Newton’s laws are great, but there’s very little students could predict with them in practice. Every question that tells them to ignore friction and air resistance is another step away from reality.

Our curriculum, for many students, is like the struggling kid at sports day, tripping over their own heels in a doomed effort to catch the pack running ahead of them. The knowledge the curriculum contains, so powerful in expert hands, is impotent for those students – a star chart with better PR.

Do we need to lower our standards to raise them?

Thinkers like Michael Young and E.D. Hirsch – whose ideas underpin our curriculum – are right: our students need to know science to participate as educated members of society. But, at the moment, most don’t. One of my Year 11s told me just before they went on exam leave that they wished Newton had never existed – that their lives would have been much easier without having to memorise his stupid laws. This doesn’t smack of being empowered by knowledge.

I’ve discussed before how little students need to know to pass their science exams in England. We claim we have high standards because of our packed, uniform, knowledge-rich curriculum; but what if we started calling out the contradictions that we saw? Success isn’t making every student work through every lesson regardless of whether they get it or not. Knowing 25% of a huge amount of stuff isn’t necessarily superior to knowing 80% or 90% of something smaller but more coherent and meaningful. Doing practicals as magic tricks or tick box exercises belittles science itself.

If pupils never move past novice stage, they lack the apparatus to judge the power of the ideas they’re learning. What if, instead, we slowed down, where we needed to? We could let that Year 9 class, who have just mastered the art of setting up a circuit, pick up where they’d left off and start taking readings the next lesson, slowly building up an understanding of what they’re doing and why they’re doing it. Without that understanding, science is no more meaningful to them than alchemy or astrology, except that they have some vague sense it’s valued by society whereas alchemy and astrology are quackery.

Nobody has asked me to write a scheme of work or a national curriculum, but if they did, I’d reverse it. I wouldn’t have a set of facts students needed to have in their head by the end – I’d start with the most foundational ideas and take it from there. Students would become expert, theory and practical reinforcing one another, so they truly appreciated the significance of each idea. ‘Pure’ discovery learning certainly isn’t the answer here; instead, we need to carefully tailor teaching to the students in front of us.

There is a strong argument against this: that slowing down for the students who struggle most is just low expectations dressed up as compassion – the very thing the knowledge-rich movement set out to fight. But we aren’t comparing a little knowledge to a lot; we’re comparing nominal, superficial coverage to genuine acquisition. Students are entitled to a curriculum they understand, not a speed read of a book in a language they only half know. And yes, it would pose problems for exams if students didn’t all cover everything at the same pace, but do we really want that one consideration to be the beating heart of our education system?

This resembles mastery learning – refusing to move on until students understand the key ideas. The EEF Toolkit rates mastery learning as having a moderate impact (+5 months), but English trials found much smaller effects. That’s hardly surprising if schools remain bound by a system in which everyone is ultimately expected to move through the curriculum together. Mastery without genuine flexibility over pace is mastery in name only.

Rather than an entitlement, the curriculum too often serves to deny powerful knowledge to those who need it most. Our teachers are waiting, ready to help students understand the real magic behind each practical. But too often they have to push on, well aware that most of the class just don’t get it, but what choice do they have? There’s a rota, a scheme of work, today’s powerpoint to get through. That’s when practical work becomes the carrot to get them through another lesson, as one of the teachers from Abrahams’ study put it.

Where facts stop and understanding begins

Equity is the right goal, but we’re going about it the wrong way. Power lies not in a series of half-remembered facts, but in understanding how science works, what it’s for and where its limits lie. To understand how science works is to know the tricks that climate change deniers pull and why those who promulgate ‘race science’ are deluded. It’s easy to produce an apparent counter-fact and claim that nothing is settled, which means that all views are equally valid. But, as Hannah Arendt argued, if we want the next generation to improve the world we’re leaving for them, we need to give them the tools to genuinely understand it first.

Instead, we teach that science is simply an accumulation of facts. When you know the facts, you know science. That route might end in you acing your exams, but it might also result in thinking that absorbing all the facts about how America staged the moon landing is a valid scientific approach.

I’m going to change how I teach science. Before each experiment, I’m going to ensure that students understand why we’re doing it. If they don’t, we’ll go back over the rationale and, if necessary, cut the practical. Why waste their time with magic tricks and box ticking?

But something else needs to change. We need to give the big ideas space to breathe. Science is the most powerful tool we’ve ever discovered for understanding reality. If children could leave school really, truly understanding that idea – that, to me, would be genuinely magical.

England has many attainment gaps. One has persisted since the 2017 GCSE reform, yet rarely attracts calls for action. Every year, girls have been considerably ahead of boys in the number of pass grades they achieve. It was surprising, then, on GCSE results day last year to see a BBC article asking: ‘Are girls being left behind in school?’

That year, girls were over six percentage points ahead of boys. Across all exam papers girls sat, this meant they passed around 170,000 more of them than they would have if they’d had the same pass rate as boys. Nevertheless, the chief executive of the Education Policy Institute was quoted arguing for an urgent need to look at what’s happening to girls’ results before it’s too late.

Subscribe now

The gap is closing – but this might not be good news

To understand the concern, we need to look at the trend in pass rates over time – something complicated by a pandemic, teacher assessed grades and then changes to pass marks.1 So, let’s skip the years in grey and concentrate on those either side.

Back when students sat the first reformed GCSEs in 2017, girls were almost 11 percentage points more likely to pass any given GCSE. The gap has shrunk consistently – leaving aside those grey years – down to 6.1 percentage points.

This sounds like a success story. Indeed, in 2025, the Daily Express announced ‘Boys close GCSE gender gap with bumper grades’ and the Guardian reported, ‘Gender gap smallest since 2016… Boys show improved results in major subjects’, although they do note the other side of the closure, too. Because two things are happening, and one is considerably more significant than the other.

First, boys’ scores are improving, but not by much (although small differences have a big impact at this scale): 63.0% in 2017 to 64.1% in 2025. At the same time, pass rates for girls have plummeted, from 73.9% in 2017 to 70.2% in 2025. Across the almost three million exam papers girls took in 2025, they now fail over a hundred thousand more than they would have at the 2017 pass rate.

One policy change accounts for some of this variation: if you fail maths or English, you now need to carry on studying it until you pass, if you are in funded education. This reform has not been universally welcomed.

But to really understand what’s happened, we need to look at individual subjects. We’ll focus on two core areas, the ones so important students need to keep retaking if they fail – English (split into language and literature) and maths, beginning with the latter:

Clearly, something wasn’t quite right with Ofqual’s grade boundary calculations when the new exams were introduced in 2017, and it wouldn’t be fair to use those initial grade boundaries, so we’ll take 2018’s figures as our benchmark.

Since then, the number of boys passing GCSE maths has barely changed (a fall of 0.3 percentage points). It’s a different story for girls, with their pass rates plunging from 59.5% in 2018 to 56.7% in 2025. Across the more than 400,000 papers girls sat in 2025, it meant more than 10,000 more failed than would have done under the 2018 pass rate.

In English, girls have long outpaced boys, but that’s been changing. And, because of some innovative research, we can see that failing to pass English can also be costly – and not simply because you lack the basic skills but because you lack the certificate.

GCSE marking has some inherent noise near grade boundaries, meaning researchers can compare students who end up just above or just below the pass mark in English. These students had near-identical exam performance – any small marking fluctuation could have tipped them across the boundary. Machin, McNally and Ruiz-Valenzuela found that those who landed just below the line were significantly less likely to progress to a good sixth-form course or start university, and more likely to be NEET at 18 – all stemming from tiny statistical differences at 16.

For English literature, the picture is a more positive one, with girls’ scores largely flat but boys’ scores consistently improving:

English language is a different story, though. While boys’ pass rates have fallen between 2017 and 2025, girls’ pass rates have dropped off a cliff:

What sound like tiny differences can bring huge costs – and we can do some back-of-the-envelope calculations to quantify them.

Let’s look at 16-year-olds only, to exclude retakers, whose data isn’t captured as well by the longer-term earnings calculations. In 2017, 78.2% of 16-year-old girls achieved at least a grade 4 in English language; by 2025 that had fallen to 75.9%, a drop of 2.3 percentage points. Applied to that year’s 307,270 papers, that’s roughly seven thousand more girls missing a pass than would have under 2017’s rate.

According to government figures, the boost in earnings for those who scraped a pass rather than failing their English language GCSE was around £18k in 2019 money.2 Multiplying the extra fails by the loss of lifetime earnings gives you an estimated £127 million in lost future earnings. So, that’s one subject – and we haven’t even added on seven years’ worth of interest.

Which gap are we closing?

Something has happened in England. We’ve been focusing on closing one gap – the socioeconomic disadvantage gap – at the expense of all others. Still, the gender gap has been closing, which is no bad thing in itself. Why say the gender gap is the ‘natural’ consequence of girls’ or boys’ ability or learning disposition if we wouldn’t say the same for disadvantage – or any other important outcome disparity?

But closing the gender attainment gap hasn’t been cost-free. It’s come at the expense of girls’ attainment, costing them and the economy many millions, by any estimate. It’s like dragging down the GCSE results of wealthier students and claiming we’d narrowed the gap between them and their disadvantaged peers. Few people, surely, would count this as a victory.

It’s likely there are many factors at play. But this doesn’t mean a shrug of the shoulders and back to business as usual. Before we continue, though, a note on gender: The data sources cited in this post treat gender as a binary. We lack robust, large-scale data for outcomes of students who identify, for example, as non-binary, and, while this gap should be addressed, it’s not something we can examine here due to these limitations.

What happens if you isolate and educate by gender?

To tell whether school environment makes a difference to boys’ or girls’ exam outcomes, we can perform a natural experiment, comparing students in mixed environments to the 10% in single-sex schools. And there’s a clear winner – students in single-sex schools do a lot better, getting almost grade 6 on average across their eight most important GCSE subjects. In mixed environments, the average is 4.6.

But there’s an issue. Single-sex schools are more likely to be selective or grammar schools, meaning students have higher prior attainment. They’re more often located in wealthier parts of the country, especially London, and have fewer disadvantaged students. Take all the measurable differences into account, as statistician Natasha Plaister did for her FFT Data Lab post, and the difference shrinks to almost nothing.

Not quite nothing, though, and the difference is important. After controlling for all the differences she could, Plaister found that boys make, on average, no extra progress in single-sex schools, but girls make one month’s extra progress – a small but meaningful difference. Girls’ scores may, therefore, be more sensitive to school environment than boys’, as evidenced by their larger pass grade fluctuations.

What’s changed in England’s schools?

If we’re interested in how England’s schools have changed in the last decade and a half, we need look no further than the educational policy changes brought about by the Conservative government in the early 2010s – not least the new exams discussed above. These changes have survived several changes in administration, including a Labour government, mostly intact. I’ve discussed a few of them in recent posts:

• Moves towards mandated teaching techniques, as if they are fully evidence-based – a practice I’ve called ‘total instruction’.

• Decreases in teacher autonomy and wellbeing.

• A curriculum which proclaims high standards whilst allowing low pass marks.

We cannot pin any change, causally, to the fluctuations in girls’ pass rates, but it is worth considering whether there might be some underlying trend affecting girls. Perhaps girls’ wider school experiences are affecting their attainment.

The World Health Organisation conduct an international survey of adolescent health and wellbeing every 4 years. They ask thousands of 11-, 13- and 15-year-olds from around the globe about their health and wellbeing, plus school-related questions regarding relationships with teachers, classmates, and general experience of school. If we compare the most recent results, from 2022, to those of 2014, we can see how school-related wellbeing has changed since these Conservative-led changes began.

Daisy Christodoulou used the data to argue that the Conservative-led changes (for which she used the umbrella term ‘knowledge-rich curriculum’) had not harmed children’s mental health. I posted a reply, in which I raised doubts about the conclusion.

Here’s attitude to school, to begin with:

What we’re interested in here is not whether students’ attitude to school has suffered as a whole – although it has clearly crashed since 2014 and was plummeting even before the pandemic. More relevant are the gender disparities. At age 11, for example, girls liked school more – until recently. Similarly, at age 13, girls and boys shared a starting point, but their attitudes have diverged.

If anything, the deviation is even clearer in terms of students’ perceptions of pressure due to schoolwork. Perceived pressure has certainly increased overall. The trends at ages 11 and 13 leave no doubt around the fact that this is a gendered story too:

Finally, perceptions of teacher support have tumbled – and fallen considerably more steeply for girls than boys. As with schoolwork pressure, at age 11, the genders have once more flipped.

Opening the envelope

In some ways, it makes little sense to call the English curriculum knowledge rich. On some papers, a student can get a ‘good pass’ with marks a little over 20%. The disadvantage gap – the one the curriculum was meant to close – remains stubbornly high, and the gender gap has fallen, but largely at girls’ expense.

This could all, perhaps, be defensible if it had resulted in more enjoyment of school or a greater sense of support. Every HBSC indicator of school wellbeing decreased between 2014 and 2022. This is not simply a case of pandemic-related worldwide collapse; the fall was in several cases steepest from 2014 to 2018 and nearly all represent collapses in England’s standing relative to the 40-odd other nations surveyed too.

Most relevant to this post, though, is the divergence between the genders. In almost every case, boys’ and girls’ perceptions of school have diverged since 2014. Something has happened within schools that has affected children’s wellbeing – and it’s affected girls most of all.

It’s hard to pin these trends on a curriculum or changes to classroom expectations. But the patterns across all school wellbeing areas are clear and unambiguous. We’ve been asking one question again and again – ‘how do we close the disadvantage gap?’ Our answer to that needs to improve, but we need better questions, too.

Every one of the statistics we’ve seen – from the tiny changes in exam pass grades to the fall in enjoyment of school or feelings of teacher support – represents a student’s experience. Opening an envelope on results day, heart sinking as another door closes. Arriving in school for a lesson taught by mandated pedagogy rather than a teacher, responsive, understanding, ready to spark curiosity and bring the lesson to life.

Are girls being left behind in school? On the strength of this evidence, it seems so. But we also need more research. We have the evidence about exam grades and from wellbeing surveys but it’s time to widen our sense of what counts as evidence here. We need to understand their experiences, by speaking to them, listening to their answers, then doing something about it.

Last night, Blossom1 (8) chose a song she’d heard that day at school – ‘The Vertebrates Song’, by Hopscotch.

Repeat after me: Mammals, reptiles, birds, amphibians, and fish.

Rosie (4) asked Blossom what amphibians are.

‘Frogs and toads,’ Blossom said. Then turned to me: ‘And newts too?’

I hedged my bets. They live in ponds, and they’re a bit slimy. The sort of thing you really need a science teacher on hand for.

Animal classification is something we teach in school – well, science teachers do. It’s part of the powerful knowledge our curriculum is built on. And according to the orthodoxy that now governs English classrooms, that knowledge belongs in school: carefully sequenced, taught in small chunks to avoid cognitive overload, with pupils securely holding the knowledge before any enquiry begins, moving from novice towards expert.

So what was Rosie doing asking about amphibians when she had only half an idea about animal families? After nursery today she asked why butterflies visit flowers as well as bees. Sometimes she’ll ask question after question, blissfully unaware she’s firmly in the novice camp.

And there’s nothing special about Rosie, beyond her love of clothing emblazoned with unicorns and her vehement dislike of anything containing vitamins or fibre. Questioning is something all kids do. No sooner have they learnt the word ‘why’ than they’ve weaponised it.

When children start school – as everyone from Bourdieu to Hirsch agrees – some arrive with more of this powerful knowledge than others. Despite the best evidence-backed methods we have, our schools struggle to close that gap.

At age 11, less than half of disadvantaged pupils reach the expected standard in reading, writing and maths combined, compared to more than two-thirds of their wealthier peers. Even if two children score the same at 11, by 16 the disadvantaged child is more than half a grade behind, on average, across all their GCSEs. While 44% of the GCSE gap is already present at 11, the remainder opens up during secondary school.

We spend £3 billion a year on Pupil Premium funding to support some 2.3 million disadvantaged pupils, and the Education Endowment Foundation has been given hundreds of millions to get at the causes.

We focus on the outcome of what happens at home, the visible bit: the fact that some kids arrive at school with more powerful knowledge than others. What we should be asking is how they got it in the first place.

Blowing up the concrete

No politicians in recent decades have shaped English education more than Nick Gibb and Michael Gove. Gibb’s holiday reading in 2005, when he was shadow Minister for Schools, was E.D. Hirsch’s The Schools We Need and Why We Don’t Have Them. Hirsch argues that educated citizens must be fluent in shared cultural knowledge; where the home hasn’t supplied it, the school must. Gibb recommended it to Gove, and when the Conservatives took power, England’s knowledge-rich curriculum was born.

Gibb had an enemy in mind. ‘I am not sure until Michael and I came in that people really understood that the problem with poor standards is progressive education ideology, the child-centred learning approach to teaching,’ he later said.2 Drastic action was required: ‘In 2010… we had to blow up the concrete to allow teachers to have these [opposing] views’.

Rightly, Gibb insisted pedagogy should be backed by evidence: ‘Poor teaching methods harm all pupils, but a growing body of research suggests it harms disadvantaged pupils most of all’. But this is not just about two men from one party. The vision survived ten education secretaries and a change of government without Labour laying a finger on it – not least the teacher training framework, which now sets the standard for good teaching in England.

England’s ‘official pedagogy’

In 2019, for the first time, England’s teacher training providers had a curriculum of their own to follow, citing familiar names: Wiliam, Dunlosky, Willingham, Sweller, Kirschner, Hattie, Rosenshine.3 It hasn’t been universally popular, especially among academics, two of whom argued it produces an ‘official pedagogy’ resting on ‘an imaginary notion of instructional practice that does not fully grasp the context of teachers’ work’.

‘Explicitly teaching pupils the knowledge and skills they need to succeed within particular subject areas is beneficial,’ teachers are told, and they should do it by ‘breaking complex material into smaller steps’.

The framework doesn’t represent poor practice. But it does, I’ll argue, represent a partial view of the evidence on learning – one that, followed to the letter day in and day out, can clash with the best evidence we have about using children’s curiosity to drive their learning. Effective teaching can certainly motivate: you just have to watch the realisation cross the face of a struggling student when an idea finally clicks. But it isn’t the only way, as the evidence on out-of-school learning shows.

Young children learning

A few years ago, when Blossom was five or six, her favourite book was one I was using for my PhD – Tizard and Hughes’ Young Children Learning. It’s a 1984 study of 30 four-year-olds, tape recorded at home and at preschool. We’d act out the transcripts; she’d be the daughter, I’d be the mother.

This is what Tizard and Hughes concluded:

Our observations of children at home showed them displaying a range of interests and linguistic skills which enabled them to be powerful learners. Yet observations of the same children at school showed a fundamental lack of awareness by the nursery staff of these skills and interests… in the world of school, the child appears to be a much less active thinker than is the case at home.

It wasn’t meant as a criticism of staff, but a point about the quality of interaction between a child and someone who knows them well, versus someone supervising a dozen others. Later work supports it: Rachel Romeo and colleagues found that – after controlling for IQ, socioeconomic status and sheer volume of speech – it was the number of conversational turns, the back and forth between child and adult, that predicted a child’s verbal skill.

Opponents like Gibb often associate child-centred learning with leaving children to their own devices. These studies show the opposite: it requires a deep understanding of the child, expressed through responsive interaction – a form of scaffolding. The knowledge gap may itself be partly a product of fewer such interactions, on average, in disadvantaged homes.

It’s also unlikely much of that back and forth is the cognitive-load-respecting direct instruction the framework prizes. It’s hard to imagine many dinner table conversations like this:

Right, before pudding, tell me the chemical composition of water, little Bethan. Two elements, yes. No – let’s try another and come back. This is CO₂; the absence of a number after the C indicates one element of… what was the C? Yes, carbon. Now, the O…

There’s also evidence that direct instruction can narrow children’s exploration. Elizabeth Bonawitz and colleagues gave preschoolers a novel multifunction toy; children explored its other features less if an experimenter explained one – an effect that held even for children who’d merely overheard the explanation. Those given no explanation explored more widely.

None of this means direct instruction is ineffective, or that parents never explain things at length (although direct instruction involves more than just this). But it’s a mistake to assume there’s no robust science behind child-centred methods. In fact, we’re just getting started.

A small step for knowledge – or should that be a giant leap?

England’s teacher training framework presses teachers to break things into small steps and avoid overloading working memory. This sounds innocuous, but to the emerging science of curiosity, this advice could pose a threat.

George Loewenstein, in his seminal 1994 paper, The Psychology of Curiosity, claimed that curiosity is triggered by a knowledge gap of just the right size – something that has now been shown across many studies. Curiosity researchers don’t treat ignorance as something to eliminate as fast as possible. They treat a carefully judged gap – uncertainty of the right size – as the engine of learning itself.

The framework encourages teachers to minimise unnecessary uncertainty by ensuring pupils master the knowledge before moving on, to ‘avoid overloading working memory’. But curiosity doesn’t want small steps; it wants a gap tailored to the learner’s confidence, and it doesn’t want that gap closed immediately – it wants you to sit with it. That’s curiosity, after all. The framework, by contrast, tells teachers to ensure pupils ‘master foundational concepts and knowledge before moving on’ and to ‘remove scaffolding only when pupils are achieving a high degree of success’. There’s a central message: knowledge first.

Manu Kapur’s work on productive failure turns that on its head. When learners attempt something and don’t succeed – and are then taught the key points – they often understand the underlying concept better and transfer it more readily. Direct instruction still features; it’s the order and format that change.

We can see productive failure as curiosity engineered into a lesson: the problem-first element forces pupils to draw on prior knowledge, and the gap they hit can activate curiosity – provided it’s well judged and students value the information required to fill the gap. That’s where responsive teaching comes in: scaffolding to size the gap, and care in choosing the problem so the payoff feels worth it.

There are some reasonable criticisms – failure can demotivate, which is precisely why scaffolding matters. Productive failure and cognitive load theory needn’t be at odds, though. With scaffolding to manage load (or distributed across a group of learners), the problem-first phase needn’t overwhelm working memory.

The problem with adaptive teaching

Nick Gibb was opposed to child-centred methods – so how then do you support disadvantaged pupils and those with special educational needs? The framework’s answer is an awkward mix of compulsion and proscription. Teachers must grasp that ‘pupils are likely to learn at different rates and to require different levels and types of support’ – blandly obvious – except that in practice strategies are ruled out: adaptive teaching is ‘less likely to be valuable if it causes the teacher to artificially create distinct tasks for different groups of pupils or to set lower expectations’.

There’s evidence from the EEF – who helped write the framework – that brings this advice into question. Across 198 studies (admittedly of varying quality), their toolkit rates individualised instruction – ‘providing different tasks for each learner and support at the individual level’ – as having ‘moderate impact for very low cost,’ and says, for pupils who have fallen behind, it ‘may allow the teacher to provide activities that are closely matched to a pupil’s attainment’.

There’s little evidence the current methods are working for the learners who most need support. England is facing what parliament calls a ‘SEND crisis’: 1.7 million pupils with SEN, ‘exhausted parents fighting for basic support, teachers stretched beyond capacity’, and a system that the House of Commons committee concluded unambiguously ‘is not working’.

It’s little better for disadvantaged pupils. The EEF was set up to find ‘what works’ to close the gap, especially for those eligible for free school meals. Despite promising findings, we haven’t found it: across the trials they studied, Bilal Ashraf and colleagues found the equivalent of 1 month’s progress for literacy and no effect at all for maths among pupils on free school meals. And even that flatters whole-class approaches.

Across large-scale evaluations, the strongest effects came from targeted one-to-one and small-group literacy interventions. When aggregated, whole-class and whole-school interventions would be classed as having ‘very low or no impact’ for disadvantaged students.

Tailoring is the common thread. A teacher working with one child or a handful of students can gauge the perfect knowledge gap for the learner in front of them – exactly the responsiveness that curiosity research points to, and exactly what whole-class delivery of fixed small steps makes hardest.

Breaking out of the straitjacket

Hirsch, Gibb, Gove and the other advocates of a knowledge-rich curriculum began with an admirable goal: to help the most disadvantaged. They saw a product – the knowledge some children already have when they arrive – and set out to make up the deficit in those who didn’t.

What they underestimated was the mechanism.

Disadvantaged children are as curious as wealthier ones; the difference, as my analysis of the OECD’s PISA data shows, is their love of learning in school. Poorer pupils who keep that love do better, but my analysis shows that England has the widest gap out of 75 education systems between general curiosity and that school-specific love of learning for the wealthier and the disadvantaged.

We’ve built an official pedagogy with a great deal to say about what children should know, and surprisingly little about how to keep them wanting to know it. We can fix that – keeping the most powerful parts of explicit teaching while experimenting with those knowledge gaps until they spark curiosity. It may mean straying beyond the bounds of the knowledge-rich curriculum, but should that really scare us?

The curriculum was always meant as an entitlement.

Why, then, have we let it become a straitjacket?

1. A randomised controlled trial carried out by a large ‘what works’ organisation with 2,000 students across 32 schools trialling a new mindset intervention.

2. Your colleague telling you about a session on ways of using formative assessment they saw at a recent conference given by a well-known education writer.

3. Interviews with 15 teachers carried out by a master’s student in the school they teach at about workload pressures.

4. A neuroscience study conducted on 18 undergrads.

5. A technique I thought of myself and used for the first time before break today.

We’ll come back to this later, but to release a little of the suspense: no, I’m not about to tell you a hunch is worth as much as an RCT.

In some cases, it might be worth more.

If I’m presenting education policy, I don’t want to be driven by my own school experience. If I’m about to teach, researching RCT mindset interventions might not be the best use of my time.

How do we tell one situation from the other? And if we’re listening to someone telling us what to do in the classroom, what standards should we hold them to?

Not all educational evidence is of the same quality and relevance, but if the intended end user is a teacher, they should be in a position to make an informed judgement and weigh up these factors.

The role of an ‘education influencer’

I write on the internet giving suggestions about what teachers might do in their classrooms. So do a lot of other people, and their ideas end up being used by hundreds, sometimes thousands, of teachers. Most of what I use in lessons started life this way, and I’m indebted to the people I borrowed it from.

But there’s a danger, which I discussed in my last couple of posts. An interesting idea, built on a single piece of research, can harden into ‘evidence-based practice’, and then into an essential feature of ‘good teaching’ that has to be visible in every lesson.

Bloggers aren’t the only ones translating research. Schools have teaching-and-learning leads. Academics sometimes write for teachers. Ofsted publishes reviews of what it takes as ‘evidence’. What standards should these ‘education influencers’ be held to? What counts as evidence? And is it ever justified to mandate one of these ideas across a school, a trust, or a country?

If we want an evidence-based profession, these are questions we have to answer.

To give a rough idea of how it might work in practice, I’m going to sell you a technique – and I’ll show my working as I go. Let’s see if you’re interested in buying it by the time I’m done.

The 6Ps

First, begin with the problem, not the technique. Too many ideas are used like the man with a hammer – they’re tried everywhere, regardless of whether they address any real need.

My problem is usually anything I think might kill my students’ curiosity. Top of the ‘wanted’ list are lessons that feel repetitive to students – nothing embodies this more than revision. My doctoral research – and teaching experience – tell me that when they think they’ve seen a topic before, students switch off.

Here’s what I’ve been trying. It’s a way of using recall questions to bookend a revision lesson, and of enabling you and your students to judge their progress. I’m calling it the 6Ps, because I haven’t thought of anything catchier:

This is based on some influential neuroscience: people remember more when they’re curious; they get more curious about an outcome once they’ve predicted it; and the memory boost is larger when the prediction turns out to be wrong.

But every one of those is a small-N fMRI or lab study, most of them on undergraduates answering trivia questions. It’s never faced the ultimate test: 30 teenagers reluctantly revising physics in an English classroom on a sweltering June afternoon.

Putting the 6Ps to the test

My Year 10 have their mock exams next week. Today, we’re revising mains electricity. I come up with a set of 10 questions on the topic. Or, rather, I tell the AI of my choice:

Come up with 10 questions on unit P5 (mains electricity) from AQA Trilogy combined science GCSE physics, so I can paste them straight onto a slide. Put the answers underneath.

I check them, obviously – but I know the spec, I’ve taught it for years, and that takes a minute or two instead of the half hour writing them from scratch would.

Now for the 6Ps:

1. I start the lesson with the 10 recall questions (pose).

2. Students answer them then predict how they’ve got on. Tick, cross or question mark next to each. Tick = 1 mark, cross = 0, ? = ½ mark. They give themselves a score out of ten (predict).

3. I hold back the answers (pause).

4. I do my normal revision lesson – go over key ideas, worksheets, the usual. Some of the answers will come up, but I don’t deliberately hand them over (present).

5. Same questions, same scoring format (pose again).

6. This time, they actually mark them (proof-check).

Then on mini-whiteboards they show me their ‘before’ and ‘after’ scores – both sets of questions, marked against the real answers. But I tell them to check their own scoring to see how accurate they were.

The point of all this is:

a) To show them they don’t know everything before we begin so they see the value of the lesson.

b) To improve how accurately they can assess their own understanding of a topic, and boost the efficiency of their revision.

c) For me to find out how much their knowledge changed from start to end and see how well I’ve pitched the lesson.

And… did it work?

The first two times I tried this, it went well. Students gained two or three marks on average, from a baseline of four or five out of ten. If I stopped there, I’d be writing this as a recommendation.

I tried it a third time and it was messier. The lesson overran, so I had to push the final pose-predict-proof-check into the next lesson; the gain came out at about a mark above where they’d started.

Maybe the effect is short-lived. Maybe those questions were harder. Maybe it was the time of day (or temperature of the room).

I asked the students what they made of the technique, on whiteboards, and the verdict was lukewarm: some wanted longer questions, some wanted more explanation, some couldn’t see how they were meant to answer questions on things they hadn’t revised yet.

This is solid science transformed into a method you could use in the classroom tomorrow, so what’s the problem?

As ever, I blame it on the children.

But at least there’s an upside. While it might have failed in the classroom, I can always try flogging it to an ed-tech company as a fallback.

There’s a bigger issue here, though. This is evidence for me, as the classroom teacher. I know the class and can judge the effect based on my sense of what they already know. But that doesn’t license me to start selling the technique on the basis that ‘it works’.

I had no control group. There’s little evidence that it would translate to another class in the school, let alone another school, subject or age group.

‘Try it,’ I can say.

‘Do it’ – less so.

Three rules for selling something

As the blogger trying to talk you into it, what am I honestly obliged to tell you? And if you’re the teaching-and-learning lead who’s just found it, what should you be asking before you put it in front of fifty colleagues on an INSET morning?

Here are some rules of thumb I think anyone trying to influence educational practice should bear in mind, applied to my own method:

Solid evidential basis. The links between curiosity, prediction and memory are well demonstrated and have been replicated. I’ve read the papers and their limitations; I’m not a neuroscientist, but other neuroscientists have cited and reproduced this work, which is about as much as I can honestly claim. So, I can say the science seems solid – but I can’t say it works in the classroom.

The size of the leap. The bigger the change you’re asking for, the stronger the evidence and the more transparent the claim has to be. If I told you the best way to learn about mains electricity was through pure discovery learning – to hand your class a set of screwdrivers and then let them loose with the sockets – I’d need to be very sure indeed. If I told you that break times should be fully structured – scaffolded use of the climbing frame, teacher-mediated discussion with your mates – same again. The 6Ps asks for almost nothing, so the cost of trying it and finding it wanting is almost nothing. That’s the whole reason I can suggest you try it while admitting I can’t prove it works.

Little danger of harm. Providing you already do some retrieval practice, you can probably judge the outcome of doing this before you start. A small leap means it’s easier to judge potential harm, and here there isn’t much room for anything to go wrong. There is a danger of potentially leaving errors uncorrected until the end of the lesson, but we’re talking about material students have met already, with responsive re-teaching at the heart of the lesson and collective marking at the end.

What counts as evidence in education?

You’ve been patient, so here are the five again:

1. The 2,000-student RCT on a mindset intervention.

2. Your colleague’s conference session on formative assessment.

3. The master’s student’s 15 interviews about workload.

4. The neuroscience study on 18 undergraduates.

5. The thing I tried before break this morning.

They can all count – but they’re not all equal. In fact, if you’re planning your lesson for after lunch, the RCT might be the least useful item on the list. It can tell you whether a mindset programme tends to work on average across 32 schools. It can’t tell you whether this exact programme will land with your Year 11 class in an hour’s time. What worked – or didn’t – last lesson is probably a more valuable guide. It's not about creating an evidence ladder or hierarchy, but fitting the tool to the problem in front of you.

The questions to ask next INSET day

Every teacher has sat through a September training session announcing a new whole-school policy. Plenty of them are improvements. More than a few would survive the obvious questions: Why this, and why now? What’s the evidence? Has anyone trialled it here, in this school – ideally the person standing at the front explaining it – before every teacher in the school was told to do it?

I’m not against change; we need it. I’m against change that arrives with all the transparency of a street-corner magician.

I’ve written before about the dangers of ‘total instruction’ – the slow removal of teacher judgement in the name of consistency – but I still think some practices are rightly mandated. Where a student has a special educational need or a disability, adjustments have to be made on the best available evidence; we owe them no less.

I’m not going to tell you what that evidence is – that’s not my area. What I can tell you about is how it feels to follow procedures I don’t understand. I used to have to photocopy worksheets onto coloured paper – up to four different colours for some classes, which is a particular challenge when you’re colourblind, like I am. Then, one year, I didn’t have to any longer.

Where did the evidence for coloured paper come from? Was it always flawed, or did more robust evidence come along that countered it? Nobody ever put either case to me. One year there, next gone. What impact did this have on those who apparently needed the coloured paper?

If we’re going to have whole-school practices, the evidence for them has to be weighed, made transparent, and shown to the people asked to carry them out. Children need good, human teachers more than ever, and the job is arguably harder than it’s ever been. If there was ever a moment to trust teachers with the evidence and let them make their own informed decisions, it’s now.

So, give the 6Ps a go, if you like the look of it – and report back.

But if your students tell you they want longer-answer questions, or more explanation first, or that they couldn’t know the answers because they hadn’t revised the topic yet, don’t say I didn’t warn you.

You’re constantly being watched and told what to do, and being checked up on. I don’t think there’s any autonomy.

Cáit (from Skerritt, 2019)

An open door

Imagine you start off teaching in a system with relatively low accountability and high autonomy – like Ireland, at least until recently. Then you migrate somewhere with low autonomy and high accountability, like England.

Maybe you feel more supported: you enjoy the extra structure, appreciate the regular feedback and feel motivated by being held accountable.

This is what Dublin City University academic Craig Skerritt wanted to know when he interviewed six Irish teachers who’d moved to the English system.

Here’s Aoibheann:

You are accountable for absolutely everything that goes on, including things you have no control over. From what I understand to be autonomy, you have no autonomy. Everything you do is checked by your head of department, then it is checked by the head of faculty, who is then checked by an assistant head, a deputy head, and it trickles up to the principal. If you put a foot wrong twenty people know about it before you do. I don’t think there’s any autonomy. Everything you do as a teacher you are told to do.

English schools are increasingly mandating a collection of practices as if they’re a coherent whole and holding teachers accountable without trusting them to exercise their judgement – a practice I’m going to call Total Instruction. It happens when schools take a grab-bag of techniques, call them evidence-based, and then remove teachers’ decision-making power.

Large-scale data, as we’ll see, can rarely prove that a lack of autonomy causes job dissatisfaction. But the voices of teachers can give us a few hints.

Skerritt’s teachers experienced regular ‘learning walks’ as managers came into their rooms unannounced. Bronagh said, ‘Oh it makes me sick. It terrifies me knowing that I’m going to get observed because it’s like, you know something bad is going to come from it. You never get an observation where you go “Oh that went great and I’m going to get told I’m fantastic”.’

Breaking down the barriers to a knowledge-rich curriculum

Robert Pondiscio wrote last week about the obstacles to a knowledge-rich curriculum. I want to show things from the other side: what happens when you arrive, apparently, at your destination.

Knowledge is good. Better knowledge-rich than knowledge-poor. But the version England built has coupled grand rhetoric with low standards and has produced the widest rich–poor ‘curiosity gap’ in the OECD’s worldwide PISA data. Even granting Daisy Christodoulou’s case that the curriculum lifted PISA scores, we have record lows in students’ enjoyment of school and record highs in pressure from schoolwork.

I don’t think building a knowledge-rich curriculum requires removing teacher autonomy. But England has done both. This is the story of what happened.

What is Total Instruction?

Ask yourself:

1. If I felt I’d planned a lesson well, then found out an observer was coming in next period, would I change something?

2. If a lesson was going well but an observer suddenly appeared in my room, would I feel a pang of panic that I hadn’t included something or done something I should have done?

If yes to either, your school may be imposing aspects of Total Instruction.

I’m not against accountability – we owe it to students to justify the good our schools are doing. I’m not against standardisation across a school – it’s impossible to run a school otherwise. I’m not against routines – which are often a part of effective parenting, as per my analogy last post, as well as a necessary aspect of school life. What concerns me is the removal of teacher autonomy under the guise of evidence-based practice, when the practices themselves outstrip any evidence.

We have two axes: horizontally, how much freedom teachers have in implementing practices – often, low autonomy is associated with higher standardisation and monitoring; vertically, how much whole-school focus there is on teaching and learning practices – high can be supportive, when coupled with autonomy, or feel surveilled without.

Supported autonomy is, in many cases, the ideal: teaching and learning leads share evidence-informed ideas and model good practice, but teachers keep the freedom to combine techniques using their own judgement.

Total freedom – little guidance, complete classroom autonomy – barely exists in England’s state sector now that national accountability has tightened.

Skerritt’s teachers can give us a sense of how it feels to fall into one of the left quadrants. Unsupported accountability is the environment that makes Bronagh feel sick at the thought of another observation. It’s the veneer of accountability without the clarity or support to meet expectations.

Then, encapsulated by Aoibheann’s experience, there is Total Instruction: ‘Everything you do as a teacher you are told to do.’ As I mentioned in the last post, nobody explicitly advocates for Total Instruction – online, in their school, in policy. Typically, it’s a hodgepodge of different ideas picked up from other schools, professional development events, books, magazines, blog posts, and leftovers from old regimes. These aren’t a problem in themselves; the key move comes when they’re formalised as ‘good teaching’ and mandated for all teachers.

This can be more or less extreme. It might be a few techniques that must be present in every lesson. Or it can be full, scripted lessons that allow for little or no teacher discretion. And the last point – teacher discretion – is key.

Where does Total Instruction come from?

As Haili Hughes put it, ‘What begins as professional guidance becomes managerial compliance.’ Total Instruction is a top-down phenomenon. England’s Department for Education’s own research shows pedagogy gets standardised when schools are failing: they’re taken over by better-performing schools or trusts, and managers assume that the techniques or practices that worked in one setting can be exported, wholesale, to another.

But when a struggling school improves, who picks cause from effect? The mandated techniques often get the credit over better management, tasks cut to free up teacher time, or a new approach to behaviour.

And autonomy is a one-way street: standardisation is the first reaction to ‘falling standards’; autonomy is rarely suggested as a tool for improvement.

Can scripts ever be justified?

There are two cases where we may be justified in curtailing teacher judgement. The first is where strong structure has good evidence behind it and no sign of collateral damage. Complete, evidence-based programmes – structured early-literacy sequences are the clearest example – are not Total Instruction. You may dispute their choice of evidence, but the evidence, the assumptions, and the translation into practice are transparent, and can therefore be contested. Total Instruction is the opposite: techniques scraped from several sources, assembled ad hoc, then mandated as though they were a coherent programme.

And even in the strongest case for structure, autonomy doesn’t vanish: you can support structure in teaching without turning teacher judgement into a problem to be eliminated. Some of the better-structured programmes separate ‘tight’ elements – the skill sequencing, to be followed faithfully – from ‘loose’ ones – pacing, use of examples, formative assessment practices – that rely on teacher judgement.

England has been running this experiment for fifteen years. Four-fifths of state schools have taken part, as the Education Endowment Foundation has pitted programme after programme against ‘business as usual’ – which, in many classrooms, still involves teachers using their own judgement.

The EEF has discovered some high-quality, well-structured programmes, but business as usual has proven hard to beat. A 2019 analysis of 141 randomised controlled trials conducted in the UK and the US, including the EEF’s – involving over 1.2 million students in all – found that interventions, on average, gave about a month’s extra progress (to use the EEF’s own metrics) compared to the control group. About 40% were statistically uninformative. Research commissioned by the EEF themselves found that the effect of seven promising programmes shrank from about three months’ extra progress to no progress, on average, when scaled up.

The second case for curtailing autonomy is narrower. Comparing classroom video and international test scores across eight countries, a UCL team found that pupils of inexperienced teachers may do better when autonomy is constrained. Even there, the teachers with the least autonomy were the most likely to report lower job satisfaction.

Stressed, controlled, directed

Every few years, in their Teaching and Learning International Survey (TALIS), the OECD ask teachers from across the world a series of questions about their workload, training and how they feel about the job, among other things.

I’m going to take the 2018 data. In part, this mitigates the effects of the pandemic, but still gives enough time to see the consequences of any major changes associated with the knowledge-rich curriculum. The main reason, though, is that England didn’t take part in 2024.1

I’m going to re-create the diagram above, plotting real countries (the code and methodological caveats are here). On the horizontal axis, I’ve looked at the degree to which teachers agree they have control over determining course content, selecting teaching methods, assessing students’ learning, disciplining students and determining the amount of homework to be assigned.

On the vertical axis we’ve got the degree to which teachers agree they have whole-school collaborative teaching practices: teaching jointly as a team in the same class, providing feedback to other teachers about their practice, engaging in joint activities across different classes and age groups, and participating in collaborative professional learning. Axes are set to ‘zero’ on the country averages. I’ve also chopped off Chile and Italy because they were way down on the vertical axis, distorting the spread.

Percentages show the top 20 countries according to how many of the teachers surveyed agreed: ‘I experience stress in my work’.

Here are the averages for workplace stress by country according to quadrant:

While differences are relatively small between countries and quadrants, England’s teaching workforce edges into second above Hungary, by this measure, in terms of work stress. And if there’s a candidate for a country that embodies Total Instruction, according to the graph, it’s England.

Reader Max asked about PISA scores per quadrant, although this analysis comes with some pretty big caveats, as per the footnote.2

Was it all worth it?

As UCL professor John Jerrim has put it, in England ‘we have lots of accountability in our school system, and also a lot of accountability-driven stress among teachers’. Education Support’s Teacher Wellbeing Index provides the latest UK-wide figures here.3

A study of over a thousand teachers found autonomy strongly correlated with job satisfaction, perceived workload manageability, and intention to stay in the profession – and that the average teacher reports less autonomy than professionals in other fields. Tellingly, it isn’t just the hours: earlier work by Sam Sims found job satisfaction tracks whether workload feels manageable, not how long the hours are.

The promise of the knowledge-rich curriculum, traceable back to E.D. Hirsch, was to close the attainment gap between rich and poor students. England’s disadvantage gap, though, is as large as ever. The pandemic hit disadvantaged students most of all, but even before that, progress was painfully slow. In 2020, the Education Policy Institute calculated it would take 500 years to close the gap. In fact, their analysis suggested that even before the pandemic, the gap had stopped closing. If moves towards mandated pedagogy have been justified on the basis of closing the gap and increasing standards, they aren’t working.

What should we do about it?

Good teaching requires what Aristotle called phronesis, or practical wisdom. This can’t be taught, only gained through experience. But a supportive school environment helps teachers learn from their experience, and allows them the freedom to experiment and combine techniques to suit their classes and approach.

A skilled teacher’s mind is like a Swiss Army knife; whole-school teaching and learning guidance can help sharpen the tools. Most weeks throw up a situation the teacher’s never faced before, but, almost automatically, they can pull out and combine the techniques needed to tackle it. Reteach or plough on? Remove the child or take them aside for a quiet word? And how, knowing the class and the subject as you do, can you ignite their curiosity?

Each decision draws upon the results of a thousand that have gone before.

The opposite is a complete script: the human made predictable, judgement stripped out and replaced with something measurable and narrowly accountable. Multinational corporations in other industries have done exactly this as a halfway point to automation.

England’s Department for Education is beginning AI tutoring trials this autumn. The tools will ‘reflect national expectations for pedagogy’ and ‘drive up standards by complementing high-quality face-to-face teaching – not replacing it’. By the end of 2027 these tools will have reached up to 450,000 disadvantaged pupils in Years 9 to 11.

An AI tutor can be rich in knowledge. It can follow a script to the letter. It can structure the learning journey and give instant, personalised feedback.

Know what it can’t do?

It can’t read Alyssia at the back of your Year 8 class. Usually, she smiles. Today, she won’t make eye contact. Tired? Bored? Something going on at home? The AI tutor logs a dip in her scores. You can’t track that, not in the moment, and not with thirty kids in the class.

You’ve got a break duty, then you’re teaching again and you’ve got a meeting after school and a stack of unmarked mock exams sitting on your desk. But you still hold Alyssia back to find out what’s going on.

Because teaching was never about following the script, or pulling the approved technique from the bag to pass an inspection. It’s about a child’s experience of school. And that’s something an AI tutor can never understand.

I’d left a centuries-old, top-performing private school. It had no behaviour policy. Nobody told the teachers what to do in their lesson. The corridors were a boisterous jumble of bodies and there were so many different sports happening at breaktime that walking across the playground was a health risk.

Although I didn’t know it at the time, the school I was interviewing at was an imitator of the Michaela Community School in London, dubbed ‘the strictest school in the country’. All I noticed then was the eerie silence as students moved between lessons and toilets without doors – to make sure nothing untoward was happening inside.

At the interview, the deputy head asked me to describe my ideal lesson.

I talked about an A level physics lesson in which I used a worksheet to lead students through a complex derivation. Most of the students were able to get there on their own, using the hints on the sheet – a considerable achievement in my book, I told him.

He looked up at me. Then asked me to discuss my approach to teaching.

I talked about exciting children’s curiosity. Using what they knew already as a bridge to discovering something new.

‘Would you say you endorse discovery learning?’ he asked, scribbling away on the sheet in front of him.

The phone call I got later that evening wasn’t a surprise: ‘we don’t do discovery learning here,’ he told me.

I later found out from a friend who was doing a teacher training placement at the school that I’d become infamous. ‘So you were the idiot who talked about discovery learning at the interview!’ she told me.

Is enquiry learning malpractice?

Greg Ashman recently published a piece arguing that there are times enquiry learning can be malpractice. In the comments section, a reader criticised an academic paper Ashman hadn’t referred to. Ashman defended the paper’s argument: if adding guidance to enquiry learning strengthens it, why not go all the way to full guidance – direct or explicit instruction?

Four reasons, I’m going to argue.

One: we can do direct instruction. Doing ‘moderately guided instruction’ (more on what that actually is later) does not preclude actually explaining things to students together with everything else that goes alongside direct instruction.

Two: direct instruction isn’t as far as we can go. There’s pressure on teachers to go further still and to strip out the judgement and flexibility that good direct instruction depends on. I’ll call that end point ‘total instruction’.

Three: there’s no decisive evidence that direct instruction beats moderately guided instruction.

Four: it’s an odd question to ask anyway. Imagine researchers asking which technique the most effective parents use to bring up their children – direct instruction or discovery learning? The fact we’re asking it tells us something about how we see the role of the teacher.

Before that, though, what was that paper the reader referred to?

Meet the academic paper that cost me a job

That interview was my first contact with the phrase ‘discovery learning’. The fact it happened in a Michaela-inspired school is no coincidence.

The Michaela Community School was ‘surely’ motivated by the 2006 work of Kirschner, Sweller and Clark, claimed schools minister Nick Gibb in a speech.

It seems an obscure thing to mention. The Michaela Community School is still going strong, but Nick Gibb is long gone and this academic paper is now 20 years old.

But the legacy of both the paper and schools minister remain. Kirschner, Sweller and Clark (2006) – KSC – is one of the most cited education papers of the century. It’s been repeatedly cited by government ministers, too. Education secretary Michael Gove claimed:

The most impressive scientific evidence on how children learn - from experts like Paul Kirschner, Richard E Clark and John Sweller - all points towards the importance of direct instruction. Their work on ‘why minimally guided teaching techniques do not work’ is hugely powerful.

More anecdotally, it’s the only academic paper I’ve seen referred to directly during school professional development – and it’s come up more than once.

A steep learning curve

Education researchers have gone over the KSC paper with a fine-tooth comb. There was a whole issue of Educational Psychologist dedicated to rebuttals, in fact (KSC replies are here).

I’m not going to do that here. I’m going to acknowledge the positive impact, then point out what we’ve lost as a result.

First, the good. When I started teaching fifteen years ago, I thought that group work and tasks with little guidance were signs of a forward-thinking teacher. I thought regular social interaction would guide my students through Vygotsky’s Zone of Proximal Development. I left students to play around with practical equipment, then got frustrated when they were confused or made little progress.

I don’t do that now, obviously, because experience has shown me it doesn’t work. More importantly, trainee teachers don’t do it either. Students learn more and teachers have a better experience as a result.

But we’ve lost something along the way.

Two things, in fact.

First, those tasks I’d dreamed up might have worked, with better planning, guidance and management. This is ‘moderately guided instruction’, and we’ll return to it below.

Second, if teachers are required to use only variants of direct instruction, it is unlikely that teachers and students will be well-equipped to get the most out of moderately guided approaches like structured enquiry, even if these were encouraged. Training, feedback, stronger models and a transition period would be required to embed these in lesson – alongside rather than as a replacement for direct instruction.1

I attended a masterclass by Tom Sherrington at researchED on Saturday about how to run paired talk.2 Since covid, he said, student talk is dying out in classrooms. Teachers struggle to manage class discussion. It’s easier not to bother.

He wasn’t suggesting a return to discovery learning. He wasn’t saying we need to let students discover scientific laws themselves. He was saying that teachers are afraid now of using well-controlled turn and talk, something that can be a central part of properly executed direct instruction.

But, to me, it’s a sign of how far the influence of KSC has spread as much as it is the effects of covid.

We’ve created a new form of teaching: total instruction.

The KSC dilemma

In the paper, Kirschner, Sweller and Clark present teaching as if it’s this:

Some of the rebuttals point out we should see it more like this:

But then, as Ashman pointed out, if you think going from the left to the middle is a positive step, why not go all the way?

At this point, I’m going to move into conjecture.

Good explicit instruction can involve peer discussion, open questioning and student independence on tasks once they have the requisite knowledge to tackle these with a reasonable expectation of success.

But I’m willing to make a bet – go into classrooms and see how often these things occur. They all involve danger, risk, teacher judgement. Much easier to cut them back and stick with the false stereotype of direct instruction – that it’s simply the teacher talking and students following instructions.

This is not the fault of KSC and it is not what they argue for, but I see it as a sign that their argument has been twisted to suit a particular position: teachers have knowledge and must pass this on didactically by talking; teacher judgement should be removed and students are passive recipients of information. This degrades direct instruction which, at its most effective, relies on exactly the judgement being removed.

This isn’t an enquiry vs direct instruction argument. You can be the firmest believer in explicit teaching and still think the teacher should be trusted to decide how to use it. Total instruction has no defenders – only a system that finds teacher judgement inconvenient.

Haili Hughes, also speaking at researchED, discussed this kind of move – stripping out what makes an approach effective to leave something impoverished in its place. Following Ed Haertel, she termed it a ‘lethal mutation.’

What are all these things?

First, a couple of definitions. Part of the problem with the KSC paper is they don’t give that many, so I’ve tracked down some given elsewhere:

Constructivism: The idea that everyone constructs ideas, theories and ‘mental schema’ for themselves. You can’t hand someone knowledge like you hand them a restaurant bill. This is how the brain works – and nobody in the debate disputes it.

Direct/explicit instruction:3 The teacher gives ‘full’ guidance (can be via demonstration, video, simulations etc); can include class discussions and activities. For problems and tasks, step-by-step instructions are given, and the guidance withdrawn slowly as students become more competent.

Unguided/minimal guidance: Students are expected to ‘discover’ some or all of the key ideas they are supposed to learn themselves, e.g. ‘students receiving partial instructional guidance may be given a new type of problem and asked to brainstorm possible solutions in small groups without prompts or hints’.

Moderately guided instruction: ‘The student receives problems to solve but the teacher also provides hints, direction, coaching, feedback, and/or modeling to keep the student on track’. Also called, in the literature, guided discovery, enhanced or assisted discovery, and scaffolded enquiry.

The key isn’t the activity but the leap: the student is scaffolded toward something they haven’t been walked through first, and this is what separates it from well-conducted explicit teaching.

The student needs enough prior knowledge for the leap to be reasonable, and that’s where a teacher’s skill, experience and judgement come in. The novice/expert distinction is fine, but we shouldn’t use it as a proxy for ‘not old enough yet’. A seven-year-old can be ready for a cognitive jump in one context, while, in another, a graduate might not be.

What does the evidence say?

Let’s start with the older evidence, that was around when KSC were writing. Mayer writes ‘there is little reason to believe that pure discovery will somehow work today’ then summarises his view of the evidence:

In many ways, guided discovery appears to offer the best method for promoting constructivist learning. The challenge of teaching by guided discovery is to know how much and what kind of guidance to provide and to know how to specify the desired outcome of learning. In some cases, direct instruction can promote the cognitive processing needed for constructivist learning, but in others, some mixture of guidance and exploration is needed.

Mayer concedes that ‘in some cases’ direct instruction is needed but in others ‘some mixture of guidance and exploration’ is needed.

Why draw on Mayer? Because it’s a key piece of evidence KSC use to argue that discovery learning doesn’t work.

Let’s move onto more recent studies.

Meta-analyses gather together a group of studies. Inevitably, some of the studies will be stronger or more relevant than others – this is always a risk with meta-analyses. Overall, they’re a useful way of summarising a large body of evidence. A 2011 meta-analysis of 164 studies first compared explicit instruction to unassisted discovery. In EEF terms, the finding was 5 months’ more progress in favour of explicit instruction. Second, though, they compared enhanced and/or assisted discovery to other forms of instruction including – but not limited to – explicit instruction. ‘Assisted discovery’ – a form of moderately guided instruction – came out on top, giving 4 months’ extra progress.

Another meta-analysis found that guidance reliably improves enquiry learning, but there’s no single correct ‘dose’, and ‘too much guidance inevitably challenges the inherent nature of the inquiry process’.

Other evidence is mixed. Large-scale international data (PISA) has been used to argue for or against highly teacher-led instruction. It also rests on students’ own reports of their lessons, so it’s hard to use it as definitive support for a particular method of instruction.

There’s also the EEF’s RCT on project-based learning published in 2016 involving 4,000 students across 24 schools. The drop-out rate was so high that the EEF themselves gave it a ‘security’ (reliability) rating of only one padlock out of five. Still, they concluded that while the project had no clear impact on literacy (and may have been detrimental):

We found that PBL was considered to be worthwhile and may enhance pupils’ skills including oracy, communication, teamwork, and self-directed study skills.

At best, a responsible commentator would conclude three things:

First, ‘pure discovery learning’ doesn’t work. But then, nobody is defending pure discovery learning. Decades ago, there were genuine calls for minimal instruction; there aren’t now. I’ve spent fifteen years in schools and half a dozen in university departments of education, and I’ve never heard anyone suggest we let kids loose to educate themselves. Even professors of education have met enough children to know they wouldn’t all race to the library the moment you shouted ‘school’s cancelled!’

What we should be debating is how much guidance enquiry needs – not whether to abandon all guidance.

Second, robustly comparing moderately guided instruction with direct instruction is hard – almost impossibly so, given the degree of teacher judgement both involve. An RCT comparing two groups’ grades also can’t capture most of what we want from education: long-term retention, transfer to novel problems, motivation, oracy, self-regulation, etc.

Third, and most fundamentally, the question assumes there’s a single best method waiting to be found. There isn’t – there are as many ways of teaching as there are teachers. The real question isn’t ‘which method?’ but how much leeway we give teachers to decide.

To see how odd the original question is, look at another place children learn: the home.

The point of parenting

Nobody is asking which instructional methods the most effective parents are using at home so that we can get less effective parents to follow their lead.

Let’s take two scenarios: a) teaching about traffic safety; b) how to use a climbing frame. Both involve physical safety, but a parent could teach their child very effectively in both situations and be justified in going to each end of the spectrum – allowing no autonomy and complete explanation about traffic, whereas allowing full autonomy on the climbing frame:

Even when we think about a knowledge-focused task like supporting children with their homework, a parent might choose a point along the spectrum depending on any number of factors like the child, their mood, the importance of the task, or how much they’re struggling with it.

Ultimately, the goal of parenting is to equip children for life. They need increasingly challenging situations. We need to understand their capabilities and build up their autonomy. We need to let them know there’s a safety net. But not to the extent they immediately let go and fall back into it without putting in the effort first.

Which parents fill their kid’s day with highly structured activities, charts, timetables and rules?

It’s not the best parents but the most worried.

And at the moment, it isn’t the teachers who are worried – education systems are worried on their behalf.

Why settle for anything less than total instruction?

I’m not arguing that direct instruction causes learned helplessness. I don’t have the data, and I don’t believe it anyway. What I do believe (but still don’t have the data for) is that only using direct instruction in every lesson is unlikely to equip students well for ambiguous situations, ones in which they have more freedom but less clear instructions or outcomes.

Coincidentally, this is also the point curiosity comes in. To be curious – to pursue curiosity – you need to be comfortable with uncertainty. You need to sit with the unknown and acknowledge your own ignorance.

That’s a discussion for another day.

To return to the central question: which is better – direct instruction or moderately guided instruction? The evidence is, at best, mixed, probably because there is no single way of teaching that is universally effective for all the outcomes we want for our students.

As with parenting, the obvious answer is that teachers should do a mixture of both. But this depends on teacher judgement and that scares some people.

As I found out when I began teaching, moderately guided instruction can be difficult to do. It requires agency, skill, understanding of the needs of the class and the individuals within it.

We could train teachers in this. We could give them scenarios and let them watch skilled practitioners at work. We could let them try it – and maybe fail, as I did – then give them feedback.

But it’s easier to tell them any form of instruction that isn’t maximally guided doesn’t work.

So the question has become the inverse of Ashman’s challenge.

Why settle for anything less than maximally guided ‘total instruction’?

First, because we value teacher judgement. Teachers should have a say in how much guidance is the right amount of guidance, not just policymakers, especially in the absence of decisive evidence to the contrary.

Second, the case for guidance is strong; the case for stripping out judgement is non-existent. Moderately guided instruction can be effective. Direct instruction can be effective. Total instruction is a lethal mutation – the end point of a system that has stopped trusting its teachers.

There’s an alternative: help teachers to develop the skills needed to judge when to use each type of instruction.

If we don’t train teachers in this, is it a surprise they do it badly? It’s a vicious cycle. We should be asking why we aren’t trying to break it.

England’s Department for Education is about to begin an AI tutoring trial, to complement rather than replace teachers.

Let’s see if they’re still saying that when the results of the pilot roll in.

Of all the times in the history of education, this is the most dangerous to be systematically deskilling teachers.

Then I looked at the papers I was handing back. ‘Well done’, I’d scribbled across a paper that had got 55%. ‘Great work’ on another that had got 48%.

Most of these kids had sat the ‘foundation tier’ – the lowest difficulty paper, the one that cut out the hardest material. Some of the questions were pretty straightforward, like calculating an average or plotting a couple of missing points on a graph. Others were even easier:

Not all of them got this question correct.

I’d always prided myself on holding my students to the highest standards. They were a couple of months off being done with school forever, never to take physics again. I was sending them out into the world knowing a fraction of the knowledge-rich curriculum I had been presenting them with for three years.

I thought I was doing everything right: Retrieval starters, mini whiteboards at every opportunity, carefully sculpted explanations with no distracting, twirling PowerPoint animations. No hint of problem-based or discovery learning in my class and the only practicals we did were the ones fully embedded in the scheme of work or required by the exam board. I’d read every kid’s SEN report and tilted the tables by sitting the most disadvantaged kids at the front of the class.

All this, and if they took the higher tier physics paper, the top scorers in the class would probably get between 20% and 30%

Where had I gone wrong?

A very Conservative revolution

When the Conservative government came to power in the UK in 2010, ministers began their revolution of teaching in England with big promises. There would be new exams, ‘designed to address the grade inflation, dumbing down and loss of rigour.’ The ‘knowledge-rich curriculum’ would now contain the ‘essential knowledge and skills which every child should master’.

Ministers also acknowledged that ‘failure to secure a good maths or English GCSE renders any student effectively unemployable.’ If students did not meet the required standards in mathematics and English, they would retake the exams until they did. And, naturally, the idea of students sitting an easier paper and being limited to just passing would be a thing of the past.

Michael Gove, Education Secretary and architect of the changes, warned that results might dip when standards increased. But that was only because they were ‘honest’ in a way previous exams hadn’t been.

The poverty of low standards

The rhetoric was strong, but underneath, Gove was undermining his own high standards. Rather than push for a system where students really understood everything on his ‘knowledge-rich curriculum,’ he let them leave school grasping only a fraction of it.

Let’s take one of those core subjects – mathematics. Let’s take the biggest exam board by entries – AQA. And let’s look at the grade boundaries for the new ‘good pass’ – a grade 5 (on the scale 1, lowest, to 9, highest).

(After 2019, pass grades changed dramatically, and so I’ve not included these. ‘Standards fell as a result of the pandemic’ would be an easy target. I’m interested in the immediate changes.)

At around 30% for a grade 5, three students could be awarded a ‘good pass’ and allowed to move on to the next stage of their education and into work, while having no overlap at all in their understanding of the curriculum, with its ‘essential knowledge and skills which every child should master’.

The situation is worse, in fact. Because Gove set the effective standard of a pass at grade 4. If you can get a fifth of the questions right on the paper, you don’t have to retake your exams. Knowing only 20% of that knowledge-rich curriculum, apparently, transforms your prospects from ‘effectively unemployable’ to government-certified school leaver.

And, as you will have spotted, Gove did not remove the ‘foundation tier’ in all subjects. Nearly every student in my class is limited to a grade 5, moving those aspirational top grades out of reach.

The Curiosity Gap

The knowledge-rich curriculum was designed to be the great social leveller. Michael Gove, drawing from others such as E.D. Hirsch, explained it in terms of the ‘Matthew Effect.’ Following the biblical passage, those who start school knowing most are the ones who gain most. They have the concepts in place to build on and so progress is easier. As Gove put it:

We cannot anymore ignore the evidence that shows that pupils from less advantaged backgrounds are less likely than their peers to access this ‘communal knowledge’ at home, who by contrast enjoy frequent guided reading with parents from a young age, as well as rich conversations at the family dinner table as they grow older.

Gove’s revolution sought to give all students access to this communal knowledge. But he forgot one vital factor: it wasn’t only access to the knowledge that differentiated between rich and poor, it was also the desire to gain that knowledge, when they were presented with it.

Some children may experience this ‘communal knowledge’ in art galleries, science museums or in discussion over dinner when their parents return from their jobs that require it. To these children, the value of school knowledge might be self-evident. To children without this range of opportunities, it may not be. That doesn’t mean they’re less curious, but it might mean they express it less at school.

Rather than guessing, though, let’s test the theory.

We’ll begin by looking at England alone, given it’s England’s knowledge-rich curriculum we’re discussing. In the 2022 PISA survey, the OECD gave children around the world tests of their science, mathematics and reading knowledge, as well as a questionnaire about their attitude to learning. Even if we restrict the data to England, we’re still left with almost 5000 students.

We’ll split the students into quintiles – five equal groups – from poorest to wealthiest. Then we’ll compare their answers to two of the curiosity questions (how much they strongly disagree to strongly agree on a 5-point scale with the following):

I am curious about many different things.

I love learning new things in school.

Between the wealthiest and poorest 20%, there’s little difference in how curious they feel. The gap comes in their attitude towards learning in school.

Another question: by 2022, we’d had the knowledge-rich curriculum for some time. Was it the great leveller it was intended to be?

Here are the test scores for the richest and poorest, which we’d expect to converge if indeed the knowledge-rich curriculum had closed the wealth gap:

Obviously we aren’t quite there yet.

At an average of almost 100 points between the wealthiest and poorest across all three subjects, England ranks 21st out of 75 PISA nations in terms of the size of its attainment gap.

If quintiles were nations, the wealthiest would rank second in the world in mathematics, just after Singapore. The poorest would be 39^th in the world, ‘statistically significantly below the OECD average’, according to the OECD.

This isn’t good, but it doesn’t mean that curiosity is the solution, nor does it mean the ‘curiosity gap’ is a uniquely English phenomenon. To investigate those two points, we need to go global.

Going global

If curiosity did correlate with better test scores, this might just be because it encapsulates a positive attitude towards learning more generally. So, let’s compare the effect of being curious to having grit or a growth mindset - while holding the effects of the other two traits constant.

Here’s the data, from over 80,000 students across 41 countries.1

According to PISA, 20 points equates to about a year’s worth of schooling. So, loving learning in school gives the poorest a boost of roughly 3 months compared to their equally-poor compatriots who do not love learning in school.

And it’s England specifically we should worry about.

We can calculate the curiosity gap (general curiosity minus love of learning new things in school) for all 75 regions in PISA. Here are the top 20:

And here’s that information for all 75 regions – with the attainment gap between poorest and richest plotted against their curiosity gap:2

In summary, the difference between England’s poorest students’ curiosity gap – general curiosity minus love of school learning – is 19.8 percentage points, against 11.0 for the wealthiest: the largest of any of the 75 regions in the comparison.

Poorer students who enjoy learning new things in school do better than those who don’t. If we believe we have a moral obligation to close the achievement gap, maybe we should be trying to close the curiosity gap, too.

The Moral Case for Curiosity

Daisy Christodoulou has argued that the knowledge-rich curriculum has held England’s PISA performance steady while the other devolved nations have fallen. I used the same data to show that English children’s enjoyment of school has collapsed over the same period.

Neither of us touched on curiosity.

The HBSC data Christodoulou cites shows only around 11% of English 13- and 15-year-olds say they like school a lot. The curiosity ranking suggests we now have a more specific problem: it’s not just that students don’t enjoy school; it could be that students - and especially the poorest - may place a lower value on the knowledge school teaches them.

Maybe it’s time to look at the curriculum in a different way.

We tell students about careers and the qualifications they need to get into university, as if this alone justifies everything we teach them. Each lesson is simply building up to this end point.

We talk about the curriculum as if it’s a gift to students. We treat it, though, like it’s something we do to them.

Justifying the curriculum – explaining why we think students should learn it – is something that should accompany the gift. We might not think students should get to choose what they learn, but if we believe it’s for their own benefit, what’s the harm in saying so?

This matters especially for disadvantaged kids who aren’t as likely, according to Gove, to enjoy ‘rich conversations at the family dinner table’.

We need to tell these kids why it matters for them to learn this ‘communal knowledge’. If we believe, as the architects of the Conservative revolution put it, that this knowledge belongs to everyone, we need to give them a case for possessing it.

In my last post, I made the case for curiosity in school from a scientific perspective. Children have access to more of their working memory when they’re curious. They remember more – even incidental information they come across. The latest neuroscience and cognitive science evidence indicates that curiosity is a high learning state.

Here, I’m giving the moral case for curiosity.

Rather than pretending the curriculum is self-evident, what if, before presenting them with the information, we told them why they were expected to learn it? From languages to mathematics, science to music, every single sentence in the curriculum would have a purpose.

And if we can’t justify each sentence? Maybe we need to rethink why we’re actually teaching it.

Curiosity – on an atomic level

This isn’t about making every lesson fun. It’s not about putting on costumes and dropping apples while pretending to be Sir Isaac Newton to learn about gravity. It’s about telling students why it matters to learn about gravity.

A couple of weeks after I gave those mock exams back, I was revising atomic structure with the class.

A student put her hand up. ‘But why do we need to know this?’

‘Because your first GCSE exam is in a month,’ I was about to snap back.

I caught myself.

Instead, I gave a rambling explanation centred on two ideas. First, the point of science is to understand the intrinsic structure of the universe – like, what the desk in front of you is made of, what you’re made of. When we find an answer, we ask a new question. Science is curiosity, systematised.

Second, I asked how long it took her phone to charge. For about ten minutes of that hour, on average, her phone had been powered by nuclear energy. Nuclear energy requires an understanding of atoms.

And the minutes that were powered by fossil fuels, driving global warming, sending species extinct and flooding whole communities? If we had nuclear fusion, fossil fuels might be a thing of the past. Maybe, when we understand more about atoms, we’ll have it.

It took a couple of minutes. Then we got back to the lesson.

During the starter, the student had been staring blankly at the desk. Now, her eyes lit up. She raised her hand to ask another question.

If only I’d tried this two years ago, I thought, when we had all the time in the world.

2

Reader Max has just asked how meaningful the trend is. Great question. We can look at three figures to help us decide, all labelled top left:

The p value means it’s statistically significant (<0.05 is normally taken as the threshold).

The R^2 means attainment gap explains about 9% of the variation in curiosity ‘gap of gaps’ (i.e. between general and school curiosity in poorer and wealthier students). A modest amount, but there are so many factors that affect an education system’s performance and whether its students say they’re curious, this isn’t that surprising.

The r is the correlation strength, which is probably the most important: about 0.3 – on the cusp of small and medium. Which isn’t huge, but education trends generally aren’t. And it would be larger (about 0.38) without the outliers (Cambodia and Uzbekistan) on the top left

Nietzsche’s guide to becoming a multimillionaire

I was reading a self-help book called Goals! I’d reached a particularly fevered section about setting your sights on an ambitious financial target. Don’t worry about how to get there, the author said. Once you have the goal in mind, you’ll find a way.

Then he quoted Nietzsche.

He who has a ‘why’ to live can bear almost any ‘how’

I wondered, briefly, about whether meeting financial goals was what Nietzsche had been getting at. Then I realised it didn’t really matter, and that it could apply to teaching, too.

Give students a strong enough ‘why’, and they’ll put up with some variation in how you get them there.

But there’s also the converse: if we don’t give them a ‘why’, does it really matter about the ‘how’?

Boring holes and whale carcasses

I’ve been testing an idea recently in the classroom. It goes something like this: pose a question, withhold the answer.

How long can a whale carcass sustain life on the ocean floor?

Why can’t astronauts just shout to each other on the moon?

How do we know about the Earth’s structure, when the deepest hole we’ve ever dug was 12km?

I ask students to make a prediction but don’t tell them the answer. The ‘delayed closure’ approach. Instead, I just let the lesson unfold, because the answer’s there, inside all the activities we’re about to do.

Take the question about the Earth’s structure. I taught that lesson earlier today. It had all the makings of a curiosity disaster: I discovered five minutes in that the students had just found out about the Earth’s structure in geography.

But I had a secret weapon – delayed closure.

I told the students that the Soviets dug for 24 years. They drilled down further than anyone ever has. What percentage of the way to the centre of the Earth did they get?

Students guessed.

The answer? Precisely nowhere, given rounding errors.

If the Earth were an apple, they wouldn’t even have broken through the skin.

So if we’ve never made it through the Earth’s crust, how do we actually know anything about the Earth’s structure?

Somebody guessed X-rays. Not far off. We do use waves passing through the Earth – but these are naturally-generated. I told them about Danish seismologist Inge Lehmann, who discovered the inner core of the Earth in 1936 using earthquake data from New Zealand.

By then, they were as hooked as a class of 13-year-olds get. They raced through a comprehension activity in silence then came up with the most imaginative models of the Earth’s layers I’ve ever seen – a Big Mac, a watermelon, a gobstopper, a crème egg with a nut in the middle (the foil as the crust) – and explained these to the rest of the class.

But will they remember it?

I tested what they knew from their geography lesson at the start, and it wasn’t all that much. When I’ve tried the delayed-closure technique over the last couple of weeks, students retained more than they had done before.

So I’m hopeful. And I’ve got the science to back me up.

Seeing faces in MRI machines

In 2009, Min Jeong Kang and colleagues published a seminal neuroscientific study on curiosity. They put volunteers into an MRI machine and asked them trivia questions. When participants were curious, the reward pathways in the brain were activated. Participants remembered more answers to questions they’d been curious about than when they hadn’t.

In a 2014 paper, Matthias Gruber and colleagues went further. Between the trivia questions and the answers, participants were shown pictures of faces. When they were curious, participants remembered the irrelevant pictures better.

When we’re curious, our memories are activated, even for information that doesn’t seem relevant.

Gruber and his colleague Charan Ranganath argue that curiosity kicks in when we notice a gap between what we know and what we want to know. If the gap feels important or interesting, we’re motivated to fill it. During this process, the brain’s dopamine system ramps up, which strengthens memory formation in the hippocampus. We remember more because curiosity enhances our attention and desire to get the information we’re missing to close the knowledge gap.

Curiosity primes us to take in and retain information.

What’s really overloading students’ brains?

But I worried – if students are thinking about the question while you’re talking, aren’t they having to juggle a lot in their heads?

Then I thought – if they aren’t thinking about the big lesson question, what are they thinking about?

It doesn’t take much imagination to speculate: What’s going on at home. What their friend said to them in the playground. What they saw on TikTok when they used their phone at break when no one was looking.

Their mind goes to these things because that’s what they’re curious about. That’s what they want to know about.

The causes of the First World War, French grammar, fractions. Most students are not, by default, curious about these things. Certainly not in their abstract form. You introduce the lesson and their mind goes back to whatever they were thinking about on the way in. The cognitive load is whatever we’re adding – our voice competing with their thoughts. The fear they might be asked a question. The worry about how to complete a worksheet when they haven’t really been concentrating.

I’d got the problem the wrong way round.

The issue isn’t: does curiosity add to their cognitive load? It’s: how can we make them curious enough about our lesson so that their other worries and concerns fade into the background for an hour?

Sure, we can overload them with information if we don’t take proper care in our explanations, instructions and task design. That’s always been the case. But the neuroscience shows that curiosity improves retention. When students aren’t curious, they remember less than when they are.

If memory is the residue of thought, curiosity is the glue that helps those thoughts stick.

Freeing up some working memory

The working memory available to students isn’t fixed in practice. How they feel affects how much of their capacity they can actually use.

When students are calm, interested, or motivated, more of their working memory is available for learning. When they’re bored, anxious, or frustrated, some of that capacity gets tied up in managing those feelings instead.

This doesn’t mean lessons must be constantly fun – positive affect is more about comfort, emotional safety, and of course curiosity, than entertainment. The trick is directing the curiosity to the lesson material.

That’s where the thought and planning come in.

Luckily, the literature on curiosity gives us hints, and we can draw on studies from neuroscience, cognitive psychology and education. Piecing them together gives us clues about what might work in the classroom.

But more on that another day.

Once we’ve got the ‘why’, the ‘how’ becomes a lot easier. We can’t promise we’ll make our students multimillionaires (although doing better in exams can’t hurt). But we can give them a reason to listen.

Another big education research conference happens at the start of September, too. A lot of people go, but when they talk about ‘evidence’ and ‘methods’, some say they’ve got the wrong idea.

Some say they ask the wrong questions.

Some say they’re closeted and partisan and do not listen to outside views.

One is called BERA. It’s for academic researchers in education.

The other is researchED; it’s practitioner-facing, for teachers who want to understand the science of learning.

But which is which?

The problem I didn’t see

We’ve got a problem, but it took me a while to see it.

I spoke at BERA a few years ago. I overheard a couple of colleagues sharing a light joke about researchED. Standard stuff at academic conferences. You’re with your tribe. The other tribe is over there. Everyone agrees they don’t quite get it. Nobody loses any sleep.

I’d put money on versions of that conversation happening at researchED too, in reverse.

Two conferences full of people giving up their time to improve education. And what do we end up talking about? The failures of the other side.

It took a comment from a reader to spell it out for me.

Taking the Argument to the other side

Earlier this month I wrote a Substack post pushing back on Kelsey Piper’s piece in The Argument, which argued that education research is broken. I thought I was defending the field. I was – kind of.

But Marc, a reader, commented underneath and put me right.

Marc, who comes at education from the effective-teaching side, took the time to acknowledge what qualitative research can do that quantitative research can’t – and vice versa. But the point that really got to me was how research gets used:

This is what causes me to think that mainstream education academics are often ideologically-driven: research is good if it supports ‘good’ ideas.

He was applying the line to one side. I think it applies to both, and I’m going to tell you why.

I’d written 2000 words defending the plurality of educational research methods, and underneath I’d taken a swipe at the people I’d most like to convince.

So this is me trying to think about it more carefully.

Trads vs Progs. No, sorry, Reformers and Pragmatists

Let me try a reframe. Imagine there are two groups of people researching and spreading ‘evidence’ about education. (Evidence in air quotes, because they don’t always agree the other side really has any worth talking about.)

Let’s call one group reformers and the other pragmatists.

If you’ve spent time in these circles, this will start to sound familiar. Bear with me. The framing matters.

Reformers tend to think the system is broken and needs to change. They’re interested in students’ experiences, teachers’ experiences, the role of schools in reproducing inequality. They use qualitative methods alongside quantitative ones. They publish in academic journals read mostly by other reformers.

Pragmatists want to get better outcomes within the system as it currently is. They speak directly to teachers – often via blogs and books. They draw on cognitive science and the science of learning. They emphasise classroom experience and what works.

These are ideal types. Most researchers sit somewhere on a spectrum. But the public-facing versions of each camp tend to be sharper. Both contain serious work. Both have produced research that has improved schools – and children’s experiences within them. But both also produce caricatures that flatten the other side.

This makes the debate sound theoretical, but their disagreements shape what happens in schools: Should children be taught through direct instruction, or through more exploratory approaches? Should they learn a knowledge-rich curriculum, or one built around skills? How much should cognitive science shape day-to-day teaching? How should classrooms be run? How should teachers be trained?

Reformers and pragmatists don’t always sit on the opposite side of the fence on these debates – the issue is that they rarely debate at all. At least, not with the other camp. When the reformers acknowledge the pragmatists, they tend to dismiss them as reductive. When the pragmatists acknowledge the reformers, they tend to dismiss them as ideological.

Outside both camps – in the press, in policy, on social media – the dismissals get cruder still.

Why we should care

In February 2025, the US Department of Education’s Institute of Education Sciences was effectively dismantled – around 90% of staff cut, more than 100 active research contracts cancelled, the Condition of Education report missing its publication deadline for the first time in its history. The American Educational Research Association filed a lawsuit; its Executive Director called the cuts “nothing short of a fundamental harm to the education research community.”

The UK hasn’t seen anything like this. But the political appetite for ‘evidence-based’ rhetoric in service of preferred conclusions is not a uniquely American problem. As far back as 2013, Michael Gove was distinguishing between ‘good academia and bad academia’ – bad academia being, conveniently, the academics who disagreed with his curriculum reforms. Former Schools Minister Nick Gibb gave a speech in which he approvingly cited the view that university lecturers ‘have to justify their existence with all that pointless theory.’ Terry Wrigley wrote a BERA blog in 2018 setting out the same dynamic at greater length.

None of this is new. But the more we caricature each other, the weaker the public case for the field becomes – and the easier it is to dismiss. If you want to discredit education research, you don’t need to do the work yourself. You just point at the most heated bit of the trad/prog debate and say look, they can’t even agree among themselves.

At the moment, the two sides are like parents: currently estranged; each thinking they’re the one doing best for the child; fighting for custody.

It doesn’t have to be this way. Instead, we could look at what they have in common: a desire to improve the education system for those in it.

We could see pragmatists differently: as people serious about evidence; willing to translate research for the people who use it; unafraid to argue in public.

We could see reformers differently: as people who hold the structural questions in view and stop us treating each cohort of children as if the system around them didn’t exist.

How can we pontificate about learning when we close our own minds at the first sign of dissent?

The personal touch

After Covid, I quit teaching. It took me a while to work out why – in fact, I had to write a whole book proposal (cheaper than therapy, at least). It was a rollercoaster ride: half autobiography, half blistering critique of the educational system.

It ended with a conclusion. The system was broken. Broken because society’s incentive structure was all wrong. Exams matter because there’s limited room at the top. Outsized benefits for those who do best. An elite who want to protect what they have.

The solution? Rebuild society. Scrap the outsized incentive structures. A few simple changes. First, universal basic income. Second, huge taxation to reduce the ‘winner takes all’ attitude. Third, election by lot, so we don’t have such a huge ratio of Old Etonians on the ballot paper.

I thought it sounded reasonable enough and sent it off to a couple of places. A few polite rejections. And those were the positive responses.

Then the days arrived where you can’t draft an email without getting an LLM to redraft it (more accurately, politely, concisely and closer to your true voice than you could do yourself – although admittedly it’s probably not hungover and two coffees short of optimal functioning). I ran my draft through Claude, hoping it would tell me I’m a genius and to forget about the literary world (which is even more dysfunctional than the world of educational research).

Sadly, it didn’t. All that talk about sycophantic AI? Doesn’t seem like it when I run my ideas through it.

Claude told me the writing was ok, but who was actually going to read it? A teacher might agree with parts, then get to the end and think, ‘right, yeah. Cheers for that.’

Had I been a bit less reformer and a bit more pragmatist, I might have got further. When we have one side without the other, we often fail to get any message across.

Finding problems is the easy bit

I am – in many ways – an extreme example. But I’m reminded of my own ‘if this doesn’t work, nothing will’ conclusions when reading some academic papers and books. Academics are great at posing problems. Great at rigorous methodology. Even better at poking holes in other people’s rigorous methodology. Not so good at the ‘what do we do about this in the real world?’ question.

There’s something comforting about the change-society-or-nothing position. The chances of society being restructured as a result of this post are fairly small. Even if Sir Keir does read it and take drastic action, I can claim he didn’t quite do what I suggested when things don’t work out.

Finding problems is the easy bit. Suggesting practical changes is harder and more dangerous. You put your reputation on the line. Try this, you say. And you’re judged by the results.

So I’m changing my approach. Moaning about the exam system while doing nothing for the actual kids I teach isn’t satisfying. I’ve spent years researching factors that promote and suppress curiosity in the classroom. I’m going to use that knowledge – in my own classroom, and in my writing, in the hope that others can benefit.

The motive is partly selfish: it will give me agency. But there’s also a genuinely more hopeful form of curiosity to focus on – one that teachers have much more control over. I want to spread the message.

I’ll carry on poking holes in the system, because it’s what I do.

I just don’t want it to be all I do.

As a reformer at heart, I specialise in seeing problems. So here’s me exercising my pragmatic muscles. A few ideas for next steps:

Recommendations

1. Don’t weaponise the term ‘evidence’.

Everyone wants teaching to be evidence-based, but when we disagree about what the term ‘evidence’ means, the arguments will rumble on. We need to use the term more broadly. We can and should dispute methods and findings, but we should apply the same scrutiny whether the conclusions slot into our worldview or not. All evidence tells us something about the world; forgetting it shrinks our potential to improve children’s lives.

2. Become comfortable with uncertainty

It’s ok not to have all the answers. It’s ok to have questions. Especially when we speak to practitioners, we shouldn’t overstate the evidence. Data from the lab might be promising. Will it translate to the classroom? Who knows, we need to try it out. Equally, just because we have a small-scale study, we should not try to hide this. Be clear about the limitations and let someone else decide how they want to use your evidence.

3. Get curious about the other side

Attend a conference you would normally stay away from. Ask questions, even when you might get pushback. Don’t be afraid to show your ignorance. This is how we learn. Don’t attend to push an agenda – on either side. That closes doors. The flipside is to ensure transparency in our own methods and findings. Where possible, ensure others can scrutinise the work we’ve done, and prepare to enter into open debate.

4. Share what you’ve found

Education research isn’t blue-skies physics. The whole point is the practical benefit for society – broadly conceived. Researchers have an obligation to share what we’ve found, not as an afterthought, but as a central aim. Especially when that research is publicly funded.

5. Work together to draw up questions

We should discuss the questions, even when we disagree about how to reach the answers. We’re closer than we might think: how do children learn? What should they learn? What are the optimal conditions for this to take place? What other outcomes do we want for them? How do we assess these – including but going beyond exam results. Who is held accountable for all of this? Once we stop talking at cross-purposes, it becomes easier to have a constructive debate.

Moving beyond the old battle lines

Those are my suggestions for the field. Here’s how I’m going to start putting them into practice – on a personal scale.

I’m going to BERA in September to talk about the architecture of the school day, and how it suppresses curiosity in ways most teachers can see but feel powerless to change – hands-up questioning as the only sanctioned form of inquiry, practical work reduced to a tickbox protocol, social pressure that makes asking a question a confession of ignorance. That’s a critical argument. It’s drawn from my doctoral work. It belongs at BERA.

I’m going to researchED to talk about how teachers can engage students’ curiosity in lessons. Make students curious and they will be thinking about the things we’re teaching – and there’s good neuroscientific and psychological evidence underpinning the links between curiosity, learning and achievement. That belongs at researchED.

I think I can do both. More posts coming on this Substack. Some about the system. Some about the classroom. Sometimes both at once.

To me, this is what it looks like to advocate for the field. Disagreement welcome – comments section is open. The version of education research that survives is the one that’s argued for in public, not just in journals.

Thumbnail photo by Arthur Tseng on Unsplash

Research claims this happens. Nobody had proved it yet. Job done.

I didn’t do it, partly because educational research is hard.

You have to get ethical approval from your university, which takes months. You have to get children to sign up and take a questionnaire. Given the difficulty of getting them to complete a worksheet when they’re right in front of me and I can threaten them with detention, that’s a challenge. You’ve also got to get their parents to agree (and you’re not even allowed to trade their signature for a promise to look favourably on their kids’ coursework).

So I didn’t do it.

Which is lucky, because the OECD had already asked tens of thousands of kids these questions and found out – in a much more robust way than I would have done – what happens to kids’ curiosity when they get older. (I’ve written it up here.) The OECD also runs PISA, the international comparative study that governments treat as the gold standard of educational testing.

It’s not the sort of thing I do, though. I watch lessons. I speak to the children themselves. It takes a different set of skills from analysing 60,000 kids’ questionnaires.

There are lots of ways to do educational research – something I don’t think Kelsey Piper realises.

Before I get to that, we need to talk about Jo.

On Jo Boaler

Piper opens with Jo Boaler. The methodological criticisms are serious – the 2012 Bishop, Clopton and Milgram paper on the Railside misrepresentations; Tom Loveless’s 2023 review; a 2024 anonymous complaint to Stanford alleging 52 citation misrepresentations, which Stanford declined to investigate as ‘scholarly disagreement.’

In summary: Boaler disputes the allegations; it hasn’t swayed her critics; her work helped shape a controversial California maths framework; she’s still employed by Stanford.

But of course there’s a larger story here, and it doesn’t reflect well on education research as a discipline.

Peer review eventually did its job, but ‘eventually’ was after California had built its new curriculum based on her work. And look at where the pushback did come from: mathematicians, statisticians and policy experts. Where were the education researchers in all this?

And then there’s the Nature reproducibility paper Piper cites: 342 education papers examined; 69 contained quantitative claims that were eligible for reproducibility analysis (admittedly a small slice of the total quant work published in education over the decade); only a handful made it through the process; zero findings reproduced exactly.

Not everyone agrees the sample was representative – it omitted key journals and excluded qualitative work entirely. Still, there’s something we should all be able to agree on: quantitative education research can and should share code and data. Again, PISA’s the gold standard here.

Piper’s made a strong case, but here’s where it weakens.

The solutions she proposes – sharing analysis plans beforehand, conducting large-scale trials, hiring trained economists and ensuring methods are transparent – are not the discipline-changing blue-skies thinking she believes they are.

In fact, there are already a whole group of organisations doing some variation on what Piper proposes – evidence‑mobilisation and ‘what works’ bodies in education. You can find them in the UK, Australia, Netherlands, Chile, Spain (and the US to some extent).

I interned in one for six months. Let me tell you more about how they work – and when they don’t.

Inside the what-works project

When I began teaching, I wondered why we didn’t just run randomised controlled trials, like medicine does. Take two groups of kids, try something with one group, try something different with the other, see who gets the best grades. As Piper puts it: test curriculum A against curriculum B and see which one gets the best results, making sure your methods are transparent. Why bother with anything else?

While they have branched out, this was the founding idea behind the UK’s Education Endowment Foundation. Piper argues ‘bring in the economists, apply the standards of their field’. We’re already doing it. When I interned in the EEF’s evaluation team, most of the team had an economics background. (I’m not an economist, but it doesn’t take much digging to see that economics has its own replication problems – as Drew Bailey, cited by Piper, pointed out in the comments.)

The EEF asks how we can improve attainment across the board and close the gap between the most disadvantaged and the most affluent students. It runs RCTs. It publishes findings. It demands the kind of methodological transparency Piper wants.

And the results are sobering.

A 2021 review commissioned by EEF themselves found that across seven interventions, when these were scaled up, the effect size dropped on average from 0.25 to 0.01. Using their own conversion, before scaling, kids appeared to make 3 extra months of progress with the intervention; after, 0 months. Another study, reviewing 141 large-scale RCTs commissioned by the EEF and its US equivalent, involving over 1.2 million students, found the interventions gave kids about one additional month of progress.

This isn’t bad educational research. These are well-designed, well-funded, well-conducted trials. The results – when successful – are used in thousands of classrooms across the country. But modelling educational research on what goes on in medicine has its limitations.

Medicine tests interventions on individual bodies where the mechanism is relatively stable across patients. Across a population, a drug that lowers cholesterol in Boston will lower cholesterol in Birmingham (although even in medicine, focus is moving towards personalised interventions). Also, as Greg Ashman points out (see the comments), a doctor can’t tell a placebo from the real article; a teacher knows when they’re trying out a shiny new intervention versus doing business as usual - that awareness is going to affect how they teach it.

A classroom isn’t a doctor’s surgery. Most teachers will tell you they’re already trying to maximise their students’ exam grades. Maybe we’re now at the point of diminishing returns, given constraints like class size, funding, mixed ability grouping and timetabling.

Neither is a classroom a laboratory, where painstakingly-designed experiments can be faithfully reproduced again and again. Step into one (or, god forbid, stand near the interactive whiteboard) and you realise they’re a seething mass of interacting factors: the curriculum, teacher, students (and the dynamics between them), lesson material, weather and whether the kids have eaten lunch yet. A reading programme that raises scores in Manchester won’t necessarily raise them in Marseille.

It’s not about bad methodology; it’s the naivety of thinking we can do the same thing across every classroom and expect the same results.

Scale it up, surely, and the effect of these other variables dies away?

Wrong.

Scale up, as we saw, and the effect of the intervention often shrinks.

This is where a different type of educational research comes in, and it’s one the EEF already does. They don’t just run the trial, they also send researchers into schools to check whether the teachers are running the programme as they should.

This is qualitative research. It doesn’t produce shareable code or replicable regression coefficients, because it’s asking other kinds of question: what does this experience actually feel like to the people inside it? What makes a teacher abandon a programme in week three? What does a child do when she can’t keep up?

These are real questions. They have real answers. But the answers don’t fit the format the Nature reproducibility paper measures. You can’t just run the same code again across fifty interviews in R.

If you reduce education research to ‘how can we use methods from economics to transparently work out the best curriculum to maximise kids’ grades’ you miss something.

You miss the collateral when all you care about are these results.

What does it feel like to be labelled ‘low attaining’?

I wrote about Saffa’s experiences in a previous post. If you realise that school isn’t for you – if you struggle in class, get kept in at break and lunch to catch up and attend intervention groups – what does that do to your sense of who you are?

You can survey a thousand kids – or ten thousand – but you won’t get the rich details that Eleanore Hargreaves, Laura Quick and Denise Buchanan at UCL did when they followed 23 London children from the age of seven who had been labelled ‘low attaining’.

Take Saffa. She loved art. Learning about pointillism was her favourite thing at school.

Did that mean she’d rather do art than maths each morning?

No – art would become ‘quite meaningless,’ she said. ‘Because you have to do plus and take away and division and stuff.’

At seven, she’d already learnt which subjects counted and which didn’t. The curriculum hierarchy had overridden her passions. She also knew the stakes. If you didn’t listen in class: ‘You’ll just be a McDonald’s cooker, just flip patties. You will be unsuccessful.’ And when she had to leave her friends to join a younger class for maths, she called it ‘the walk of shame.’

What if we got Saffa a couple of extra points in her GCSE exams? Wouldn’t that transform her life chances? Maybe. But she’s seven. She’s already doing the walk of shame. Will those points be enough to transform how she feels about herself?

What about Chrystal, too, who feels low attaining kids like her end up alone: ‘no-one cares’ because ‘they have no friends to stand up for them’? Or Neymar, hiding in the toilets to avoid a maths test, despite how disgusting he finds the smell?

Seven years old. With nobody to tell their story except a group of researchers who may or may not have economics-level statistics knowledge, but who do have the deep training required to do qualitative research properly.

The problems with qualitative research

Qualitative research brings its own challenges, and transparency is one of them. The UCL team could upload their transcripts and observation notes, but that brings logistical challenges (my PhD interview transcripts total 150,000 words) and ethical ones – something in there might identify a participant. Name the school – as Piper suggests, and the comments underneath push back on – and you often have enough to name the participants.

These challenges aren’t insurmountable, though, and qualitative researchers should keep pushing for the highest standards (we just need to agree on what these are).

There’s also the issue of interpretation. Which of Saffa’s words do the researchers choose? How do they connect them with other students’ experiences? The UCL team are experts at this. Read their papers and you can see the depth required to do the work. If you think it’s easy, try turning 150,000 words of transcript into a coherent and representative narrative.

In choosing to interview students and observe lessons, I found something much more interesting than I could have got from a survey alone. I caught the moments where a student asked a question and their classmate commented under their breath, or a teacher cut it off. And I saw the moments where children’s faces lit up during a practical, or a teacher took a question and ran with it, bringing the lesson to life.

These studies are not better than quantitative research. They sit alongside it. Numbers can tell you how children’s curiosity changes between ten and fifteen. But unless you watch the lessons and speak to the students and the teachers, you cannot fully understand why those changes happen.

The case for education research

A couple of days ago, my daughter saw a book on the shelf – Guy Claxton’s What’s the point of school?

‘Isn’t it obvious?’ she said. ‘Learning.’

Even when I started teaching, I would have agreed. But the more time you spend in schools, the more children you speak to, the more you wonder: is this the best we can do for them?

Schools are happy to tell us what they’re for. They proudly display banners with their grand mission statements. Do they do what they say? Who holds them to account?

Inspections are a form of research. We use exam results to compare schools. Accountability is essential. But someone needs to hold the education system to account on behalf of the students.

There’s a place for large-scale and smaller-scale studies that investigate learning. The science of learning is complex and multidisciplinary, and there’s a lot we’re only beginning to understand about how the brain processes information. Piper’s prescription – better methods, more transparency, larger trials – would improve some of that work.

But it can’t tell Saffa’s story.

What Piper is calling for is a narrower field, not a stronger one. A field that produces only the kind of evidence her preferred standards can measure. Yes, it would be cleaner and easier to audit. But it would also be silent on most of what matters about what schools do to children.

If we jettisoned the rest of education research, we’d lose Saffa. And Chrystal. And Neymar. They’d disappear back into exam stats, where the only question we’d be allowed to ask is whether some intervention raised their scores by 0.06 standard deviations.

That’s not a stronger field. That’s a smaller one.

Thumbnail image by kyo azuma on Unsplash

There are empty seats in my classroom. When those students do come into school, it’s clear that many of them are struggling to cope.

Can we pin these struggles on school policy, or are the causes broader than this?

This is the question Daisy Christodoulou took up in a recent post.

I’m going to split my reply into three parts:

First, what I agree with.

Second, I’ll investigate the data she didn’t include.

Third, I’ll ask what the knowledge-rich curriculum aimed to do, and judge it against its own targets (well, one in particular).

One – what I agree with

Too often, we talk about what happens in UK schools when what we mean is England.

Also, teen mental health has fallen across rich countries for reasons that have little to do with whether or not we now celebrate ‘our island story’, as Gove famously asked teachers to do.

And in terms of PISA data (which I assume she’s referring to), while England hasn’t improved in absolute terms, given the international falls since covid, treading water is an achievement in itself. England has improved in comparison to the G7, whole sample averages and the other UK nations.

All of this seems reasonable enough.

She also conducts a ‘quasi-experiment’ using the devolved nations of the UK. Compare England to Wales and Scotland: if Scotland and Wales chose very different curriculum paths and ended up with similar wellbeing results, then curriculum policy can’t be doing much here.

This seems fair. And the numbers appear to support it.

But a closer look at the data she cites muddies the water.

What she argues

Christodoulou’s claim rests on the WHO’s Health Behaviour in School-Aged Children (HBSC) survey. She uses three charts of 15-year-olds in England, Scotland and Wales – life satisfaction, liking school, and feeling pressured by schoolwork – to point out how similar the three countries look.

From this she concludes:

Their education policies and attainment scores have definitely diverged. But happiness and wellbeing have not.

Her argument: if curriculum policy mattered for wellbeing, English 15-year-olds would be doing worse than their Scottish and Welsh peers. There is no difference. So no smoking gun.

That’s reasonable from the charts she shows.

But what about the ones she doesn’t?

What happens at 11 and 13?

HBSC doesn’t just survey 15-year-olds. It also covers 11- and 13-year-olds.

If curriculum reform is affecting school experience, these ages are exactly where we would expect to see it most clearly. Especially on the two questions that cover attitudes to school.

Christodoulou doesn’t include that data. Here’s why.

First, the proportion of pupils who say they feel pressured by schoolwork, in 2014 and 2022:

HBSC 2014: 11-, 13-year-olds who feel pressured by schoolwork, by gender and nation.

HBSC 2022: the same question, eight years later.

At 11 and 13, the picture is different.

On every one of these measures, the English increase is the largest in the UK, and on several it’s double the Welsh or Scottish change.

The second measure is the proportion who say they like school a lot. We’ll talk about why this is such an important question for the knowledge-rich curriculum in a moment.

First, here are the numbers at 15 again:

Christodoulou concludes: ‘The vast majority of English teens do not like school a lot. But the vast majority of Welsh and Scottish teens don’t either.’

The fall is largest for English 15-year-olds, but the findings are stark – lots of children don’t like school a lot.

Now let’s see what happens to those bright-eyed 11-year-olds just starting secondary school, and their 13-year-old peers.

HBSC 2014: pupils who like school a lot.

HBSC 2022: the same question, eight years later.

In 2014, English 11- and 13-year-olds were top of the liking school charts.

By 2022, England had fallen to the bottom.

For 11-year-old girls, liking school fell from 50 per cent to 21 per cent. For 13-year-old girls, from 26 per cent to 8 per cent — a drop of more than two thirds, the steepest fall of any UK group on any age band.

Again, the English change is larger than that of Wales or Scotland on almost every comparison.

These aren’t small gaps. With roughly 1,500 to 2,500 students sampled per nation at each age band, the 15-point divergence between England and its neighbours is unlikely to be statistical noise.

Christodoulou’s headline claim – that wellbeing has not diverged between the three nations – is true only at 15. At 11 and 13, it has, using her own source (and her own words), ‘definitely diverged’.

It’s also worth asking here if we’re looking at general wellbeing or something more specific – how students feel about school, and what school is doing to students.

But wasn’t this just Covid?

The obvious objection is timing. English pupils took the survey between December 2021 and April 2022, when the country’s schools had things besides a questionnaire to contend with: Covid’s Omicron wave, staff shortages, Year 6 pupils who had just sat the first post-pandemic SATs in a chaotic catch-up year.

It’s not all that surprising English 11-year-olds felt pressured.

But Covid hit Scottish and Welsh 11-year-olds too. If the pandemic was the dominant story, we would expect similar shocks across all three nations. Instead, the English change is substantially larger than the Welsh and Scottish change, both on school pressure and on liking school.

The pandemic affected every student in each country. Something above and beyond this is showing up in the English data.

Is it catch-up intensity in English primaries, the weight of the English SATs regime, accountability pressure passed down from secondary league tables, a cultural shift in how English primary schools operate?

It might not be the knowledge-rich curriculum, but Christodoulou could do a little more in investigating why the pressure ramped up for English students.

The symmetry problem

Christodoulou argues that global trends in teen wellbeing can’t be explained by English school policy alone.

But this isn’t quite what the data says.

She slides between three things – clinical mental health, general wellbeing, and how children feel about school – as though they’re the same category. They’re related, of course, but those changes in English children’s school experience deserve their own explanation, above and beyond smartphones and the pandemic.

There’s a bigger problem: Christodoulou also credits English curriculum policy for the fact that English attainment has held up better than Scottish or Welsh attainment on PISA, against a backdrop of global decline.

This is important. The Gove reforms were sold not just as attainment policy but as a route to greater motivation and enjoyment of school. If curriculum policy is powerful enough to buck a global attainment decline, it’s powerful enough to be part of the explanation when children report pressure and dislike of school – especially when its own architects promised it would do the opposite.

Why don’t students like school?

Michael Gove’s reforms were sold, in part, as a route to greater motivation.

In his 2012 speech ‘In praise of tests’ he told his audience:

Exams matter because the happiness I have described sustains future progress. We know that happiness comes from earned success.

He cited the cognitive science of Daniel Willingham – author of Why Don’t Students Like School? – and Willingham’s argument that genuine mastery produces the kind of satisfaction that shallow engagement can’t. In ‘The purpose of our school reforms’ two years later, Gove praised Willingham and Christodoulou by name as the thinkers whose work underpinned his curriculum changes.

Gove’s reforms weren’t intended just to gain a few more PISA points. Motivation and enjoyment of school were a key part of their justification.

Yet, sixteen years on, English children report the steepest decline in liking school of any UK nation.

The proportion of 11-year-old English girls who feel pressured by schoolwork has doubled.

We can’t necessarily blame the failure of Willingham’s cognitive science – again, this is a complex topic. But if Gove wanted us to judge the reforms on their ability to improve children’s enjoyment of school, we need to ask serious questions about whether they’ve been successful here.

More bluntly, if curriculum reform was meant to affect how children feel about school, we can’t now say that how children feel about school is none of curriculum’s business.

Why this matters

Just as we can’t prove that the curriculum changes caused England’s PISA success relative to its UK neighbours, the HBSC data can’t prove that knowledge-rich curriculum reform caused English children to feel worse. There are enough candidate explanations for the English divergence at 11 and 13 to publish a book on it.

But Christodoulou argues that the HBSC data lets us rule curriculum reform out as a factor.

It doesn’t.

11- and 13-year-olds’ perceived pressure due to schoolwork in England has increased relative to children in Wales and Scotland. Their enjoyment of school has plummeted. Any honest reading of the evidence has to engage with that rather than crop it out.

‘Sometimes the problems we think are unique to our own system are not,’ Christodoulou writes.

That’s true. Teen wellbeing – both general satisfaction and school-related measures – have fallen. They’ve fallen across wealthy countries, and Anglophone ones in particular, as she points out. No single policy lever will fix it.

But we could say the same for PISA scores. Well, they’re complex. Let’s just accept that’s the way they are.

An education secretary who tried this would be out the door the next morning.

Why don’t we pay the same attention to the statistics that are more meaningful to students themselves? The one in five children in England who are persistently absent from school. The 11-year-olds who feel pressured by their schoolwork. Those 13-year-olds who no longer like school.

Christodoulou is right that we shouldn’t blame Michael Gove for everything. But ‘not everything’ and ‘nothing’ are different claims.

The data she cites supports the first.

It doesn’t support the second.

In Assessing Children’s Learning, Mary Jane Drummond tells the story of Jason.

Jason is seven and a half. He’s been at school for nearly three years. Now he’s taking a maths test in the school hall with the rest of his class.

Out of thirty-six questions, Jason gets one correct.

Jason hasn’t learnt much maths, Drummond says. But he’s learnt quite a bit about how to be a good pupil:

‘He has learnt how to take a test. His answers are written neatly, with the sharpest of pencils. When he reverses a digit and sees a mistake, he crosses it out tidily. He places his answers on the line or in the box as instructed…’

Jason doesn’t understand a thing on the paper, but he knows he can’t get angry and turn the table over. He can’t shout. He can’t speak to the child next to him.

He knows he’s sitting a maths test, and he knows he has to write numbers.

What Jason has really learnt

Exams, as Jason has shown us, breed compliance.

I used to lead a team of exam invigilators. It was only when I found myself standing through my tenth assembly on exam regulations that it occurred to me what a strange ritual exams are:

Silence.

No communication of any kind.

No phones. No writing on your arm. Clear pencil case. You’re recognised by a candidate number, not your name.

Why go to such extremes?

Because pass or fail, you do so on your own merit. In that exam hall, there can be no excuses. No unfair advantages.

And the stakes are real.

In 2015, England’s Department for Education announced that ‘knuckling down and succeeding in school puts an average of £140,000 in a young person’s back pocket.’

(How much of a difference does ‘knuckling down’ at school really make? I looked at what the data actually shows in my previous post.)

By 2021, inflation had struck: the DfE estimated that gaining one grade higher across a full set of GCSEs was linked to around £200,000 more in lifetime earnings. The conclusion? Develop new policies ‘to help pupils achieve better GCSE grades’.

For the DfE, exams are the focus. They drive policy – and the rules ensure a level playing field.

Sure, there are concessions. Extra time for pupils who need it. A separate room if sitting the exam in a hall would be too much. Different coloured paper, different sized scripts.

All carefully designed to make exams appear scrupulously fair. There can be no claim that the odds were stacked against you. If you fail maths, that must mean you’re a failure in maths.

What other explanation could there be?

I’m bottom set in maths too

I recently taught a lesson to a class of 14-year-olds. It didn’t go all that well.

One of the students put her hand up:

‘Sir, are we bottom set?’

She’d nailed it.

For those unfamiliar: in many secondary schools in England, kids are sorted into classes by ability – based on previous test results. This class hadn’t done well on theirs. Research shows that having a lower household income or special educational need is a powerful predictor of those who end up in the bottom set, regardless of ability.

I didn’t want to answer in front of the class. I told her to stay back at the end if she wanted to talk about it.

I was hoping she’d forget. She didn’t.

‘I’m bottom set in maths, too,’ was all she said before heading off to lunch.

Not, I’m in the bottom set, but I am bottom set.

A couple of weeks earlier, in my first lesson with the group, another student told me she was stupid. It was one of the first things she said to me, as if preparing me for disappointment.

You’ll just be a McDonald’s cooker, just flip patties

These are not unusual moments. Any teacher in a comprehensive school will recognise them. What’s frightening is how early the process begins.

Eleanore Hargreaves, Laura Quick and Denise Buchanan at University College London followed 23 children labelled ‘low attaining’ from the age of seven.

Saffa loved art – learning about pointillism was her favourite thing at school.

Did that mean she’d rather do art than maths each morning?

No – art would become ‘quite meaningless,’ she said. ‘Because you have to do plus and take away and division and stuff.’

At seven, she’d already learnt which subjects counted and which didn’t. The curriculum hierarchy had overridden her passions.

She also knew the stakes. If you didn’t listen in class, ‘You’ll just be a McDonald’s cooker, just flip patties. You will be unsuccessful.’

And when she had to leave her friends to join a younger class for maths, she called it ‘The walk of shame.’

The children understood that the way to succeed was through what the researchers called ‘compliant hard work’: listening, concentrating, doing what you’re told.

Bob, who was struggling himself, prescribed the same medicine for a classmate who wasn’t doing well. What should the teacher do? ‘Let her stay for her whole lunchtime… Work!’

But compliance wasn’t protecting them. Jake worried that struggling children would be told: ‘Oh you’re bad at mathematics, oh you’re bad at English… Oh you’re not smart.’

Chrystal felt that children like her would end up alone: ‘no-one cares’ because ‘they have no friends to stand up for them’.

Ben dreaded passing the headteacher’s office, terrified they’d call him in and tell him he had to go down a class. The fear existed only in his head – but stayed there across several terms.

Neymar hid in the toilets to avoid a maths test, despite how disgusting he found the smell. When the teacher found him, he complained of a tummy ache.

All except two said school was boring.

Seven years old. Already ranking themselves. Already afraid.

Already convinced that hard work was the answer – inside a system that had already decided they weren’t good enough.

Stop napping!

These children’s experiences don’t happen in a vacuum. They’re produced by a system that has steadily constricted around what can be measured.

In the US, social studies, art, music and physical education have all been squeezed to make room for tested subjects. Even time spent on science – supposedly a core subject – has decreased by a third to make room for English and maths.

The pressure of standardised testing was so great that one school superintendent stopped kindergarteners from napping. Another school cut lunch to less than fifteen minutes.

In England, Alice Bradbury and colleagues found that children as young as five are being grouped by ability based on phonics knowledge. And it intensifies as high-stakes SATs approach at 10–11. They surveyed 288 primary headteachers. 35% agreed that ‘SATs mean we have to group by ability’ in English. In maths, this rose to 47%.

Several heads were uncomfortable with this. One said: ‘Pupils get into a psyche of failure because they’ve always been in the bottom set… You’re just written off.’

Another felt ability grouping was necessary so that the children would learn to cope with the test pressure, but said: ‘I worry about the effect this has on children’s maths mindsets and it goes against everything I believe in.’

But they felt they had no choice.

The measurement becomes the purpose

It’s impossible to put a date on the beginning of mass education in England, but 1862 was a significant year: Viscount Robert Lowe’s Revised Code.

Schools would receive government funding based on how many children passed tests in reading, writing and arithmetic. Inspectors would visit, examine the pupils, and the grant rose or fell according to the results.

When their salary depended on it, what did the teachers do?

They taught to the test.

Matthew Arnold – Victorian England’s most celebrated poet, who also spent 35 years as a schools inspector – watched it happen: ‘it is now found possible, by ingenious preparation, to get children through the examination in reading, writing and ciphering, without their really knowing how to read, write or cipher.’ (It would be cool if kids used to learn codebreaking, but cipher here just means maths.)

Arnold could see it coming.

He warned that tying school grants to a minimum test standard ‘must inevitably concentrate the teacher’s attention on the means of producing this minimum and not simply on the good instruction of his school. The danger to be guarded against is the mistake of treating these two… as if they were identical.’

The original idea of a good schooling would be forgotten. Passing exams would become a good education.

Five years later, the vote was extended to the working classes. Lowe announced: ‘we must educate our masters.’ Now the working classes could vote according to their beliefs, it was imperative they hold the right kind of beliefs.

The Revised Code was eventually abandoned. But the instinct behind it – measure children, fund accordingly, trust the numbers – never went away.

As psychologist Donald T. Campbell put it over a century later: ‘The more any quantitative social indicator is used for social decision-making, the more subject it will be to corruption pressures and the more apt it will be to distort and corrupt the social processes it is intended to monitor.’

In schools, no social indicator is more powerful than exams.

We’ve built a system around measuring children, and the measurement has swallowed everything else. The curriculum narrows to fit what’s tested. Children are sorted by their results. Teachers who value curiosity and creativity find there’s no room for either.

Who’s failing the ‘low attainers’?

It is not the teachers who are failing Saffa, Ben and Neymar. The teachers are keeping seven-year-olds in over break to catch up when they’d probably rather be having a cup of tea in the staffroom.

Headteachers are grouping students by ability to prepare for high-stakes testing at age 11. They believe they’re doing the best for the students, often against their own instincts to protect them.

Exams consistently favour those from wealthier backgrounds – as I explored in my previous post. In every one of the 19 countries where the PISA data allowed a comparison, wealthier students who didn’t believe in the value of hard work outperformed poorer students who saw themselves as gritty – by a wide margin.

From government ministers to classroom teachers, everyone agrees that these inbuilt biases are wrong and that we must do everything we can to correct them.

But who deals with the fallout when the kids realise the system has placed them on the bottom rung of the academic ladder?

Not the government ministers.

What if nothing works because exams are designed to discriminate?

What if the main effect of everything we do to correct the gap – grouping by prior attainment, extra work over break, intervention classes, all the mindset programmes – is to form an identity in low attainers as low attainers?

It’s such a strange system that, at some point, it stops feeling like a design fault and starts to feel like a design feature.

Someone left a comment:

A child’s success in life is overwhelmingly determined by the family they come from.

Wealthy, socially-politically connected, educated, higher IQ parents. Born into a family culture of discipline and thoughtfulness.

Advancing in young adulthood on the family wealth cushion, that grants that child the security, freedom, time, and informational access to fly above the mean survival grind, …..

An individual child’s innate characteristics, or how much they’re able to “bootstrap” themselves by self-aware conscientious effort and psychological conditioning (with or without institutional education) has some impact, in some cases.

But it is NOT the overarching life success determinant for the overwhelming majority of children.

So, is the commenter right?

Getting curious about grit and growth mindset

Recently, schools have been pushing ‘grit’ – a concept popularised by psychologist Angela Duckworth, that is usually taken to mean persistence. How hard are you willing to work at something before you give up?

Before that, schools were all about Carol Dweck’s growth mindset: success comes by believing that your academic ability is malleable rather than fixed. You can improve your intelligence by working at it.

In the OECD’s PISA questionnaire, they survey hundreds of thousands of kids from across the globe. As well as testing for achievement in mathematics, reading and science, they ask about the kids’ persistence (which I equated with grit), growth mindset and curiosity.

My question: which has the strongest association with test scores?

Crunching the numbers

Over 140,000 kids across nineteen countries were asked about all three ‘social and emotional skills’ (as the OECD call them).

In that post, I ranked the top 10% in each trait as if they were countries. We’ll come back to this later. Here, I want to show the results a bit differently. If you want the full methodological detail on how I handled the PISA data, it’s in the last post. The same caveats apply here.

Let’s take the top 10% of students in grit (or growth mindset, or curiosity). What are their chances of being in the top 10% in one of the tests – reading, maths or science – versus those in the bottom 10% for grit?

This is called an ‘odds ratio’.

There are about 140,000 kids in our sample. 14,000 of those are in the top 10% in any test result. Let’s make up some numbers to show how odds ratios work.

Imagine that out of the top 14,000 students in maths, 4,000 of these are in the top 10% of grit versus 2,000 in the bottom 10% of grit. This gives us a ratio of 2. Top-10% grit students are twice as likely as those lowest on grit to be top scorers.

The baseline is obviously 1. If grit has no effect on test score, you’ll have the same number from the top 10% in grit as the bottom 10% in grit.

Anything below 1 means that it’s better to have a lower grit score.

Anything above 1 means grit (or growth mindset or curiosity) wins out.

So, how do our gritty, growth mindset and curious students compare across the three subjects?

It’s good to be gritty. You’re between 50% and 80% more likely to top-score in maths, reading or science if you’re in the top 10% for grit than someone in the bottom 10% for grit.

Even better, though, to have a growth mindset: you’re twice as likely to ace the tests than someone with very low growth mindset.

Best of all is to be curious.

Those who are top in curiosity are between 3 and 4 times more likely to be found in the top 10% of the test scores than those who are lowest in curiosity.

Case closed. Why push grit or growth mindset when you could focus on curiosity instead?

But was I ignoring something with a far bigger impact than any of these?

Family circumstances

Wealth, personal connections, better educated parents with a higher IQ – we can’t test all of these, let alone a family culture of discipline or thoughtfulness.

But we can test some of them.

Let’s take household wealth: in PISA, students go through a list of 30-odd household items. Do they have a room of their own? A desk to study at? How many screens? And more categories of books than you’d ever thought about. PISA doesn’t ask about household income, but the HOMEPOS (home possessions) category is the closest thing we’ve got.

Next is students’ parents’ occupation. These are coded against a standard classification which maps each job to a prestige score. It’s a ranking system, where a surgeon would outscore a shop assistant.

We even have a category that gives us parents’ education, which asks how long each parent spent in education and takes the highest scorer (e.g., if mum did a master’s course, she’d get around seventeen years). The categorisation is too rough to get the top 10%, so we can’t include it in a fair comparison, but we will bring it in later.

Then we let the computer programme do the magic.

For household wealth and parental occupation, how many of those in the top 10% do we find at the top of the maths, reading and science charts versus those in the bottom 10%? This gives us our odds ratios.

That changes things.

Curiosity, which looked so towering and impressive on the last graph, has shrunk next to parental occupation and household wealth.

If one or both of your parents are in a high-prestige job, you’ve got between six and seven times more chance of being at the top of the test ranking than someone with a low-prestige job.

And the difference is even starker with household wealth.

Those at the top of the wealth charts are between ten and eighteen times more likely to be found in the top 10% of the maths, reading and science scores.

But parental occupations and wealth are linked. Those surgeons are likely to be making much more than the shop assistants. Our measurements will overlap.

So PISA rolls it all – wealth, education and occupational status – into a single measure: socioeconomic status (SES). It’s PISA’s best single measure of family background.

If the commenter was right, we’d expect the odds from having a high SES to massively outweigh those of grit, growth mindset and curiosity.

Do they?

Family Status – the overwhelming factor driving success?

Yes.

Socioeconomic status wins out. It dwarfs any effect from grit, growth mindset or curiosity, although it’s fairly similar to the effect of household wealth (which has a much larger impact on reading scores, interestingly).

Let’s imagine that those in the top 10% for socioeconomic status were their own country and let’s plot them again the other countries in our sample, as we did last time with grit, growth mindset and curiosity.

Anything that gets you above Singapore is enough to make headlines and change educational policy.

I can almost hear the education ministers clamouring: ‘what is this thing SES and how do we give our kids more of it?’

For reference, the OECD say that a score of 20 points higher on the PISA tests corresponds to about a year’s progress. That means our curious kids are a year ahead of the average kid in the sample.

Those top of the socioeconomic status lottery?

They are three years ahead of the average student (I couldn’t bring myself to plot where those in the bottom 10% would be).

I would love to say I’m the first researcher to stare at a computer and say: ‘wow, so a kid’s family background affects their test score!’ It’s such a well-known effect that unless you control for wealth (i.e. compare two kids with the same wealth), people will blame any effect you find on wealth.

Kids with more books in the home do better on tests? It’s because of their socioeconomic status.

Kids who are more curious do better on tests? It’s because of their socioeconomic status.

Kids who are born in September do better on tests? It’s because of their socioeconomic status.

In fact, it’s so well-known we’ve become almost blind to it.

Well, that’s not quite true. Many countries now spend billions of pounds a year (or whatever your local currency is) trying to reduce the effect.

In the UK, the government spends £3 billion a year on the pupil premium – extra funding for schools to close the gap between disadvantaged children and their peers.

The US has spent around $18 billion a year on Title I – its programme to close the achievement gap for disadvantaged students. A Brookings analysis found it averages around $500 per pupil, with little evidence it works. The programme currently faces proposed cuts of more than a quarter.

Ultimately, that’s why we have grit and growth mindset.

Because if we can get the poorest to try hardest, they can close that gap on their wealthy peers.

Right?

Pulling yourself up by your bootstraps?

The commenter accepted that ‘bootstrapping’ has ‘some impact, in some cases’, but was ‘NOT the overarching life success determinant for the overwhelming majority of children’.

I’m taking this to mean that if a kid works hard, they can ‘pull themselves up by their bootstraps’ and overcome disadvantage.

We can test this, too.

I’m going to use some provocative language to make a point.

I know these are real (albeit anonymous) kids and I have no idea what’s going on in their home lives.

But I want to test the claims we make about whether kids really can ‘pull themselves up by their bootstraps’, in the way the commenter described.

So let’s take two groups. The first, I’m going to call ‘rich and lazy’. These are in the top 25% of family SES and the bottom 25% for grit. (Top/bottom 10% would give us too few students, but 25% gives us about 8,000, which is a fairer comparison with our other groups.)

It’s worth noting that we can’t directly test the effort these kids are willing to put in (except, e.g. by how much of the questionnaire they bothered completing…). I’m using their self-reported perseverance – whether they agree with statements like, ‘I complete tasks even when they become more difficult than I thought.’

Then we’ve got the ‘poor and hardworking’. These are the bottom 25% for SES and the top 25% for grit.

Can the ‘poor and hardworking’ kids really outperform their richer counterparts who are low on grit?

No, in short.

Nowhere near.

Remember, 20 points is about a year of schooling, so the poorer kids can be near the top in grit and yet be several years behind their richer counterparts.

What about those odds ratios we saw earlier?

Rich, lazy students are over-represented in the top of each test score, even compared to the average student. The poor and hardworking students are an endangered breed at these rarefied altitudes.

But a sceptic might wonder if this is fair.

Maybe all the rich and lazy kids are concentrated in a couple of countries like… Spain and Australia.

(And they’d be right, to a degree. More than a third of the ‘rich and lazy’ kids – by this measure – were Spanish or Australian!)

So we can check the gap in each country separately. Does it hold even within one society and one education system?

In all 19 countries in our sample, across each subject, the privileged students who don’t try hard outscore the disadvantaged students who do.

The average gap is 53 points: roughly two and a half years of schooling.

Tipping the table in your favour

This isn’t some quirk of the PISA data. A meta-analysis of 240 studies – including over 100,000 students in total – found intelligence is the strongest single predictor of academic achievement.

In the US, the SAT correlates so well with IQ scores it can be used as a proxy. In the UK, a child’s score on a cognitive abilities test at age 11 predicts their GCSE results at 16 with striking accuracy.

And the link between social class and exam performance runs partly through the schools themselves – schools with wealthier intakes recruit and retain better teachers.

Put on a few inches and you’ll be shooting hoops in no time

The PISA data – and the other studies I’ve shared – bear the commenter out. But there are caveats. Growing up in a wealthy household doesn’t guarantee success, it just makes it (considerably) more likely.

Second, I see the concerns about ‘conscientious effort and psychological conditioning’. This is my worry with mindset interventions – they can only go so far; often, that’s not very far at all.

More importantly, we should ask why we use them as a crutch.

We claim that people just need to try harder to get the top grades when the effect of perseverance on scores is massively outweighed by factors such as IQ and socioeconomic status.

Sure, you can work hard and be successful, but when you’re from a disadvantaged background, the odds are against you.

From this view, the education system feels a bit like everyone is being asked to play in the NBA when they grow up. Every kid has to attend coaching academies; if they’re not getting enough baskets, the coaches tell them to carry on their stretches at home and keep taking their vitamin supplements to make up those inches on the tall guys. One day, if they just try hard enough, they’ll be playing in the big leagues.

But basketball is entertainment. We all accept it’s biased towards the tall. Welcome to the world of sports.

School should be different. Everyone plays, whether they want to or not, and it determines your life chances, as we’re always told.

So we have a moral obligation to ask uncomfortable questions.

Especially the fundamental one: is the system fair?

If we don’t think it is (and we don’t, because we spend billions each year looking for ways to sort out the fact that poorer kids are at a huge disadvantage), are grit and growth mindset interventions enough to fix it?

At some point, the questions have to get bigger.

Nobody chooses their IQ. You can’t decide to be born into a wealthier family. Even how hard you’re willing to try is affected by your genes and environment – neither of which you choose.

So what is the point of school?

Partly, as economist and education scholar Alison Wolf argues, it’s a signalling system – a way of ranking kids to work out who gets the best jobs. But the measure has become the purpose.

Which might explain why the interventions we reach for – grit programmes, growth mindset workshops – are always about helping individual kids run faster in a race that’s already been fixed.

Rather than asking whether the race is worth running, we just tell the slower runners to try harder.

It’s time to get curious about what school could look like if exam results weren’t the whole point of it.

Psychologist Susan Engel and her student Kellie Randall advertised for some teachers to run a science experiment. Each teacher was brought in separately and paired with a child.

The teachers were split into two groups, both with the same task – guiding the children as they mixed baking soda, vinegar and water, then dropped raisins in.

But while half the teachers were asked to help the child learn about objects floating and sinking, the other half were simply asked to help the child complete the task and worksheet.

Teachers were given only eleven minutes to work with the child – only just enough time to complete the worksheet – so that the teachers (like teachers everywhere) were under time pressure.

Those told to help the children learn made encouraging comments, asked their own questions and gave positive feedback. Those who were helping the children complete the worksheet made ‘restrictive’ comments, according to the researchers, criticising and correcting what the children were doing: ‘The procedure says to add 8 raisins. Make sure you are adding exactly 8; count them out first.’ And, ‘No, no, wait, let’s read through this first.’

And that was even before the kid went off script.

On the bench were some things that weren’t on the worksheet – sweets, crackers, marshmallows. Halfway through the experiment, the child picked up a Skittle from this pile and dropped it into the liquid.

If the teacher asked, the kid was to say they just wanted to see what would happen.

The teachers who had been told to help the child learn were mostly encouraging. They asked follow-up questions: ‘Wow, what’s happening with the Skittle?’ and ‘Do you think the water is going to turn orange now?’

The other half, who wanted to get through the worksheet, were frustrated. ‘You were not supposed to do that,’ one said. ‘Show me where it said to put that in there. It didn’t.’

Another said, ‘Well, curiosity is a very good thing, but probably we should stick to the directions.’

The most depressing part?

After the trial, the teachers who had been told to help the child complete the worksheet thought they’d done just as good a job of encouraging the children’s curiosity as the other group.

When curiosity researchers leave their curiosity at home

That’s one experiment. Conducted in a lab rather than classroom.

But what if this sort of thing does go on in classrooms? And what if it happens lesson after lesson, year after year?

What effect would that have on children’s curiosity?

As well as a favourite experiment, I also have a favourite book about curiosity: The Hungry Mind.

It’s also by Susan Engel.

In fact, it inspired me to start a PhD researching children’s curiosity.

But one line in it bothered me:

‘When older children are asked about their enthusiasm for domains like science, many indicate a great interest in the topic itself, but say they have little interest for the way it is presented in school.’

This is a big statement, and she makes similar claims elsewhere.

It’s certainly plausible. As a secondary science teacher, I can confirm a lot of students have little interest in the way it’s presented in school. Or at least, the way I present it (and I try pretty hard some days to make it as interesting as I can).

Other academic researchers have picked up on what Engel says.

‘Despite its association with positive outcomes, one of the most common manifestations of curiosity – children’s questioning – typically decreases once children enter formal schooling.’

Yes, there’s some evidence for this, but one of the main studies cited was conducted with preschoolers at home and at kindergarten. Nearly all such studies (and there aren’t many that look at how questioning changes with age) are small-scale.

Another, more recent paper cites Engel: ‘typical schooling does not reliably increase curiosity, and might even diminish it in some contexts, i.e, when learning information by rote memorization is emphasized.’

Sure, maybe. But Engel gives no evidence to support an effect of rote memorisation on curiosity. And other authors follow suit (e.g., here and here).

What’s missing is curiosity about the original data.

What does the evidence really say?

If only we had a big, global sample of children to ask

Engel’s ideas have caught on because they’re appealing. Also, in a field dominated by questionnaires and quantitative analysis, she actually watches children and teachers to see what they do.

And she does this in the classroom.

She nails the mechanisms by which teachers might halt curiosity. The sort of thing that, repeated day in, day out, might lead kids to think that curiosity has no place in schools.

In one case study, Engel’s watching Mrs Parker, who has given ten-year-olds some equipment to model how the ancient Egyptians might have built the pyramids.

There’s a sheet to fill in. When one group comes up with some particularly creative ideas, Mrs Parker tells them, in front of the rest of the class, ‘OK, kids. Enough of that. I’ll give you time to experiment at recess. This is time for science.’

In another class of fifteen-year-olds, Engel reports the teacher halting a class discussion: ‘I can’t answer questions right now. Now it’s time for learning.’

These are great vignettes. And I’m all in favour of small-scale research involving classroom observations – it’s the basis of my PhD work. I see many of the same things Engel saw, although I also see teachers patiently answering question after question (ones which are often very tenuously related to the lesson).

But to make universal statements about what happens to children’s curiosity at school, you need more.

You’d need to include tens of thousands of children, from across the world – Bogota in Colombia to Helsinki in Finland.

Imagine you asked all those children how curious they were. Then you’d have something solid to work with.

Then you could begin to make bigger statements about what happens to children’s curiosity as they get older.

Luckily, someone has done exactly this.

Thanks OECD – tell us what you’ve got

If you want big educational surveys, you go to the OECD.

Their PISA study of 600,000 15-year-olds every few years makes headlines and forces educational policy changes.

(If you want to know what PISA tells you about curiosity, head here.)

But they run another study which is, in its own way, even more fascinating – their Survey on Social and Emotional Skills, or SSES.

In 2019, they gave their questionnaire to over 60,000 kids in ten cities across the world. This time, they weren’t creating a league table for academic achievement, but finding out which ‘soft skills’ help kids succeed. They asked about creativity, self-control, cooperation and – crucially – curiosity.

And, more importantly, they include 10-year-olds and 15-year-olds in their study. The kids were asked questions and their answers to these questions were weighted and made into a ‘curiosity score’.

All we need to do is compare the curiosity scores of the two age groups.

This should tell us everything we need to know. If the fifteen-year-olds’ curiosity is, on average, lower than that of the ten-year-olds, then Engel’s right.

So, here we go:

(Technical detail: even if you give this the full stats work-over, you find the same thing. I’ve run the model with more levels than a multi-storey carpark and in every city, the 15-year-olds have lower curiosity than the 10-year-olds).

So, job done.

Schools kill curiosity.

We need to stop all those Mrs Parkers in their tracks and tell them: Experimentation is science! Questions are learning!

But not so fast.

What were those questions asking? What exactly was it the children told the OECD about their curiosity?

Do you like learning new things? Or do you love learning new things in school?

Here are the individual questions. All questions were on a 5-point scale — strongly disagree / disagree / neither agree nor disagree / agree / strongly agree.

To what extent….

1. Are you curious about many different things

2. Are you eager to learn

3. Do you like to ask questions

4. Do you like to know how things work

5. Do you like learning new things

6. Don’t you like learning

7. Do you love learning new things in school

8. Do you find science interesting

If we split the results by question, a more interesting story emerges.

Let’s start with the first one: If kids are curious, surely they’ll agree or strongly agree with question number one.

How do 10-year-olds and 15-year-olds vary in their responses?

Wait.

What was that?

There’s not much in it, but more 15-year-olds are curious about many different things than 10-year-olds. So, curiosity, by that measure, actually increases (slightly) as kids get older.

What about question three – the percentage who like asking questions?

This is where things get interesting. Although students feel at least as curious on average at the age of fifteen, they don’t like asking questions. Not as much as the 10-year-olds, anyway.

(Note that we aren’t looking at what happens to the same children over time – we’re looking at two different groups of children. And this would make a difference if something like the pandemic had happened, and might affect different age groups differently. This is an issue with later SSES data but not this 2019 cohort.)

So, curiosity stays the same (or increases) but enjoyment of asking questions decreases.

But maybe this has nothing to do with school. Maybe this is developmental.

It could be that 15-year-olds don’t need to ask as many questions, because they have other ways of finding out (although even on Google and ChatGPT, you need to ask questions to resolve your curiosity).

There is, however, a way of testing this.

Look at questions 5 and 7. They ask whether students…

(Question 5) like learning new things.

(Question 7) love learning new things in school.

There are two differences between these questions.

If students’ answers differ, it might simply be because they like learning, although they don’t love it.

But there’s another difference, one that jumps straight out: question 7 asks specifically whether students love learning new things in school.

Most students – aged 10 and 15 - agree or strongly agree with both questions. But looking at the number who don’t agree or strongly agree makes the changes a bit clearer:

The number who don’t agree that they love learning new things in school is much larger than the number who don’t agree they like learning new things.

Even at ten years old, it’s fifty percent larger. But by fifteen, that gap has doubled. Now, twice as many 15-year-olds don’t agree that they love learning new things in school compared to the number who don’t like learning new things.

Houston, we have a problem

We can even look at this ‘curiosity drop’ city by city.

I averaged the scores for every student in each city at age 10 and age 15.

On the y (vertical) axis, I’ve plotted their agreement that they love learning new things in school.

The x (horizontal) axis tells us whether they like asking questions.

We can then see the changes between 10- and 15-year-olds in each city.

Of course, we have to be cautious. We only have twenty data points. But an interesting pattern begins to emerge.

Those arrows are shooting back towards zero (although note that the axes only start at about 3.5).

In our diverse sample of cities from across the world, something changes for the average student between age ten and age fifteen. They don’t like asking questions as much, as their love of learning new things in school is falling.

Why, though? What’s causing this?

So who’s responsible? Lining up the suspects…

This is where we have to be cautious. It’s easy to fall into the trap I gave at the start of this post by pointing the finger at the most obvious candidate.

Researchers have been quick to blame the Mrs Parkers of this world, squashing questions and pulling Skittles out of chemical concoctions.

But let’s look at this from Mrs Parker’s perspective.

Her mind could have been on an upcoming inspection, or the headteacher was going to check through her class’s books later that day: ‘oh look, this kid (who’s in the disadvantaged group) didn’t manage to complete his worksheet.’

The teachers in Engel and Randall’s study were randomly assigned to a group. They hadn’t decided to focus on completing the worksheet – they’d been told to do that.

We know from survey evidence that teachers value children’s curiosity – something I’ve heard again and again in interviews for my doctoral research. But teachers work within constraints: a scheme of work to get through, exams to prepare for, and – crucially – the need to ‘prove’ that learning has taken place.

If you’re a teacher with limited time to cover a lot of material, a simple action like a student dropping a Skittle into liquid, or asking an off-topic question, could become a subversive act. You can embrace it, but that sends out a message to all the kids that this kind of behaviour is ok. At best, it puts you further behind in the endless race against the scheme of work. At worst, your classroom authority is challenged and it’s even more difficult to get through everything you planned to.

Or you can put a stop to the behaviour and plod through the lesson’s content. And at the end, you’ll have the worksheets to show for it.

The curious teacher has nothing.

Where is that curiosity going?

So is it school that kills curiosity? Is it developmental – something that happens when kids hit their teens? Or is there something the OECD’s questions are failing to capture?

I don’t think the data can tell us.

But it can tell us something important.

Look again at the difference between those two questions. Only sixteen percent of 10-year-olds don’t agree they like learning new things. The same proportion of 15-year-olds.

Curiosity, in the broadest sense, survives adolescence intact.

But ask whether they love learning new things in school, and the picture changes.

At age ten, there’s a gap between students’ answers to these questions.

By fifteen, that gap has doubled.

The curiosity is still there. It just doesn’t feel as welcome in school.

A final question: what would schools look like if students felt their curiosity was welcome?

Enjoyed this? Subscribe to Get Curious for more content to pique your curiosity.

Subscribe now

· Joyous Exploration – delight in learning for its own sake. This is curiosity as wonder: finding something fascinating and wanting to understand it better, not because it’s useful, but because ideas are genuinely exciting.

· Deprivation Sensitivity – a restless need to know. People high on this dimension work relentlessly at problems, often losing sleep over them, not because they enjoy the struggle, but because they simply can’t rest without a resolution.

· Stress Tolerance – comfort with uncertainty and the ability to manage the anxiety that comes with it. High scorers don’t necessarily seek out the unknown – they just don’t let the discomfort of not-knowing stop them.

· Social Curiosity – interest in other people. And here’s where it gets interesting: the researchers found two distinct flavours. Overt social curiosity is the art of asking questions – turning towards people, probing, engaging. Covert social curiosity is quieter: watching, listening, picking up on things others miss.

· Thrill Seeking – the appetite for adventure. This is curiosity with an adrenaline edge: a preference for the novel, the risky, the spontaneous. People high here don’t just want to know – they want to feel.

Wondering which you might be higher or lower on? Well, I’ve got a way you can find out, but more on that in a minute.

Empathizer or avoider?

The researchers discovered something else, too. They found patterns in the results. People taking the quiz could be grouped into four distinct curiosity ‘types’:

The Fascinated score high across the board. They’re lit up by the world – ideas, people, places. Learning energises them; uncertainty feels like an invitation. They’re assembling a richer picture of how everything connects.

The Problem-solver is intensely, narrowly curious. When something grabs them, they won’t let go. They’re less interested in everything than in this thing, right now, fully understood. Persistence is their superpower.

The Empathizer is high on social curiosity, often lower on thrill-seeking. People are their laboratory. They explore life through connection, and the question they ask most often – consciously or not – is why do people do what they do?

The Avoider is the most interesting type, and probably the most common. They score lower across the board, particularly on stress tolerance. But this isn’t someone without curiosity. Their curiosity is quieter, more cautious – often unexpressed, because uncertainty feels uncomfortable. The pull towards the unfamiliar is there. It’s just competing with something stronger.

Go on – it might change your life*

Want to know which category you fall into? Glad you asked! Because I’ve put together a quiz to help discern the problems solvers from the avoiders, those high on joyous exploration or lacking in stress tolerance.

At this point, I should give a disclaimer and say something like ‘the test is just for fun’ or ‘don’t take this as a diagnosis, quit your job and become a yoga teacher in Kuala Lumpur or Scunthorpe or wherever.’

But it has just changed my life.

Ok, maybe that’s a slight exaggeration. It made my Saturday ever so slightly easier.

I was finishing off the code for the quiz and Daughter Number 1 (who identifies as Blossom for the purposes of this post) asked what I was doing.

We were supposed to be doing errands, then going to the bookshop and getting lunch, but she couldn’t be bothered and wanted to play at home.

Also, she thought the takeaway curry might have onions in it.

As a distraction, I said she could take the test.

Each time I read an academic paper and come across a new curiosity measure, I test it out on her (it’s easier than recruiting actual research guinea pigs; the KCL ethical process is a nightmare). She loves being asked questions. I even ran a trial for my doctoral research where I asked her to examine a range of houseplants and come up with questions about them.

She scored high across most dimensions on the quiz, coming out top on stress tolerance (not always evident at bedtime) and getting the ‘fascinated’ profile. She was delighted about getting the curiosity rubber stamp.

‘Imagine what we might discover in the bookshop,’ I told her. ‘You can try some new food at the curry place.’ She was all for it – her fear of onions forgotten.

Science in action: no guarantees you’ll get the profile you’re hoping for…

So, go on and try the quiz. It’s just for a bit of fun.

But there’s a tiny chance it might just change your life – or at least make your day a bit easier.

Terms and Conditions:

*No refunds available if taking the quiz does not turn out to be a life-changing experience.

More on curiosity

If you’ve ever wondered what happens to kids’ curiosity at school - Do Schools Kill Curiosity? It’s more complicated than you think.

Or, on what actually predicts exam success - Over 140,000 kids were tested to see which is most important for success: grit or growth mindset. The Answer? Neither of them.

In Victorian England, women did not become adventurers. They certainly didn’t become solo travellers, not if they were from respectable families.

Marianne North didn’t care for convention. She didn’t care for her own safety. She didn’t care what other people thought.

What she did care about was exotic plants.

A vice takes hold

Originally, Marianne North wanted to be a singer. She had a problem: ‘a most provoking habit of nervousness; when told to stand up and show off, the room seemed to go round, and I could not keep myself from shaking all over’.

So she developed another passion. Two, in fact: plants and oil painting. To North, painting was a vice like whisky, ‘almost impossible to leave off once it gets possession’, as Milbry Polk quotes in Robin Hanbury-Tenison’s The Great Explorers.

Marianne North was born into privilege. She didn’t have much formal education but was surrounded by inspiring figures during her childhood. These included Sir William Hooker, pioneering botanist and director of Kew Gardens, who sparked North’s love of plants.

After her mother died, North promised to look after her father. Which she did her way – on a series of road trips they took together to Europe and the Holy Land. North’s sketchbook was never far from her hand.

On one trip, her father fell ill. At home in Hastings, days after North turned thirty-nine, her father died.

North was devastated.

It was unthinkable, at that time, for a woman to travel without a chaperone. But North disdained convention. She called marriage ‘a terrible experiment.’ She hated the idea of simply being someone’s wife.

She wanted freedom.

The pleasure of recognition

It has become easy, in the twenty-first century, to travel to faraway places in comfort.

A new city, a different continent, another culture. Experiences that should give us a new perspective on life simply become another box, ticked.

As journalist and literary critic António Guerreiro puts it, we travel for ‘the pleasure of recognition and not the curiosity of discovery’.

This is a polite way of saying we travel to take photos.

Or, at least, to inhabit those photos that haunt our dreams. The photos we used to gaze at in the glossy travel agency magazines and that now comprise Instagram.

There is no unknown, no ambiguity. The only risk is that the hotel might not match up to the Tripadvisor description.

What kind of curious are you?

Risk taking excites me

I prefer spontaneous adventure to planning

When I have free time, I like to do things that are a little scary

Agree? You might score highly on the thrill-seeking dimension of curiosity.

For Todd Kashdan and his colleagues, curiosity isn’t a single trait. It can be split along five dimensions:

• Joyous Exploration — delight in learning for its own sake.

• Deprivation Sensitivity — a restless need to close knowledge gaps.

• Stress Tolerance — comfort with uncertainty.

• Social Curiosity — interest in other people (overt asking vs covert observing).

• Thrill Seeking — a desire to explore through novelty or risk-taking.

People who score higher in stress tolerance, for example, are more extraverted. Joyous explorers are more open to new experiences. The socially curious – unsurprisingly – are more likely to gossip. (Want to know where you rate? Try this quiz I put together – just for fun – and get more info about the five dimensions).

It’s easy to feel that adventure should be for the thrill seekers, or those who are fully stress tolerant. But that’s an excuse we use – it’s too big. I don’t know where to start. I haven’t got the time, or the experience, or the money to do that.

We can turn those excuses on their head. Rather than thinking of adventure as something that exists ‘out there’, we can build adventure into our daily lives.

Embrace the inconvenience

We don’t have to start with the epic travel, writes explorer Belinda Kirk in Adventure Mind. A change to routine, a new environment, the chance to rely on our wits over modern technology – small changes can be our starting point.

And adventure brings a whole host of benefits, Kirk argues. As well as improving our mental health, it builds new skills and helps us use our brain in a way that the modern office doesn’t allow us to.

Adventure might involve following a map rather than a SatNav, cooking our own food on a burner stove in the great outdoors (and probably under a drizzle) rather than mustering the courage to risk the street food stall.

This is adventure as journey rather than destination.

Insta-travel is about where you end up. It’s about what it looks like from the outside.

In adventure travel, the inconvenience becomes the story: getting lost, being cold, wet, tired. The kind of things we spend our lives avoiding become part of the story.

Challenging yourself by getting outdoors can be transformational, Kirk writes in Adventure Mind.

You’re in a new environment, which focuses your mind on your immediate surroundings, not on the email you still haven’t answered.

Paradoxically, it is novelty of the situation, and the cognitive demands it places on you, that frees your mind:

Before you leave on the adventure, you have a million and one things to do, but as soon as you set foot on the plane, sit on the seat of your bike or turn away from your car with a full backpack, life immediately becomes much simpler. And nagging feelings dissipate and are replaced by something much more exciting: the challenge, the unknown.

One fingerhold from death

The stages of Joseph Campbell’s Hero’s Journey are so well known that they have become a cliché – and it’s easy to apply them to our lives. We wait for the call to adventure – for someone to tell us it’s time. Someone to tap us on the shoulder and hand us the itinerary and map with the route already marked out.

It’s easy, too, to think of adventure as for someone else. A different kind of person.

We look at people who do extraordinary things and assume they must all be like Alex Honnold.

Honnold climbed one of the tallest rock faces in the world without ropes or harness. The friction between his chalked fingertips and the tiniest protrusion of rock was all that separated him from death.

A few years later, he climbed one of the world’s tallest skyscrapers, again without safety equipment.

There is a scene in the 2018 film Free Solo in which Honnold is put into an MRI machine. He is shown images that would terrify a normal person. His amygdala barely lights up. He needs these terrifying experiences to feel alive, in the same way someone might feel a thrill while giving a PowerPoint presentation at work which they’ve only practised a couple of times.

To most people, adventure is both more mundane and more extraordinary than it is for Honnold. Mundane, because few people are going to make an Oscar-winning film about it. Extraordinary because it requires us to overcome what our amygdala is telling us. (Run!)

It’s more like Bilbo Baggins in The Hobbit, who lived a quiet life in The Shire, and was respected by his neighbours because he never had any adventures or did anything unexpected. Until one day he ran off after a group of dwarfs in such a hurry he forgot his hat and walking stick.

And it’s more like Marianne North, who lived a proper life in polite society until, at the age of forty, she began a series of solo adventures that would last for the rest of her life.

Snakes, fever and enormous boulders

Marianne North had a very particular idea about what she wanted to escape to.

She was driven by ‘the dream of going to some tropical country to paint its peculiar vegetation on the spot in natural abundant luxuriance.’

Nowadays, heading off to the tropics and posting snaps of some of the weird flowers on a group chat doesn’t sound that revolutionary. What Marianne North did, though, challenged the way people saw plants.

Traditional Victorian botanical art was often a lifeless affair. A sprig was laid out and painted with watercolours.

North had her own ideas about how to paint flowers.

Marianne North, ‘Foliage, Flowers and Seed Vessel of the Opium Poppy’ (1870s). Marianne North Gallery, Royal Botanic Gardens, Kew. Public domain via Wikimedia Commons.

Her oils brought life and colour back into the plants. She used a limited palette, and avoided black, giving a vibrancy to her paintings. Plants appeared in their habitats, creating a naturalism that had been dispensed with.

She began her travels in Egypt, but was encouraged by Lucie Duff Gordon, another female explorer, to travel more widely.

Like Bilbo, facing his trolls and goblins, North’s travels were not without their dangers.

In the Seychelles, she was perched so high on a boulder that ‘the slightest slip or cramp would have put an end to both the sketch and to me.’

In Sri Lanka, she thought someone had left a plant specimen on her stool. She reached out her hand to pick it up – realising just in time that it was a poisonous snake.

Her sister later wrote that North led a ‘charmed life. She could apparently sit all day painting in a mangrove swamp, and not catch fever.’ She could survive without food and sleep and return home ready to enjoy a ‘flattering reception’ in London.

By challenging what people thought proper for a respectable woman, she left her mark on botanical history.

Don’t forget your hat and walking stick

In retrospect, it seems as if success came easily to Marianne North.

She was friends with Charles Darwin, who encouraged her to visit Australia. He thanked her afterwards for the specimens she sent back to him.

In Borneo, North painted an enormous pitcher plant. This carnivorous plant was unknown to science and later named in her honour. Several other species first depicted by North have also been named after her.

Along with James Fergusson, North designed the gallery that bears her name at Kew Gardens. Over 800 of her paintings still cover the walls.

As North’s biographer Milbry Polk puts it:

She was one of the truly fortunate people who discover what they love to do, have the means and courage to follow their passion, and the gift to share their discoveries.

North had the money to travel – and to refuse the marriage that may have tied her down. Her father had encouraged her hobbies, and she had grown up around the greatest scientists of the day. Had she been born into different circumstances, this path may have been unthinkable.

Today, many of us tell ourselves similar stories.

I don’t have enough money

I don’t have the right connections

I need people behind me

But often what we mean is, I’m scared.

We can’t all be Alex Honnold, dulled to the fear that grips everyone else when contemplating something out of the ordinary.

But maybe we can be a bit more like Bilbo Baggins – worried about what people might think, but taking that first step on the road to adventure anyway. Even if we do forget our hat and walking stick as we rush through the door.

Debunking the myths

Google ‘education myths debunked’ and you’ll soon come across something about learning styles.

Each child had a preferred way of learning – visual, auditory or kinaesthetic – the idea went, and teachers had to accommodate this.

This was based on evidence.

Until it wasn’t.

This is how teaching works. Strategies are supposedly evidence based. But it’s funny how often the evidence is called into doubt just as the next fashionable theory arrives.

Whose test paper will you take a peek at?

A few years ago, ‘growth mindset’ was in fashion.

The original concept came from Stanford University professor Carol Dweck and colleagues.^[1] If you have a growth mindset, you believe your intelligence and abilities are malleable. It isn’t just what you’re born with, it’s what you do to improve them that really matters.

The classic test of growth mindset went a little like this – a kid takes a test. After, they’re given the opportunity to someone else’s test. Do they choose the test of the student who did better, to learn from them, or the one who did worse, as an ego boost? First choice = growth mindset, second choice = fixed mindset.

Each school had posters up about how to foster it. In 2015, the education secretary Nicky Morgan wanted to ‘nurture a generation’ through promoting growth mindset.

The academic work on growth mindset was complex, but schools often stripped this nuance away. Inevitably, when the classroom interventions didn’t match the effects seen in the original studies, it was this translation that was blamed.

In 2019, a randomised controlled trial involving over 100 schools and over 5000 students found that students exposed to training in how to build a growth mindset made …. no extra progress compared to those without the training.

As with all fashions, growth mindset eventually faded from view. The posters became tattered and the school assemblies moved onto new topics.

Growth mindset never completely disappeared. It’s still mentioned from time to time. The fossilised remains are still there, on outdated teaching and learning pages on forgotten corners of school websites.

But there’s a new kid on the block.

The secret to West Point success – and everything else

By 2019, the year growth mindset’s lustre faded in England, the UK government was already talking about the importance of developing ‘character traits’ like conscientiousness and perseverance.

By 2025, this had become a central focus for the Department for Education. Developing ‘much-needed grit’, as the Education and Health Secretaries Bridget Phillipson and Wes Streeting put it, would help young people in all areas of life.

Grit, according to Phillipson, is ‘the resilience, the ability to cope with life’s ups and downs, about the challenges that are thrown at you’. In fact, it’s ‘essential for academic success’.

The concept – grit – was popularised by psychologist Angela Duckworth in her book of the same name.^[2] She led much of the pioneering research in the area. And grit seemed to explain things no other concept could.

Take the elite West Point military training academy in the US. Out of 14,000 high school applicants – who are nearly all star athletes as well as top academic performers – only 1,200 are admitted. Of those, twenty percent drop out before graduation, including a substantial number in the first weeks.

Fitness didn’t predict who would leave; nor did academic achievement. Then Duckworth stumbled on the idea of grit. Suddenly, just by giving cadets a few questions, she could predict who would make it to graduation and who would leave a few weeks in.

And grit didn’t just explain who would graduate from West Point. It seemed to explain success in areas of life as diverse as who’d be crowned spelling bee champion to who would top a company’s sales figures.

But what exactly is grit?

Grit, for Duckworth, has two aspects – perseverance and consistency of interests. Perseverance obviously encapsulates how hard you are willing to work. Consistency of interests looks at whether you flit between hobbies and activities or stay focused year after year.

But interest is often forgotten, especially in schools. It doesn’t matter whether you’re interested in English or maths, you still need to work hard towards your exams.

Duckworth’s own work doesn’t help with this simplification. She sums up grit in two equations:

talent x effort = skill

skill x effort = achievement

Effort, then, is the most important aspect of achievement. Many schools can be forgiven for reducing grit to the idea that you need to work hard (even when you don’t feel like it) to succeed.

The question is, are they right?

Does persevering really make a difference?

So, which is best?

In an ideal world, we’d take a hundred thousand kids from across the world and give them a questionnaire. We’d ask them about grit and growth mindset.

We could then, for example, take the top 10% in each group – grit and growth mindset – and compare their scores in maths, reading and science.

Who did best, those top in growth mindset or grit?

Luckily, we don’t need to – someone’s already done it for us. The OECD’s PISA test is taken by hundreds of thousands of fifteen-year-olds around the world every few years. As well as assessing which country’s students do best in mathematics, reading and science, they look at ‘soft skills’ including growth mindset and perseverance.

Perseverance? What happened to grit?

First, as the Education Secretary suggests, while grit is a complex concept, most schools are only interested in the perseverance part. It’s about trying hard when things get tough.

Also, PISA’s perseverance construct isn’t that far away from Duckworth’s original grit scale.

Take these two:

Although I should probably just say ‘perseverance’, ‘grit’ is a bit catchier. I’ll use the two interchangeably from now on, but bear in mind that Duckworth’s scale includes consistency of interests too, e.g. ‘My interests change from year to year’ (reverse scored, so ticking ‘agree’ gets you a low score).

The growth mindset scale is a cut-down version of Dweck’s original questions. The key question is this one: ‘Agree/disagree: Your intelligence is something about you that you cannot change very much.’

So, what happens when we put growth mindset and grit (perseverance) head-to-head? Which has the strongest effect on test scores?

But how WEIRD is our sample?

The analysis below is quick and simple. It’s an approximation. A thorough analysis would take many more stages and deal with a few issues a bit more rigorously than I have.

One of the main problems is that kids taking the PISA tests don’t answer all the questions, only a selection of them. If we look at just the countries where at least 90% of kids answered questions about growth mindset and perseverance, we’re left with only 19 countries out of the 80 that took part.

It doesn’t sound that much, but the sample still includes 145,000 students.

The main worry is whether these countries are really that representative, globally speaking:

In a 2010 paper, psychologist Joseph Henrich and colleagues warned that a lot of published research findings were skewed towards Western, Educated, Industrialized, Rich and Democratic (WEIRD) nations.

Did these findings hold in non-WEIRD nations? There wasn’t enough evidence to say. And the sample of 19 countries above isn’t immune from this criticism: most obviously, there are no African or South American nations included at all.

But the sample is more diverse than it might look at first glance: alongside Western European nations, the sample includes Türkiye, Mexico and several high-performing East Asian systems whose educational cultures differ substantially from Western norms.

Then there’s a second simplification (and complication).

PISA doesn’t give each student a single score for mathematics, reading and science; instead, it estimates a range of possible scores. A thorough analysis would take all of them into account.

I’m just going to use what’s called the first ‘plausible value’. So, bear in mind it’s an estimate, but it will get us close enough. Even published, peer-reviewed papers like this Nature article do this sometimes.^[3]

Before we release the results, a word about what the scores actually mean.

Can you really make airplanes out of dogs?

Six birds flew over the trees

The window sang loudly

Airplanes are made of dogs

Do these sentences make sense?

If you can answer correctly, you’re on the way to a decent score on the reading fluency section.

There are trickier and more interesting comprehension questions – on a made-up blog about Rapa Nui, for example, and discovering whether you can give aspirin to chickens via a chat forum.

The OECD keeps a tight hold of most of the questions, though, which allows them to compare scores. If the UK scores 500 points in science this year, is this really an improvement over the 490 last time or is this grade inflation?

If everyone answers the same questions every four years, it’s easy to compare.

What’s a ‘good’ PISA score? In some ways, it depends how your country did last time. Beat the previous score and your nation’s Education Secretary will be on the phone to the papers to organise a press conference.

It also depends on how your rivals do. And, of course, the pace setters, like Singapore, Korea, Japan and their neighbours like Taipei (Taiwan), Macao (China) and Hong Kong. These top countries might be pushing 550.

The ‘average’ used to be 500. It’s slipped a bit since then to around the 470s. And this is obviously where you’ll find the mid-ranking countries like Israel, or Malta or the Slovak Republic. Countries nearer the bottom, like Kosovo or Cambodia, fall below 380.

To give an idea of the spread of results, here are the reading scores for our 19 countries:

Enough preamble.

How do the gritty students compare to those with growth mindset?

And the winner is…

Let’s take the students who score in the top 10% for perseverance and compare them with those who score in the top 10% for growth mindset (obviously the same students might appear in both samples).

Then we see who gets the highest scores on mathematics, reading and science.

Time to cut to the results:

So, we have a clear winner.

It’s good to be gritty: students in the top 10% for perseverance outperform the average student in our sample by over 10 points across all three subjects.

That’s not bad going. According to PISA, 20 points equates to about a year’s worth of schooling.

Growth mindset does even better, though, adding between 14 and 17 points on top of the baseline across all subjects.

A thought experiment: Let’s imagine that those in the top 10% of each trait –grit and growth mindset – were countries. Where would they fall in our list of 19?

They’re both doing well, but growth mindset edges it.

So, we’ve solved the conundrum. Who has the Education Secretary’s number? Someone WhatsApp her the graph and let’s dust down the growth mindset posters.

Not so fast!

There’s another concept we haven’t talked about yet (in this post, at least): curiosity.

The PISA test asks students how curious they are, too.

What if we took kids in the top 10% for curiosity? How would they compare?

Getting curious about which trait matters most

First, a word about curiosity.

Neuroscientists define it as the intrinsic motivation to learn.^[4] It’s about wanting to do well rather than simply acing a test or doing it to get rewards (or avoid punishments).

Who’s curious, according to PISA? People who strongly agree with the following:

· I am curious about many different things.

· I like to ask questions.

· I like learning new things.

And disagree with these:

· I get frustrated when I have to learn the details of a topic.

· I find learning new things to be boring.

As a gritty, persevering, growth-minded curiosity scholar, I have tried to trace the origins of these questions (the dark winter evenings just fly by in our household). Their genealogy, as Nietzsche might have put it, isn’t always that clear.

Still, people who agree ‘I am curious about many different things’ are probably more curious than people who don’t agree with it.^[5]

So, enough preamble. How do those in the top 10% of the curiosity ratings compare to the gritty and growth mindset students?

The big reveal

As before, we take the top 10% of students in grit, growth mindset and curiosity. We look at the average scores across the three tests. Then we put them in a table and…

Voilà.

And the same thing in the form of a helpful graph (note the dotted black representing the average for our 19 countries across all three tests and the fact we’ve started the y-axis at 490 points):

Again, we have a clear winner.

Not only that; the differences are sizeable. A few points in PISA can make a big difference. The seven points that separate curiosity from growth mindset – its nearest rival – in mathematics are meaningful.

Say you started off with Latvia on 483 points in 21^st place. An extra 7 points would allow you to leapfrog Finland, Slovenia, Czech Republic, Australia, Austria, Poland, the UK, Denmark, and Belgium into 12^th position.

In science, the difference is starker: if you could switch from being in the top 10% of grit to the top 10% of curiosity, you’d make almost a year of progress in an instant (ignoring all the caveats mentioned above about missing data and using only one of the ‘plausible value’ scores etc).

Insert curiosity into our ‘what if traits were countries’ table and the gap is clear:

Time for another thought experiment.

Let’s imagine that some schools went all-in on the educational interventions and promoted grit and growth mindset at the same time.

If we had a group of students who excelled in both areas, surely they would beat our curious students.

Right?

Flipping the table

Wrong.

Curiosity is no ordinary trait: it’s a learning superpower.

If we add the perseverance and growth mindset scores together and take the top 10% of students in both traits… these students still don’t reach the scores of the most curious students.

In fact, in reading, these gritty, growth mindset students are still the equivalent of three months behind their curious peers.

Let’s do a crazy thought experiment.

Let’s add together the perseverance, growth mindset and curiosity scores for all students. Take the top 10% again and compare them to the ones who are in the top 10% simply ranking by curiosity.

The difference?

This time, the triple-combo takes the crown, but there isn’t much in it. The difference is smaller than the gap between curiosity and either of its competitors.

Don’t worry about grit and growth mindset; you’d be as well simply concentrating on getting into the curiosity elite.

Time to put 145,000 students into a line. And then another line…

The top-10% comparison is the most intuitive way to look at the data, but there are two more rigorous tests worth mentioning — and they tell the same story.

The first is a correlation test.^[6]

Let’s take 10 students and line them up by curiosity scores, then test scores.

If they’re in the same order both times, their correlation scores would be 1 – a perfect correlation.

Let’s imagine they’re in completely the opposite order. The student top first would be bottom in the second case, and so on. This time, the correlation would be -1 – a perfect negative correlation.

No relation between the two? Correlation is 0 – no relationship.

For context, the correlation between height and weight in adults is around 0.5: most people in a sample who are heavier are also taller.

Here’s how our three traits score:

First thing to note – all correlations are positive. Higher traits = higher test scores.

Second thing – all the scores are pretty low.

As a rough guide, correlations below 0.2 are often considered small in social science research, but in the case of the PISA test data, this is to be expected. Test scores are complicated, affected by hundreds of factors – teaching quality, home environment, sleep, what someone posted about you on social media the morning of the test.

A single psychological trait isn’t going to explain everything.

But within that context, across all subjects curiosity’s correlations are consistently larger than the alternatives – up to double, in some cases.

This finding holds across our sample.

Curiosity has a stronger correlation with all three test scores than growth mindset in all 19 countries and beats perseverance in 17 of 19.

Wherever you look, the pattern holds: curiosity correlates most strongly with test scores.

Note that this doesn’t mean curiosity causes higher test results, merely that the two go together. A rigorous analysis would look at curiosity’s effects over the longer term. We can’t rule out the effect that test success causes higher curiosity (or, of course, that both are caused by another factor).

But what if curiosity is piggybacking on the other two?

Correlation testing tells us that when one thing goes up, so does the other.

But we can do a bit better than this.

A regression model lets us ask a more specific question: given what we know about a student’s curiosity, how well can we predict their test score?

Even more importantly, it tackles another concern: what if curiosity only looks stronger because curious students also happen to be grittier or have growth mindset?

Well, we can look at how curiosity affects test scores while ‘controlling’ for perseverance and growth mindset. Effectively, we take all the students, imagine they have identical scores for perseverance and growth mindset and plot their curiosity against their score in each subject.

Then we do the same for growth mindset (keeping curiosity and perseverance scores constant) and for perseverance.

The steeper the slope, the more dramatic the effect, say, being more curious has on your mathematics score.^[7]

In every case, we can see that curiosity – the red line – is the steepest.

The gradients (or ‘betas’) of the lines tell us how many extra points you get in each subject when your trait score – curiosity, perseverance or growth mindset – increases.

A student who moves from average to one standard deviation above (i.e., from the 50^th percentile to the 84^th), perseverance and growth mindset staying the same, would score 14 points higher in reading. This is equivalent to about 9 months of schooling, by PISA’s measure.

Conclusion? Get curious.

In comparison, going from average to one standard deviation above in growth mindset adds just over six points.

The same shift in perseverance adds only half that.

The three traits aren’t just measuring the same thing in different ways, curiosity is doing something distinct.

And when it comes to PISA test scores, it makes a real difference.

The upshot

None of this means perseverance and growth mindset are useless.

They’re not.

And none of these traits operate in isolation — you can be curious and give up easily, or incredibly gritty about the wrong things. The strongest performers were those in the top 10% on all three measures together, which shouldn’t surprise anyone.

But if you had to bet on one trait — if you were designing a school culture, or thinking about how to raise a child, or deciding where to put your energy — the data makes a fairly unambiguous suggestion.

But wait a minute. Are we really measuring curiosity or grit or growth mindset here?^[8] For curiosity, it’s hard to say for sure without doing a battery of other reliable tests (which don’t really exist).

But we probably can say that the students who score highest on curiosity have the most positive attitude towards curiosity.

And, therefore, the students who most strongly think curiosity is a good thing are likely to do better on the PISA tests than those who strongly think perseverance or growth mindset are good things.

Curiosity isn’t just an add-on. It isn’t something to tag onto your education policy or school’s mission statement. It shouldn’t be an afterthought to grit or growth mindset.

Having curiosity might be the thing that matters most.

Which raises an awkward question for anyone who spent the last decade putting up growth mindset posters, or the last year giving assemblies about grit: why aren’t we talking about curiosity?

Acknowledgements

My extremely patient PhD supervisor Dr Richard Brock looked over the code and post beforehand, and made lots of very sensible suggestions about how to improve the method to make it more robust. My excuses for taking coding shortcuts at every stage appear in the footnotes below. Run the full thing - with weightings, all the PVs and an MLM and you’ll get almost the same values, honest.

Sorry Richard! All errors are, obviously, mine.

Endnotes and references:

^[1] Carol Dweck (2006) Mindset: The New Psychology of Success

^[2] Angela Duckworth (2016) Grit: The Power of Passion and Perseverance

^[3] Also, using all the PVs is complicated, takes a long time and my computer gets very hot while it’s doing it. The analysis is unweighted, too. PISA’s complex sampling design ideally requires student sampling weights to correct for unequal probability of selection, which would slightly widen our confidence intervals but is unlikely to change the substantive findings.

^[4] E.g. Baranes, A., Oudeyer, P.-Y., & Gottlieb, J. (2015). Eye movements reveal epistemic curiosity in human observers. Vision Research, 117, 81–90; Gruber, M. J., Gelman, B. D., & Ranganath, C. (2014). States of Curiosity Modulate Hippocampus-Dependent Learning via the Dopaminergic Circuit. Neuron, 84(2), 486–496; Van Lieshout, L. L., De Lange, F. P., & Cools, R. (2020). Why so curious? Quantifying mechanisms of information seeking. Current Opinion in Behavioral Sciences, 35, 112–117.

^[5] See Note 7.

^[6] Growth mindset scores were non-normally distributed, so we should use a different correlation test with these, but Spearman rank correlations tell a similar story.

^[7] These are very simple linear models. One thing a fuller analysis would do is also control for socioeconomic background — wealthier students tend to score higher on tests and may also report higher curiosity, simply because they have more opportunities to pursue their interests. We haven’t done that here. It’s possible that some of curiosity’s apparent advantage reflects this overlap. That said, the consistency of the finding across 19 very different countries — from Mexico to Singapore, from Greece to South Korea — makes a pure wealth explanation harder to sustain. A more rigorous analysis would also use multilevel modelling (MLM), which accounts for the fact that students aren’t independent observations — they’re clustered in schools, which are clustered in countries, each with their own characteristics. Our simple regression treats all 145,000 students as if they were independent, which slightly overstates our confidence in the results. In practice, the direction and rough magnitude of the findings are unlikely to change, but the standard errors would be larger and the significance thresholds harder to reach. Think of it this way: knowing that two students from the same school both score highly tells you less than knowing the same about two students from different schools on opposite sides of the world — the MLM approach adjusts for this.

^[8] The PISA questionnaire uses what’s called ‘self-report’ to measure traits. You can’t cheat on the test – or so the idea goes. You can tick strongly agree on all the curiosity or growth mindset questions, though. Are those who score highest on curiosity really the most curious students? We’ve got no way of knowing. Many people have tried other ways to measure curiosity; few have succeeded convincingly. It’s easier to make the case that we’re genuinely measuring grit or growth mindset on PISA than curiosity. Grit and growth mindset are both psychological constructs. They’re defined by getting a high score on a questionnaire. It’s a bit more debatable with curiosity. There are several competing ‘instruments’ for measuring curiosity (and PISA have created their own). There are even people (like me) who disagree that we should think of curiosity purely as a psychological construct – it’s a bit more complicated than that.