The Gender Trap: why we're seduced by obvious explanations

How we mistake patterns for causes, nature for destiny and moral preference for evidence

It’s the first day of a new academic year with no mistakes. New lanyards, paper cups of instant coffee and last year’s GCSE results projected on a screen in the school hall. The head gets up to say there’s much to celebrate — of course there is — attendance ent up, maths results improved, English did better than expected. But — and it’s always the same but — boys’ performance has dipped further across all GCSEs. The girls’ bars are higher; the boys’ are lower. The quickest route to school improvement, it sometimes seems, would be to change the admissions policy and only accept girls.

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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