The number that should stop the conversation
Anne Kielland of the Fafo Institute and Jing Liu of the University of Hong Kong surveyed 274 Norwegian neurodivergent adolescents aged 13 to 17 — children with diagnoses of ADHD, autism, and Tourette syndrome — about their experience of school. The headline finding, published in the Scandinavian Journal of Educational Research in June 2026, is stark. Across the sample, 86% described periods when they felt some or frequent fear of going to school. Just under half — 128 of 274 — reported periods when they often dreaded it. Another third reported being sometimes scared. More than half had experienced periods of high absence.
The study uses the term involuntary school absenteeism rather than the more familiar “school refusal,” and the distinction is the whole argument. “Refusal” implies a child who will not attend. The involuntary framing, which Kielland and Liu adopt because it is the term families themselves now use, describes a child who cannot — one whose absence is, in the authors’ framing, a rational, sovereign act of self-preservation rather than a behavioural problem to be corrected. The study found that actual truancy in this population was, echoing earlier work, almost non-existent. The children staying home were not avoiding effort. They were protecting themselves from harm.
The reframing matters because the entire apparatus around school absence is built on the opposite assumption. Standard advice operates on the principle that “absence yields more absence” and that “every day counts,” pressing for the fastest possible return. Extensive resources go into child psychiatric assessment, individual therapy, and behavioural programmes aimed at the child. Child protection services had been contacted in half the families in the sample. The system reads a frightened child who cannot attend as a child who will not, locates the problem inside the child, and intervenes accordingly. The 86% figure is what that misreading produces at scale — a population of children afraid of the place they are legally compelled to be, met with interventions calibrated for a problem they do not have.
It isn't the family — and the data is unusually blunt about it
The study tested associations between school outcomes and three domains: characteristics of the child, of the family, and of the school. The family result is the one that should change the conversation, because of how completely it comes back null. None of the parents’ socioeconomic characteristics — marital status, education level, employment, household income, ability to manage an unexpected expense, home ownership — were associated with any absenteeism outcome. Not one. The standard assumption that school absence tracks family disadvantage, poverty, or low parental education simply did not hold in this neurodivergent population.
The school domain told the opposite story. Every single school-related factor the study measured was significantly associated with the outcomes, and all in the expected direction. Having at least one teacher who cares. Most teachers being kind. The other pupils being kind. Having a best friend at school. Feeling a sense of academic mastery. Each of these was strongly protective against absence, dropout, and school-related fear, and strongly associated with still attending full-time. A best friend at school carried one of the largest effects in the entire analysis.
Set the two results side by side and the structural reading is unavoidable. The thing that predicted whether a neurodivergent child could attend school was not anything about the child’s home, income, or family stability. It was whether the school was a place where they had a friend, a teacher who cared, and an experience of mastering the work. The variable is the environment. The data does not merely suggest this — it draws the line about as sharply as social-science data ever does, with the family domain returning nothing and the school domain returning everything.
This is the demand-structure argument arriving as evidence rather than assertion. My corpus has made this case across domains — that what gets caught as individual deficit is the substrate meeting an environment calibrated against it. Here the calibration is measurable. Where the school environment met the child — a caring adult, a friend, accessible work — the child attended. Where it did not, the child became afraid and eventually could not. The mismatch is not a metaphor in this dataset. It is the regression table.
Schools are being told to make children "robust"
The study situates its findings against two decades of Norwegian education policy, and this is where it becomes genuinely uncomfortable reading. Successive government whitepapers have escalated the demand that children acquire “self-regulation” — defined across documents as the ability to manage one’s own actions, feelings, and thinking, to resist distraction, to work purposefully, to adapt in interaction with others. The 2019 whitepaper mentioned self-regulation 63 times. These are precisely the capacities that the neurodivergent profile, by its nature, finds hardest to deliver on demand — and the policy framework treats them as the baseline every child must reach.
The 2024 whitepaper, titled “A More Practical School,” sharpened the position rather than softening it. Confronted with rising absence, its stated principle was that when students experience stress and discomfort meeting the school’s demands, the school must not reduce those demands but must instead teach children to handle situations they will face for the rest of their lives. The document emphasised building “robustness” and “resilience,” and framed absence as a problem of children’s “lowered endurance” to be corrected. The demand structure, encountering evidence that it was producing distress, concluded that the children needed to be made tougher.
This is the round hole tightening, documented in government text. The policy does not ask whether a school environment that produces fear in 86% of its neurodivergent students might be the thing requiring redesign. It asks how to make the children more able to withstand it. Kielland and Liu cite Robert Chapman’s Empire of Normality and the narrowing of the “child norm” — the way success increasingly depends on characteristics that overlap less and less with the strengths found in neurodivergence. Each whitepaper raises the bar of expected self-regulation, and each increment widens the gap between what the structure demands and what the neurodivergent substrate can supply against that structure. The absence is the gap made visible. The policy response is to demand the child close it.
The authors note the obvious tension the policy ignores: training a neurodivergent child to “fit in” and “toughen up” cannot be done without driving more masking — the suppression of natural regulation, including the movement and stimming that neurodivergent children use to manage stress in the first place. The robustness agenda does not build capacity. It teaches concealment, and concealment carries its own well-documented mental health cost.
Attendance was never the outcome that mattered
The study’s sharpest methodological move is to refuse attendance as the measure of success. A child returned to full-time school who remains afraid every day has not been helped; they have been relocated. Kielland and Liu therefore shift their analytic focus from whether children attend to whether they are afraid — and the picture that emerges is of children for whom physical presence and genuine safety have come apart. Among children still often scared at the time of the survey, 15 were attending full-time and 25 part-time. Present, compliant, and frightened. The attendance statistic counts them as a success.
This is where the study’s quietest finding does the most work. The only child-related factor associated with reduced fear was being on medication. Medication showed no association with whether absence happened at all, but it was positively associated with current attendance and negatively associated with fear. Read structurally, that is not a treatment success. It is the same pattern I’ve traced through workplace AI and ADHD prescribing — medication functioning to make the substrate tolerable to an unchanged environment, so the child can keep attending the structure that frightens them. The authors ask the question directly: does medication become a too-convenient substitute for investing in the enabling schools these children need?
The answer the data points toward is that the entire apparatus is aimed at the wrong target. The intervention resources go to making the child attend — psychiatric assessment, behavioural therapy, medication, and the pressure to return fast. The thing that actually predicted a child being in school and unafraid was a caring teacher, a friend, and a sense of mastery: properties of the environment, none of which appear on the intervention list. The system measures attendance, optimises for attendance, and in optimising for it produces frightened children who are present. The measure that mattered — whether the child is safe — was never the one being managed.
What the study leaves as an open question is also its quiet challenge. It asks what a school environment in which neurodivergent students can flourish actually looks like, and how such environments are built — and notes that the research has not yet answered it, because the research, like the policy, has been busy with the child. The 86% are not a measure of how fragile neurodivergent children are. They are a measure of how poorly the environment has been built for them, and how completely the apparatus has declined to look at the environment as the thing that could change.
Citations
Kielland, A. & Liu, J. (2026) — School absenteeism and fear of school among Norwegian neurodivergent adolescents 13–17 years of age — Scandinavian Journal of Educational Research
Chapman, R. (2023) — Empire of Normality: Neurodiversity and Capitalism — Pluto Press
Totsika, V., Hastings, R. P., Dutton, Y. et al. (2020) — Types and correlates of school non-attendance in students with autism spectrum disorders — Autism
Botha, M. & Frost, D. M. (2020) — Extending the Minority Stress Model to Understand Mental Health Problems Experienced by the Autistic Population — Society and Mental Health
Chapman, R. & Carel, H. (2022) — Neurodiversity, epistemic injustice, and the good human life — Journal of Social Philosophy
