Falling grades, rising risk: what pupil records reveal about offending

A pupil’s exam results are not just a measure of academic ability. According to a major new study, they may be one of the most practical early-warning signals available for identifying children at risk of entering the criminal justice system.

Research from King’s College London, tracking 4.3 million pupils in England born between 1990 and 1997, has found that students whose performance at school declines relative to their peers face a higher risk of contact with the criminal justice system. The study, published in the Journal of Developmental and Life-Course Criminology, is thought to be the largest of its kind, and its authors argue that school data — already collected on every child in the state system — could offer forces and partner agencies a rare, scalable opportunity to intervene before offending starts.

Researchers linked two national administrative datasets: the Department for Education’s National Pupil Database and the Home Office’s Police National Computer. Pupils’ attainment was tracked through the three points at which statutory tests are taken in England — age 7, age 11 and age 16 — and each child was sorted into one of five trajectory groups based on how their relative performance moved over that period, from consistently average or high, through to consistently low, with two “mover” groups in between who either improved or declined relative to their peers.

Against those trajectories, the team then checked Police National Computer records for first convictions or cautions, both before the end of secondary school and afterwards, up to age 21. Crucially, the offence categories covered were not limited to low-level misconduct: everything from theft and criminal damage to drug offences and serious violence was included.

Overall, around 8.8 per cent of the cohort, 369,557 pupils, were first convicted or cautioned before the end of Year 11, and a further 5 per cent, or 210,936 pupils, received a first conviction or caution in young adulthood. But those averages concealed sharp differences by trajectory.

Among pupils whose relative attainment declined through school, one in three had received a first conviction or caution before leaving secondary school, and one in ten went on to a first conviction or caution in young adulthood. These were both markedly higher than any other group. Pupils who performed below average throughout school were also at elevated risk, though less acutely than the “decliners”.

Perhaps the most operationally useful finding, however, concerns pupils who started out behind. Among those performing below average early in school, the odds of offending in young adulthood were 53 per cent higher than average attainers if their relative performance stayed low, but only 4 per cent higher if it improved relative to their peers. In other words, closing the gap, even just a little, appeared to all but neutralise the added risk associated with a slow start.

The pattern held across offence types and was, if anything, more pronounced for serious violence — the offence category of most direct interest to forces running Violence Reduction Units and contextual safeguarding programmes.

The study’s authors are careful not to claim that falling grades cause offending. Attainment, they argue, functions as a signal rather than a driver — a visible marker of underlying difficulties that could just as easily be at home, in a child’s mental health, or in their peer relationships, as in the classroom itself.

Lead author Dr Alice Wickersham, a Research Fellow at King’s and at Administrative Data Research UK, put it this way: schools records could help identify when pupils “might need support,” treating attainment as a proxy for wider circumstances rather than a judgment on ability alone.

Co-author Professor Stephen Scott, Professor of Child Health and Behaviour at King’s, pointed to parenting programmes, including online delivery models, as an intervention already shown to reduce the antisocial behaviour that typically precedes offending, while also improving attainment among disadvantaged pupils.

One counterintuitive finding merits scrutiny: the elevated offending risk usually associated with being male, from a black or mixed ethnic background, eligible for free school meals, or in receipt of special educational needs support was, in this dataset, weaker among pupils on declining or consistently low attainment trajectories than among their peers on stronger trajectories. The authors suggest this may mean non-academic risk factors such as motivation, emotional regulation, family and neighbourhood pressures, carry more weight for these groups than attainment itself. However, they flag that the finding runs against prior assumptions and needs further work before firm conclusions are drawn.

For forces, YOTs and Violence Reduction Units already working the “public health approach” to crime prevention, the study adds empirical weight to an argument long made anecdotally by school liaison officers: that a sudden dip in a pupil’s grades is worth as much attention as a change in their friendship group or attendance record.

Because attainment data is already collected nationally and systematically, the authors suggest it could be folded into existing information-sharing arrangements between schools, social care and policing partners with comparatively little additional cost. Though the researchers are careful to flag the risk of labelling effects, and stress that any response should be support-led rather than enforcement-led.

The research was funded by Administrative Data Research UK, part of UK Research and Innovation, and the authors note that unmeasured factors — family circumstances, IQ, temperament — mean the link cannot yet be called causal. A separate sibling-based study elsewhere has found evidence consistent with a causal pathway, which the King’s team suggest as a direction for further work.