How understanding risk patterns creates possibilities for intervention

"The pathways to criminality resemble an intricate combination lock - complex, unpredictable, yet vulnerable to disruption through preventive scaffolding. We cannot predict which specific alignment of adversities will unlock harmful behaviour in any individual, but we can systematically introduce protective factors that prevent these alignments from occurring. This is not just good practice; it is our moral obligation to every child..."

• Professor Stan Gilmour, Oxon Advisory

Imagine trying to unlock a safe with not just one combination lock, but five concentric dials, each containing at least ten numbers. The mathematical complexity is staggering - 100,000 possible combinations. Now imagine that these dials are constantly shifting, influencing one another, and responding to external forces in unpredictable ways. This is the reality of how risk factors for criminalisation operate across individual, family, peer, community and societal domains.

Winston Churchill observed "Out of intense complexities, intense simplicities emerge" and Robert Sapolsky noted (odd bedfellows, I acknowledge): "While little in childhood determines an adult behavior, virtually everything in childhood changes propensities toward some adult behavior".

These observations capture perfectly the paradox at the heart of our understanding of childhood adversity and the prevention of criminalisation. The pathways from early trauma to later criminalisation involve bewilderingly complex interactions, yet from this complexity emerges a simple truth: prevention is both possible and profoundly effective.

The inherent complexity of this system fundamentally undermines deterministic thinking. Recent Welsh prison research, documenting intergenerational patterns of adverse childhood experiences (ACEs), might initially suggest a fatalistic conclusion: that children's futures are predetermined by their parents' experiences. Nothing could be further from the truth.

"To predict a person's future is to deny them the very essence of humanity - the capacity to choose, to change, and to surprise even themselves. Our lives are not predetermined algorithms but unfolding stories where each page can bring an unexpected turn."

The complexity of human life helps us understand the crucial distinction between prediction and forecasting, a distinction that moves us beyond binary thinking about developmental risk. Prediction, in its traditional sense, implies certainty and determinism, a binary outcome where something either will or won't happen. This approach categorises children as either "future offenders" or "non-offenders," creating a false dichotomy that fails to capture the nuanced reality of human development.

Forecasting, by contrast, operates in the realm of probabilities and possibilities. Rather than making definitive claims about individual outcomes, forecasting acknowledges uncertainty while still providing valuable insights about potential trajectories. Just as the weather forecaster doesn't tell us with certainty whether it will rain tomorrow but instead offers probabilities that inform our decisions, developmental forecasting recognises patterns of vulnerability without claiming to predict individual destinies.

Several features of human development make deterministic prediction impossible, even while allowing for meaningful prevention:

First, the sheer mathematical complexity defies prediction at the individual level. With ten potential risk factors across five domains, the possible configurations are vast. Each configuration represents a unique developmental pathway, and small variations can lead to dramatically different outcomes. This complexity means that even the most sophisticated risk assessment tools fail to accurately predict individual trajectories.

Second, timing matters tremendously. The same risk factor experienced at different developmental stages may have entirely different impacts. Parental separation when a child is two years old creates different vulnerabilities than separation when the child is twelve. These temporal dynamics introduce another layer of complexity that undermines deterministic predictions.

Third, human beings demonstrate remarkable heterogeneity in their responses to adversity. Two siblings experiencing identical family circumstances may respond in profoundly different ways due to variations in temperament, interpretation of events, or exposure to different external influences. This differential susceptibility means that even identical risk profiles can produce divergent outcomes.

Fourth, protective factors intervene in unpredictable yet powerful ways. A supportive teacher, a talent-based opportunity, or a positive peer relationship can emerge unexpectedly, disrupting what might otherwise appear to be a predetermined pathway. These protective factors don't just add to the equation, they fundamentally transform it.

Fifth, human agency and meaning-making capacities introduce an element of freedom that defies deterministic models. Children are not passive recipients of risk factors but active interpreters of their experiences who make choices within constraints. This capacity for meaning-making introduces an irreducible element of unpredictability into human development.

“The combination lock analogy helps us understand why prediction is impossible while prevention remains entirely possible.” 

We cannot predict which specific combination of risk factors will "unlock" criminalised behaviour for any particular child. But we can systematically interrupt potential combinations through preventive scaffolding.

This forecasting approach allows us to think in terms of gradients rather than categories. Rather than classifying children as either "at risk" or "not at risk," we recognise that vulnerability exists along a continuous spectrum that changes over time. A child might move toward greater vulnerability in one period of development and toward greater resilience in another.

It acknowledges the biopsychosocial roots of multifinality, the multiverse of developmental pathways, the principle that similar starting points can lead to diverse outcomes. Two children with similar risk profiles might follow dramatically different developmental trajectories due to timing, protective factors, or individual differences in susceptibility. And also recognises equifinality, the principle that different developmental pathways can converge on similar outcomes. There are multiple routes to both positive and negative outcomes, which further undermines deterministic thinking.

The United Nations Convention on the Rights of the Child establishes that every child has the right to survival, development, protection, and participation. These rights are not conditional on behaviour or family background. When we view childhood adversity through this lens, our focus shifts dramatically from predicting which children might offend to ensuring all children receive the support necessary for healthy development.

“Preventive scaffolding works by introducing protective factors that disrupt potential risk configurations.” 

A stable, supportive adult relationship acts as a jamming mechanism that prevents other risk factors from aligning. Quality education creates alternative pathways that divert from potential harm. Healthcare, stable housing, and economic security provide the foundation that allows children to withstand pressures in other domains.

ACEs research is valuable precisely because it helps us understand patterns without determining individual destinies. It shows us where to build protective scaffolding, or where to focus our preventive efforts to have the greatest impact. When we know that children of parents with multiple ACEs face heightened vulnerability, we can ensure these families receive enhanced support without assuming negative outcomes.

The most effective prevention approaches recognise that while we cannot predict with certainty which individuals will experience negative outcomes, we can identify cohorts where concentrated support will yield the greatest preventive benefit. This population health approach acknowledges complexity while providing practical guidance for resource allocation.

By embracing complexity rather than denying it, we create space for hope. The very features that make human development unpredictable: sensitivity to timing, responsiveness to relationships, capacity for meaning-making, also make it amenable to positive influence.

Perhaps most importantly, a child-first approach recognises that children have an inherent right to supportive environments regardless of predicted outcomes. We provide scaffolding not because we believe children will become offenders without it, but because all children deserve the opportunity to develop their full potential.

This approach transforms how we understand intergenerational patterns of adversity. Rather than seeing these patterns as destiny, we recognise them as opportunities for targeted prevention. Research doesn't tell us which children will follow their parents into the criminal justice system; it tells us where to focus our preventive efforts to ensure fewer children experience this outcome.

In embracing complexity, we find not determinism but possibility, and in this possibility lies the true value of understanding risk factors for criminalised behaviour. This profound insight illuminates one of the most persistent challenges in social policy: how do we measure the efficacy of prevention when success means the absence of an event?

Traditional evaluation frameworks struggle with this fundamental paradox: how can we document, measure, and value what hasn't happened? The challenge has led to chronic underinvestment in prevention across social domains, from healthcare to criminal justice, with funders and policymakers reluctant to invest in outcomes they cannot directly observe.

The combination lock model of risk factors offers a revolutionary reconceptualisation of this challenge. Rather than attempting to measure the absence of negative outcomes directly, we can measure shifts in the configurations of risk factors that make those outcomes more or less probable. This shift transforms prevention from an unmeasurable non-event to a measurable change in risk landscapes.

The Data Integration Revolution

Modern approaches to prevention measurement leverage collaborative data sharing across multiple agencies to create a comprehensive picture of risk configurations at individual, family, community, and societal levels. Each agency holds different pieces of the combination lock:

Education authorities track not just academic achievement but attendance patterns, relational connections with teachers, peer dynamics, and behavioural indicators. Health services document developmental trajectories, mental health needs, and patterns of service engagement that signal emerging challenges. Social care maintains records of family functioning, parental capacity, and protective networks surrounding vulnerable children. Youth services capture data on engagement, interests, and positive relationship development. Housing authorities track stability, conditions, and neighbourhood characteristics. Police and justice agencies hold information about neighbourhood safety, family justice involvement, and community cohesion indicators.

When integrated through appropriate data-sharing agreements and robust privacy protections, these disparate data sources create a multidimensional map of risk and protective factors across a population. This integration enables several approaches to measuring prevention efficacy:

Risk Configuration Tracking

By tracking the prevalence of high-risk configurations over time, we can measure whether our preventive interventions are successfully disrupting potential alignments.

For example, by implementing an integrated data system that tracks combinations of school exclusion, family adversity, and community risk factors. Evaluations can measure prevention success not by counting crimes that didn't occur, but by documenting reductions in the prevalence of high-risk configurations known to precede criminalisation. When the data shows fewer young people with multiple aligned risk factors following preventive intervention, these findings can reasonably claim prevention efficacy without needing to prove a counterfactual.

Protective Factor Amplification

The Welsh Flying Start and Enhanced Health Visiting programme, for instance, measures prevention efficacy through documented increases in parent-child attachment, maternal wellbeing, and family resilience indicators. These protective factors are known to reduce the probability of subsequent child maltreatment and later antisocial behaviour. By measuring increases in protective configuration rather than decreases in negative outcomes, they transform an unmeasurable counterfactual into measurable positive change.

Developmental Trajectory Analysis

When multi-agency data is collected longitudinally, it becomes possible to identify typical developmental trajectories that lead toward or away from negative outcomes. Prevention efficacy can then be measured by documenting shifts in these trajectories following intervention.

The UK's Violence Reduction Units analyse developmental pathways using what integrated data from health, education, social work, and justice systems is currently available. They identify common trajectories toward violent offending, such as the sequence from early trauma to school disengagement to antisocial peer association to initial violence. Their prevention frameworks measure success by documenting disruptions in these typical pathways, with the expectation of showing more divergent, positive trajectories following preventive intervention compared to historical patterns.

Early Warning System Efficacy

Multi-agency data integration enables the development of early warning systems that identify emerging risk configurations before they fully align. Prevention efficacy can be measured through the system's ability to accurately identify vulnerability and the subsequent impact of earlier, less intensive intervention.

The Early Help Assessment (EHA) framework used across English local authorities integrates indicators from multiple agencies to identify families where risk factors are beginning to accumulate. Prevention efficacy is measured both by the system's ability to accurately identify emerging vulnerability and by the reduced need for subsequent statutory intervention. When families identified through early warning systems show lower rates of escalation to child protection or justice involvement compared to similar families identified later, prevention value is demonstrated.

Economic Return Measurement in Prevention Programmes

Economic return measurement represents one of the most compelling approaches for policymakers considering investment in prevention programmes. By tracking service utilisation comprehensively across multiple systems over time, analysts can document precisely how preventive interventions reduce costs across various domains, providing a robust financial justification for early intervention approaches.

The Washington State Model

The Washington State Institute for Public Policy (WSIPP) has pioneered this sophisticated approach, using integrated administrative data to calculate economic returns from prevention programmes with remarkable precision. Their methodology documents how early support interventions reduce costs across education, health, social care, and justice systems over extended timeframes, sometimes spanning decades.

The WSIPP model transforms the seemingly unmeasurable concept of "prevention" into concrete financial returns by measuring reduced service utilisation as a proxy for prevented negative outcomes. This approach provides compelling evidence for resource allocation decisions.

The Adolescent Diversion Project (ADP) Case Study

The Adolescent Diversion Project provides an alternative to formal sanctions in the juvenile justice system by matching diverted youth with volunteer caseworkers who deliver tailored community-based services focused on skill building. The benefit-cost analysis demonstrated remarkable returns.

Relevance for UK Policy Context

While the American example is instructive, this approach has significant relevance for UK policy contexts. Similar methodologies could be applied to evaluate programmes like Youth Offending Teams, Family Nurse Partnerships, or Sure Start Children's Centres.

The approach aligns well with HM Treasury's Green Book guidance on economic evaluation, particularly its emphasis on measuring costs and benefits across multiple domains and stakeholders. It also provides a robust framework for responding to the increasing emphasis on value for money in UK public services.

This methodology offers a powerful tool for making the economic case for prevention, transforming what can sometimes seem like an act of faith into a credible investment proposition with measurable returns.

Cohort Comparison Approaches

When randomised trials are impractical, multi-agency data enables robust quasi-experimental designs that compare similar populations receiving different levels of preventive support. These comparisons provide evidence of prevention efficacy without requiring perfect counterfactuals.

The Better Start Bradford programme, for instance, uses integrated data to compare developmental trajectories of children in intervention areas with those in similar demographic areas without enhanced prevention services. By controlling for baseline risk factors through sophisticated statistical matching, they identify differences in outcomes attributable to preventive scaffolding. This approach allows them to measure prevention efficacy without requiring randomisation.

The Ethical Foundations of Data-Driven Prevention Measurement

This data-driven approach to measuring prevention must be grounded in ethical principles that respect human dignity and potential. The combination lock model understands risk factors as creating vulnerability, not destiny. Measurement approaches must therefore avoid deterministic labelling or stigmatising categorisations.

Several ethical principles should guide this work:

• Data should focus on modifiable factors rather than fixed characteristics, preventing fatalistic thinking.
• Analysis should emphasise population patterns rather than individual prediction, protecting privacy and dignity.
• Measurement should track both risk reduction and protective factor amplification, maintaining strength-based rather than deficit-focused approaches.
• Communities should be involved in determining what outcomes matter and how they are measured, ensuring relevance and empowerment.
• Data governance must balance integration benefits with rigorous privacy protection through appropriate anonymisation and security.

Embracing Complexity, Enabling Prevention

By reconceptualising prevention measurement through the lens of risk configurations rather than non-events, we transform prevention from an act of faith into an empirically verifiable investment. This shift carries profound implications for policy, practice, and resource allocation.

The combination lock model recognises that human development is neither completely predictable nor completely random, it exists in the complex middle ground of probabilistic patterns influenced by countless factors interacting across multiple systems. This complexity makes binary prediction impossible, but it makes prevention entirely possible.

When we measure changes in risk configurations rather than attempting to count prevented negative outcomes directly, we create a robust framework for evaluating prevention that respects human complexity while enabling empirical verification. We move beyond the false choice between deterministic prediction and complete uncertainty, finding a middle path that acknowledges patterns without claiming certainty.

“In embracing complexity, we find not determinism but possibility. And in this possibility lies not just the value of understanding risk factors, but the practical ability to measure our success in creating conditions where harmful outcomes become less likely and positive development becomes more possible for all children.”

The combination lock may contain too many variables, too many shifting interactions, for deterministic prediction. But these same features that make prediction impossible make prevention entirely possible and, through sophisticated data integration, increasingly measurable. In this way, we transform the prevention paradox from an insurmountable challenge into an opportunity for more sophisticated, humane, and effective approaches to supporting children exposed to vulnerabilities.