“The intervention worked on average. The children stubbornly persisted in being individuals. The system interpreted this as an implementation challenge.” The rise of evidence-informed practice has been one of the most positive developments in education over recent decades. The UK is home to cutting-edge researchers whose work around education and inclusion provides an increasingly robust foundation for educational decision-making. This plays a pivotal role in the current SEND reforms. Strengthening ordinarily available provision, improving inclusion and increasing the use of evidence-informed approaches before children require statutory support are sensible and commendable ambitions. But there needs to be confidence that support is robust, equitable and grounded in sound evidence for these ambitions to succeed. The challenge lies in building systems that combine evidence, accountability and professional judgment in ways that remain responsive to individual children. This is particularly important for children experiencing barriers to learning, where variability is the norm rather than the exception. Research findings are inherently probabilistic. Evidence from inferential statistics provides insight into what tends to happen, on average, under certain conditions and according to certain measures. As these findings travel from journals into policy, implementation frameworks and practice, the caveats that accompany them often fade from prominence. “On average” is interpreted as “always”. “Under these conditions” is applied “in all circumstances”. “Associated with improved outcomes” becomes reified as “the correct approach”. Attraction This is the certainty trap, the point at which findings that are probabilistic at the level of populations become expectations that are deterministic at the level of individuals. Part of the attraction is understandable. Education is characterised by complexity, ambiguity and competing demands. Research offers a degree of stability while providing a shared language that can support consistency, fairness and accountability. Education is shaped by inescapable tensions between consistency and responsiveness, standardisation and adaptation. The success of the SEND reforms may depend less on resolving these tensions than on recognising them and navigating them well. Difficulties often emerge when interventions do not produce anticipated outcomes. Educational change is rarely generated by a single ingredient. Progress reflects an interaction of teaching approaches, relationships, expectations, confidence and context. While research can identify approaches associated with improved outcomes, the precise mechanisms often remain uncertain. Trust, belonging and responsive teaching are difficult to standardise and measure, yet few practitioners would deny their importance. Professional judgment is often the means through which these relational factors are enacted in practice. The certainty trap has consequences. When interventions supported by research do not produce expected outcomes, explanations can increasingly focus on implementation. Sometimes this is entirely appropriate. At other times, legitimate questions about context, individual difference or the nature of the barrier itself risk being overlooked. Professional dignity Professional observations that do not fit expectations can be interpreted as resistance rather than information. The response then centres on monitoring, fidelity checks and compliance. This has important implications for professional dignity. When observations that do not fit expectations are approached primarily as implementation problems rather than potentially valuable sources of information, expertise risks becoming associated with adherence rather than interpretation. Professional practice, however, derives much of its value from the ability to exercise judgment under conditions of uncertainty. Diminishing professional judgment and dignity can result in defensive practice, reducing curiosity, responsiveness and ultimately effectiveness. Response to intervention (RTI) offers one way negotiating these tensions. RTI begins with approaches that have strong research support and expects them to be implemented thoughtfully and as intended. Monitoring response is built into the model, preserving accountability, transparency and a focus on outcomes. Central to RTI is the recognition that a child or young person’s response to intervention constitutes an important source of evidence in its own right. When outcomes diverge from expectations, that divergence becomes the basis for further inquiry, reflection and adaptation. Practitioners begin with what evidence from research suggests is most likely to be effective, check that approaches are being delivered appropriately, observe what happens and adjust where necessary. Professional judgment comes into its own when reality diverges from expectations. Certainty is more comfortable As evidence-informed packages of support are developed through the SEND reforms, avoiding the certainty trap will require retaining the caveats that accompany the evidence itself. This means creating frameworks with sufficient structure to secure fairness and sufficient flexibility to respond when children diverge from the patterns observed in research. Holding uncertainty is hard work. Certainty is more comfortable, more defensible and easier to scale. Yet the children and young people facing the greatest barriers to education are often those whose needs do not align neatly with what we might expect. They are, almost by definition, the group for whom population-level expectations are least likely to provide a complete explanation of their experience. They deserve the effort required to remain curious when certainty would be easier. Co-authored by Duncan Astle of the University of Cambridge, Jackie Masterson of the UCL Institute of Education and Maria Ionides, senior educational psychologist at Cambridgeshire County Council.