There is a phrase that appears in almost every school improvement plan, every self-evaluation document, and every job interview for a promoted post: "evidence-based." We are committed to evidence-based practice. Our improvement priorities are evidence-based. Our approach to teaching and learning is grounded in evidence. The phrase has become so prevalent that it has achieved a kind of semantic weightlessness — everyone uses it, everyone nods when it is used, and very few people stop to ask what it means or whether the evidence it refers to actually exists.
The instinct behind the phrase is sound. Schools should base their decisions on evidence rather than assumption. Teaching should be informed by what the research says works. Improvement priorities should be grounded in data about what is happening rather than in hunches about what might be. None of this is controversial. The problem is not the principle. The problem is the gap between claiming to be evidence-based and actually being evidence-based — a gap that is, in my experience, wider than most schools acknowledge.
Three different things called "evidence"
Part of the confusion is that the word "evidence" is used to describe at least three different things in education, and they are not interchangeable.
The first is research evidence — the findings of academic studies, meta-analyses, and randomised controlled trials that tell us, in general terms, what works. This is the evidence that Hattie synthesises, that the EEF publishes guidance reports on, that Rosenshine distilled into his ten principles. It is the evidence that underpins the phrase "evidence-based practice" when it is used by researchers, and it refers to a body of knowledge that has been produced through systematic inquiry, peer review, and replication. When a school says it is evidence-based because it has adopted Rosenshine's Principles, it is drawing on this tradition.
The second is school-level evidence — the data that a school generates about itself. Attainment results. Attendance figures. Exclusion rates. Survey responses. Observation records. This is the evidence that self-evaluation frameworks ask schools to engage with, and it is the evidence that an inspector will ask to see. It is local, specific, and contextual. It tells you what is happening in this school, with these young people, at this point in time.
The third is professional evidence — the accumulated knowledge that teachers and leaders hold about their practice, their pupils, and their context. This is the evidence that informs daily decisions in every school: the teacher who knows that this class responds better to structured tasks than open enquiry, the pupil support leader who knows that this young person's behaviour deteriorates when there is disruption at home, the headteacher who knows that this department's improvement trajectory has stalled for reasons that do not appear in any data set. It is experiential, tacit, and often unrecorded.
This section includes an interactive chart — open the page in a browser to explore it.
A school that claims to be evidence-based is, usually, drawing on some combination of these three — but the combination is rarely made explicit, the quality of each component varies enormously, and the connections between them are almost never systematic. A school might cite Rosenshine's Principles in its improvement plan (research evidence) without having any observation data showing whether those principles are enacted in classrooms (school-level evidence), and without any mechanism for connecting the professional knowledge of individual teachers to the improvement priorities that are supposed to guide their practice (professional evidence).
This is not hypocrisy. It is the consequence of a system that has adopted the language of evidence without building the infrastructure to produce it.
The school-level evidence gap
Of the three types, it is school-level evidence that presents the most significant gap in most schools — not because schools do not generate data, but because the data they generate is often the wrong data, or the right data held in the wrong form.
Attainment data is plentiful. Every school can tell you, in considerable detail, what percentage of pupils achieved what grades in which subjects. This data is important and it drives significant decisions — about curriculum design, about staffing, about intervention. But it is also retrospective. By the time the attainment data tells you that a cohort has underperformed, the cohort has left. The data describes what happened. It does not describe why.
Observation data — the evidence of what is happening in classrooms — is, in most schools, thin to the point of absence. Not because observations do not happen, but because the observation systems in use are not designed to produce data. They are designed to produce paragraphs. A paragraph of impressionistic prose, however well-written, is not evidence in any meaningful analytical sense. It cannot be aggregated across observations. It cannot be compared across departments. It cannot be tracked over time. It is a snapshot, captured once, filed, and forgotten. The school that claims its improvement priorities are evidence-based but cannot show you structured observation data from the last six months is a school that is evidence-based in aspiration but not in practice.
Attendance data exists but is frequently underused. Most schools monitor attendance as a compliance function — which pupils are present, which are absent, which have hit the threshold that triggers a referral. Far fewer schools use attendance data as an analytical tool — tracking patterns over time, identifying the early warning signs that precede significant disengagement, connecting attendance trajectories to the attainment and pastoral data that might explain them. In my own school, we have invested heavily in this kind of analysis, connecting attendance to attainment and wellbeing indicators so that the pattern becomes visible before it becomes a crisis. The work has been transformative. But it is work that the existing systems do not do automatically. It has to be built.
Wellbeing data is the most fragile of all. Most schools capture some form of wellbeing evidence — through surveys, through pastoral conversations, through SHANARRI-based assessments — but it is almost always held separately from the academic data, stored in a different system, owned by a different team, and used for a different purpose. The young person's wellbeing and their attainment are treated as parallel concerns rather than as interconnected dimensions of the same experience. The consequence is that neither the pastoral team nor the teaching staff has the full picture. Each holds a fragment and makes decisions based on that fragment, unaware of what the other fragment would reveal.
The evidence-informed distinction
There is a distinction in the literature between "evidence-based" and "evidence-informed" practice, and it is a distinction worth taking seriously. Evidence-based practice, in the medical sense from which the term originates, means that clinical decisions are made on the basis of the best available research evidence, integrated with clinical expertise and patient preferences. It implies a direct, causal relationship between the evidence and the decision. In medicine, this is achievable because clinical trials can control for variables in ways that educational research cannot.
Education is different. The variables that affect whether a teaching practice works — the pupils, the teacher, the subject, the context, the time of day, the relationship between the teacher and the class, the thirty other things that happened before the lesson started — cannot be controlled in the way that a drug trial controls for placebo effects. The research can tell us that retrieval practice is, on average, more effective than rereading. It cannot tell us that retrieval practice will work with this specific class, taught by this specific teacher, in this specific school, on this specific Thursday afternoon. The context mediates the evidence in ways that the evidence itself cannot predict.
For this reason, many researchers — and I agree with them — prefer the term "evidence-informed." It signals that research evidence is one input to professional decision-making, not the sole determinant of it. The teacher's professional judgement, their knowledge of their pupils, their understanding of their context — these are not obstacles to evidence-based practice. They are essential components of it. A teacher who ignores the research is operating on instinct alone, which is insufficient. But a teacher who follows the research mechanically, without adapting it to their context, is not being evidence-based. They are being compliant. And compliance is not the same thing as professional practice.
The implication for schools is that being evidence-informed requires both access to the research and the capacity to interpret it in context. It requires school-level data that is structured enough to reveal whether the research-informed practices are being enacted. And it requires professional cultures that treat evidence as a tool for inquiry rather than as a script to be followed. The school that says "we are evidence-based because we have adopted Rosenshine's Principles" has completed the first step. The school that says "we are evidence-informed: we adopted Rosenshine's Principles, our observation data shows they are being enacted in 60% of lessons with mixed quality, and our professional learning programme is targeting the specific practices where enactment is weakest" has completed the cycle. The difference is not semantic. It is the difference between claiming to use evidence and actually using it.
What the infrastructure looks like
If I have learned anything from years of trying to make evidence-informed decisions in a real school, it is that the quality of the decisions depends on the quality of the evidence, and the quality of the evidence depends on the systems that produce it. Good intentions are not sufficient. A school that values evidence-informed practice but captures observation data in narrative paragraphs, stores attendance data in one system and attainment data in another, and holds pastoral information in the professional memory of individual staff members is a school that has the values but not the infrastructure.
The infrastructure does not need to be elaborate. It needs to be structured. Observation evidence needs to be captured in a form that is specific enough to describe what was seen, consistent enough to be compared across observations, and connected enough to aggregate into patterns. Attainment, attendance, and wellbeing data need to be accessible in a single view — not necessarily in a single system, but in a way that allows the connections between them to be visible. Professional knowledge needs to be recorded and shared rather than held in individuals' heads.
This is, ultimately, what Learning Lens is designed to provide — not a replacement for professional judgement, but the evidence base that professional judgement needs to operate effectively. The structured observation data that tells a school not just that practices are being used but how they are being used, and whether the pattern across departments and over time matches what the improvement plan intended. The connections between classroom evidence and the wider data about young people that give context to what the observation revealed. The infrastructure that turns "evidence-based" from a phrase in the improvement plan to a description of how the school works.
Many schools have intelligent, committed professionals making the best decisions they could with the evidence available to them. The problem was never the people. It was the evidence. And the evidence was limited not by the school's commitment to using it, but by the tools available to produce it.
Jamie Scobie writes from extensive experience in Scottish secondary education, including pastoral care, data for improvement, and school self-evaluation. This blog is an independent publication: he writes in a personal capacity as the creator of Learning Lens, writing about classroom observation, teaching evidence, and education policy. He speaks here only for himself and for Learning Lens, not for any employer or other organisation. He holds an MSt from Cambridge (Distinction) and a Masters from Stirling.