In simple language
What leaders should take from the research
Methods guidance: changed outcomes and attribution differ. UK evaluation guidance distinguishes observing a change from showing that an intervention caused it. A credible comparison strengthens attribution; a single before/after observation leaves many alternative explanations. HM Treasury & Evaluation Task Force (2026)
Implementation guidance: fit includes people and systems. EEF recommendations identify the approach, infrastructure and people who enable change as contextual factors. Suitability and feasibility need examination alongside what a study found. EEF (2024)
Editorial interpretation: keep three statements separate. “The study reported…” describes the source; “we think this matters because…” explains the connection; “we propose…” identifies the local action. A proposal does not gain evidence strength simply by appearing beside a citation.
Proposed action: test the important assumption. Identify the difference most likely to affect success, such as language, learner age, timetable, expertise or available support. Decide what local evidence would justify continuing or adapting.
Evidence strength and limits
Useful, with boundaries
This explainer applies UK methods and England implementation guidance to school decision-making. It is an editorial explanation, not a systematic research review or an evaluation qualification. Randomisation is useful for causal questions but does not guarantee transfer; qualitative evidence can explain experience without estimating an average causal effect. A local pilot should be proportionate and appropriate to the decision.
For your next leadership meeting
Seven questions for your leadership team
Use the prompts to examine your own context and decide what, if anything, to change.
What decision are we making?
- Which alternatives remain plausible?
- Who will experience the choice?
What material did we actually read?
- Was it full text, selected sections or an abstract?
- Which relevant detail remains unverified?
What does the design establish?
- Is the claim causal, associational or descriptive?
- What uncertainty is reported?
How similar is the setting?
- Which population, language or curriculum differs?
- Which difference could matter to the mechanism?
What would implementation require?
- Do people have time and support?
- What other work would need to change?
How will we review locally?
- What baseline and comparison are feasible?
- Which alternative explanation should we record?
What would change our decision?
- What outcome or burden would make us stop?
- Who owns the review and when?
What kind of evidence answers this question?
| Question | Useful evidence | Limit to retain |
|---|---|---|
| Did this approach cause an outcome? | A well-designed trial or credible comparison analysis | Design assumptions, uncertainty and evaluated setting |
| How did people experience it? | Interviews, observation and contextual accounts | Whose accounts are included and missing |
| Was it delivered as intended? | Participation, dosage, support and implementation records | Delivery is not the same as a learner benefit |
| Is it worth trying here? | Relevant research plus local need, feasibility and stakeholder evidence | The decision is a judgment under uncertainty |
Proposed school action
These are Cited Leadership discussion proposals, not a tested intervention package.
- Decision: name the choice, affected learners and alternatives.
- Source: record title, date, material actually read, design, setting, outcome and limitation.
- Connection: explain why this evidence is relevant and which differences could break the proposed mechanism.
- Implementation: name people, time, materials, training and work to remove.
- Review: choose indicators, timing, owner and a reason to continue, adapt or stop. State what the design will leave uncertain.
Fictional example: a school sees better assessment results after introducing collaborative planning. A new cohort and curriculum change also occurred. Leaders can report improved results and review the planning process; that observation alone does not show that planning caused the improvement. The example is invented and contains no client data.
Primary sources
References
- HM Treasury & Evaluation Task Force. (2026). Quality in policy impact evaluation (QPIE). Supplement to the Magenta Book. Updated 15 May 2026.
- Design
- Evaluation-methods guidance.
- Setting
- UK government policy evaluation; applied here as a methodological analogy.
- Material actually reviewed
- Original HTML executive summary and sections 2–4 on attribution and research designs.
- Limitations
- An appropriate comparison can strengthen attribution; it does not automatically remove confounding or establish transfer to another school.
- Source reviewed
- 12 October 2026
- Education Endowment Foundation. (2024). A School’s Guide to Implementation (third edition). Published 24 April 2024. Web overview and recommendations on behaviours, contextual factors and process.
- Design
- Evidence-informed implementation guidance, not a trial of a Cited Leadership resource.
- Setting
- Guidance developed for schools and other education settings in England.
- Material actually reviewed
- Original HTML overview and all three recommendation pages: behaviours, contextual factors and process; full PDF and underlying review not reviewed for this refresh.
- Limitations
- A framework for implementation decisions does not establish an effect for a particular programme or validate these worksheets.
- Source reviewed
- 12 October 2026
Reviewed source route 1 · Reviewed source route 2 · Reviewed source route 3
These questions are prompts for local discussion. The cited research does not determine what any one school should do.