Begin with the claim you need to test
Evaluation should be designed around a decision: whether to continue, adapt, expand or stop an intervention. That decision determines which questions matter and how strong the evidence needs to be.
Monitoring can establish what was delivered and whether outcomes changed. Evaluation examines how and why change occurred and, where the design permits, the contribution made by the intervention. HM Treasury's Magenta Book provides general guidance for planning evaluation and selecting suitable approaches.
Match the claim to the design
Before-and-after measurement can describe change but cannot, on its own, separate intervention effects from wider trends or differences in participants. Use an appropriate comparison where feasible; otherwise be explicit about the limits of causal inference and the evidence supporting a contribution claim.
Full briefing
Get the complete briefing
The full version includes the practical framework, the questions to ask and where to start. Request it and we will send it to you, free of charge.
Inside the full briefing
- Distinguish the evaluation questions
- Use a theory of change to expose assumptions
- Make proportionality a design choice
- Questions for commissioners
- Where to start