Measuring change and explaining what caused it are different tasks. Knowing the distinction helps you commission the evidence a programme actually needs.
A baseline establishes a reference point
A baseline describes conditions at a defined starting point. Its usefulness depends on consistent indicators, a clear population and sufficient documentation for later comparison. A baseline dataset without definitions or collection details can be difficult to use when the programme reaches its next reporting stage.
Agree which outcomes matter and how they will be measured. Record who is included, the period covered, the collection method and relevant contextual conditions. If implementation has already begun, document that timing instead of presenting the study as a pre-intervention measure.
An endline measures conditions at a later stage
An endline can show whether measured conditions differ from the baseline. The comparison needs care if respondents, definitions, geography, instruments or collection seasons changed. Some apparent differences may reflect the measurement process rather than a change in the underlying outcome.
Where the same individuals are followed, investigate attrition. Where separate samples are used, examine whether their composition is comparable. A clear report explains how the comparison was constructed and which limitations remain.
Impact evaluation asks a causal question
The causal question is what would have happened without the intervention. The World Bank’s Impact Evaluation in Practice explains why this counterfactual matters. Before-and-after figures alone cannot rule out other explanations for change.
A suitable design might use random assignment or a defensible quasi-experimental comparison, depending on how the intervention is implemented and what data exists. The design must be justified for the context; choosing a sophisticated model after collection cannot automatically repair a weak comparison.
Evaluation criteria are a separate set of lenses
The OECD-DAC criteria cover relevance, coherence, effectiveness, efficiency, impact and sustainability. They help organize evaluative judgement. Their use should follow the purpose and context of an evaluation, rather than requiring every study to give equal attention to every criterion.
A programme review may examine implementation, stakeholder experience and the plausibility of the results pathway without estimating a causal effect. That can still provide useful evidence if the report accurately describes the question answered and the limits of its design.
Commission the study around the decision
If the programme needs a starting reference, prioritize indicator definitions and a credible baseline. If it needs to understand progress, plan monitoring and a comparable follow-up. If a funding decision depends on causal attribution, assess evaluability and the comparison strategy early enough to influence implementation and data collection.
Also consider whether the priority is to understand how or why something happened. Process evidence and qualitative research may be needed to interpret quantitative outcomes and identify implementation constraints. A mixed-method design is useful when the components answer connected questions, not simply because it includes more methods.
Request an evidence map before fieldwork
Ask the evaluator to map each question to a source, collection method, analysis and expected limitation. Identify what would count as evidence against the proposed explanation as well as evidence in its favour. Agree how conflicting findings will be handled.
A useful final report makes its reasoning visible. It distinguishes observed results, explanations supported by the design and issues that remain uncertain. That distinction helps programme teams act without asking the evidence to carry a claim it cannot support.
Sources & further reading
Sources reviewed 10 September 2026. The practical scoping recommendations in this article are Descriptive’s editorial synthesis; referenced guidance is not a claim of affiliation.