Choosing an M&E method: start from the evaluation question
Questionnaires, interviews, quantitative surveys, focus groups, document review. Faced with the diversity of available methods, methodological choice quickly becomes a question of tools. Yet in evaluation, the first question to ask is not “how will we collect the data?” but “what do we need to understand?”
A project can have numerous indicators, produce regular reports and collect a considerable amount of data without actually being able to answer the essential questions about its results.
This is why the choice of an M&E method must start from the evaluation question, and not from the tool the team already masters or intends to use.
This sequence makes it possible to build a design that is coherent and proportionate to the purpose of the evaluation.
The evaluation question determines the type of information sought
Not all evaluation questions seek to produce the same type of information. Take three questions.
“To what extent has programme coverage evolved between 2024 and 2026?”
Here we need to measure an evolution, compare values and possibly analyze gaps between groups or territories. A quantitative approach will probably be central.
“Why do certain categories of beneficiaries use the service less?”
This question no longer concerns magnitude, but mechanisms, perceptions and constraints as they are experienced. Figures alone will not answer it.
“To what extent did the project contribute to the observed improvement, and which factors facilitated or limited that contribution?”
This last question concerns the observed change, the contribution of the intervention and the conditions under which that contribution occurred, all at once. Several data sources and methods may then be necessary.
Several broad intentions can therefore be distinguished: describing a situation, measuring the magnitude or evolution of a phenomenon, comparing groups, periods or territories, explaining a result or a gap, understanding mechanisms or experiences, examining the contribution of an intervention to an observed change, or forming a judgement on the performance or relevance of an intervention.
This distinction is the first level of methodological reasoning.
A simple grid to guide the choice
Each evaluation intention corresponds to a dominant approach and to collection methods that are generally suited to it. The grid below makes these correspondences visible.
Quantitative, qualitative, mixed: what each approach allows
Behind these correspondences lie three distinct logics, each with its own zone of relevance and its own limits.
The quantitative approach: measuring without necessarily explaining
The quantitative approach is particularly suited when the evaluation needs to measure levels, proportions, gaps or trends. What share of beneficiaries adopted a new practice? What is the coverage level of the service? What evolution can be observed between two periods? Is there a difference between two groups?
Quantitative data make it possible to standardize measurement, compare populations and analyze certain variations. But they do not necessarily explain those variations. An indicator may show that use of a service has declined without establishing whether the decline relates to cost, accessibility, service quality, existing practices or a shift in context.
Measuring a change and explaining that change are two different questions.
The qualitative approach: understanding mechanisms and context
The qualitative approach is particularly relevant when the evaluation seeks to understand the perceptions, experiences, behaviors or mechanisms associated with a result. It can draw on individual interviews, key informant interviews, focus groups, observation, document review or case studies.
It is especially relevant when the question concerns the “why”, the “how” or the conditions under which a change occurred. Its value lies not in the number of people interviewed, but in the depth and relevance of the information gathered.
It nevertheless demands a specific rigor: purposive selection of participants, a coherent protocol, traceability of the analysis and explicit acknowledgement of limits. Qualitative findings are not automatically generalizable in statistical terms to an entire population. They can however be essential for understanding the mechanisms and contextual factors that indicators cannot capture.
The mixed approach: articulating data rather than adding it up
Some questions require combining both approaches. Imagine a project supporting producers in adopting a new practice. A survey shows that 42% of the producers interviewed have adopted the practice, against a target of 70%. The project now knows the size of the gap. But one essential question remains: why does adoption stay below the objective?
Interviews can then explore the factors likely to explain the situation: cost, availability of equipment, accessibility, existing practices, perception of the new technique or context-related constraints.
The two approaches therefore provide different information: the quantitative measures the magnitude of the phenomenon, the qualitative helps understand its mechanisms and explanatory factors. A mixed approach does not, however, consist in mechanically adding a survey and a few interviews. The methods must be designed together and their findings articulated in the analysis. The aim is not to accumulate data, but to produce the evidence needed to answer the question.
From a simple tool to an evaluation design: do not confuse the levels
Another source of confusion comes from the fact that the terms “tool”, “method” and “approach” are often used as if they designated the same thing. Yet they operate at different levels.
These elements do not serve the same function. An evaluation may use a theory of change to structure its reasoning, a survey and interviews to produce the data, then several analysis methods to interpret it.
It is therefore more useful to think in terms of evaluation design than to look for a single method that would, by principle, be the best.
Logical framework and theory of change structure the reasoning
The choice of method must also remain coherent with the logic of the intervention. The logical framework, the theory of change or the results matrix represent the links between activities, outputs, outcomes and expected changes.
They help determine what must be observed, at which level of change, which data are needed, which assumptions must be examined and which evidence is expected. This articulation assumes, however, that the logical framework itself stays up to date, which brings us back to the question of its evolution during implementation.
The distinction between levels of results matters here. The fact that an activity was carried out does not necessarily mean the expected output was obtained. Likewise, achieving an output is not enough to demonstrate that a change occurred among beneficiaries. And observing a change does not automatically allow the conclusion that the change results from the intervention. This is exactly the limit revealed by reading achievement rates in isolation.
The more an evaluation question concerns the upper levels of the results chain, the more explicit the methodological reasoning must be.
The evaluation matrix then makes it possible to connect evaluation questions to the indicators or evidence sought, to data sources, to collection methods and to analysis methods. It also helps identify a frequent weakness: a question deemed important at evaluation time, but for which the monitoring system planned no sufficiently relevant data.
The purpose of the evaluation also influences the choice
Methodological choice does not depend on the question alone. It also depends on how the findings will be used.
A formative evaluation may seek to understand quickly the difficulties encountered during implementation in order to adjust an intervention. A summative evaluation may focus more on appreciating the performance or results obtained at a given point. In both cases, several methodological approaches can be mobilized.
What changes is the level of evidence sought, the timeline, the resources available and the nature of the decision to be informed. A question intended to improve an ongoing activity will not require the same design as a question intended to inform a major funding or renewal decision.
Five criteria to examine before choosing a method
Even when the question is clear, several criteria must be examined before settling on a method.
1. The nature of the question
Are we seeking to measure, explain, compare, understand or examine a contribution?
2. The level of evidence sought
Does the decision require a simple description of the situation, an in-depth understanding, or elements allowing a more rigorous examination of a causal relationship or a contribution?
3. Availability and quality of data
Are existing data complete, reliable and comparable enough? An available dataset is not necessarily a usable dataset for every analysis.
4. Context and resources
Access to beneficiaries, institutional constraints, sensitivity of the subject, team skills, budget and timeline all shape what design is possible.
5. Intended use of the findings
Who will use the findings? For which decision? At what moment? This question often determines the right level of complexity.
A relevant method is first an aligned method
The choice of an M&E method should therefore not begin with a questionnaire, a platform or a collection technique. It should begin with a series of questions: what do we want to know? What must we observe in order to answer? What level of evidence is needed? Which data do we already have? Which data must we produce? And above all, which decision should this information inform?
This logic helps avoid two frequent pitfalls: collecting data that answers no important question, or asking evaluation questions the monitoring system is unable to answer.
In development projects, this coherence must also extend into day-to-day data management. Teams must be able to retrieve information, track indicators, consolidate data from different sources and work from a structured view that feeds analysis and steering.
A design that follows the question, not the other way round
Delta Monitoring structures indicators, consolidates data from different sources and connects each piece of evidence to the question it is meant to inform. Teams keep traceability between evaluation questions, data collected and decisions taken, whatever the methodological approach chosen.
A method is never good or bad in the absolute. It is aligned, or it is not. A modest design coherent with the question asked will produce more useful material than an ambitious design built without knowing what it seeks to establish.
This is where the maturity of an M&E system plays out: not in the sophistication of the techniques used, but in the capacity to connect every piece of data collected to a question that matters, and every question to a decision awaiting an answer.