Evaluating well avoids blind decisions. In social and environmental programs we can analyze potential for change (before executing), measure effects over time (before and after) and estimate controlled effects (comparing with those who did not participate). Each type answers a different question: whether we are on the right track, whether a change occurred, and whether that change is due to the intervention. Here I explain when to use each one and what to expect from your findings.
Potential for change (exploratory assessment)
This level is used before executing or escalating. It serves to align the strategy with the theory of change and validate key hypotheses: population, problem, mechanisms and risks.
Improve the design to avoid investing blindly.
Typical methods: evidence review, interviews with actors, co-creation with users, small-scale pilots and indicator tests. Deliverables: prioritized problems, critical assumptions and definition of outputs, results and impacts.
Tip: Use it when there is pressure to “start now” but the path is not clear. A week of exploration can save you months of rework.
Effects over time (before and after)
It seeks to answer if there were changes in those who participated between the beginning and the end, here the official definition applies: Evaluate over time, that is, measuring the beginning and end of an intervention.
Evidence of progress, not causality.
It requires a baseline, monitoring of outputs and results, and consistent instruments at both times. It is ideal for ongoing programs, with clear coverage and defined cycles. Limitation: without a comparison group, we cannot affirm that the change is exclusively due to the intervention.
If you don't have a baseline, no problem! There is an option B: the retrospective baseline. It consists of collecting at the end people's perception of how they were before and after, and what part of the change they attribute to the program (self-attribution). It is useful for reconstructing trends when you are late and is ideal for ongoing programs, with clear coverage and defined cycles.
Controlled effects (quasi-experimental)
The before-after evaluation includes a group of people very similar to that of the program (age, context, initial conditions), but who do not receive the intervention in that period; By comparing how both groups change, we can estimate how much of the change is due to the program and how much would have happened anyway. This is what we call evaluating with control, in other words, it is measuring the beginning and end of an intervention, and what would have happened if it did not occur.
It allows the change to be attributed with greater rigor.
You can use statistical matching, differences-in-differences, or eligibility designs. It requires more ethical and operational planning, it is most useful when you need to demonstrate shared value to funders or scale public policies.
How to choose?
| Type | Key question | When to use | Expected evidence | Technical requirement | Findings |
|---|---|---|---|---|---|
| Potential for change | Does the change strategy make sense? | Before booting or scaling | Clear risks, hypotheses and metrics | Low | Preliminaries: Theory of Change validated and characterization |
| Effects over time | What changed in the participants? | Make agile decisions based on evidence | Differences between baseline and closure | Medium | Agile and rigorous: magnitude of the observed change |
| Controlled effects | Is the change due to the intervention? | When attribution is required | Valid comparison with non-participants | High | Detailed: estimate of attributable effect (counterfactual) and effect size |
Frequently asked questions
Conclusion
Choosing the type of evaluation is a matter of intelligent decisions: first validate the route, then check the progress and, when necessary, attribute the change with control. This way you protect resources, improve the intervention and demonstrate social impact -change the lives of others- with clear evidence. Let's talk and put together the plan your program needs.