Baseline, midline and endline studies are the backbone of most donor-funded evaluation frameworks. Done well, they show whether a programme is changing lives and give managers the evidence to adjust course. Done badly, they consume budget, frustrate respondents and produce numbers nobody trusts. The difference lies mostly in decisions made before a single questionnaire is administered.
What each study is for
|
Study |
When |
Purpose |
|
Baseline |
Before or at the very start of implementation |
Establish the starting position on key indicators, inform realistic targets and sharpen programme design. |
|
Midline |
Around the midpoint of the programme |
Test whether the programme is on track, identify what is and is not working and enable course correction. |
|
Endline |
At or near the end of the programme |
Measure results against the baseline and targets and generate lessons for the next phase. |
The three studies are designed as a set. Each one is only as useful as its comparability with the others, which is why the design decisions below matter so much.
Design decisions that make or break a study
- Start with the indicators. The questions you ask should flow from the programme’s results framework, not the other way round. If a question does not feed an indicator or a learning question, leave it out.
- Decide who to survey, how many and how they will be selected. The sample must be large enough to detect the change you expect and representative of the population you want to talk about.
- Where feasible, include a comparison group so that you can separate the effect of the programme from changes that would have happened anyway.
- Panel or repeated cross-section. Following the same respondents over time gives sharper evidence of change but requires tracking and allows for attrition. Fresh samples each round are simpler but less precise on individual change.
- Use the same definitions, questions, sampling frame and methods in every round, so that results can be compared.
Fieldwork quality
Even the best design fails if fieldwork is weak. The essentials are well-trained enumerators who understand the purpose of the study and every question; a pilot test of the tools; digital data collection with built-in validation checks; active field supervision; and back-checks, in which a supervisor re-contacts a sample of respondents to verify that interviews took place and responses were recorded correctly.
Ethics matter as much as accuracy. Obtain informed consent, explain how the data will be used, protect confidentiality and comply with the Kenya Data Protection Act, 2019 when handling personal data. Take particular care with children, survivors of violence and other vulnerable groups, and make sure referral information is available if sensitive issues arise during interviews.
Analysis and reporting
A useful report goes beyond averages. Disaggregate results by sex, age, disability status and location so that gaps between groups are visible. Compare findings with baseline values and targets. State confidence levels and limitations plainly, including non-response and attrition.
Most importantly, finish with conclusions and recommendations that the programme team can act on, and make sure there is a session to discuss them with the people who run the programme, rather than only a report to file. Findings that are presented and debated change practice. Findings that are merely delivered rarely do.
Learning, not just compliance
Donors require these studies for accountability, but the best programme teams use them for learning. A midline, in particular, is an opportunity to ask why results are ahead of or behind expectations, to revisit the assumptions in the theory of change and to adjust activities while there is still time. Treating the midline as a box-ticking exercise wastes the most valuable moment in the programme cycle.
Common mistakes
- Commissioning the baseline after activities have already begun, which weakens the starting picture.
- Changing indicator definitions or questions between rounds.
- Using samples too small to detect the change the programme expects to achieve.
- Ignoring attrition in panel studies until the endline.
- Treating the midline as optional and leaving no budget for it.
- Producing a long report that nobody reads or uses.
Plan time and budget early
Evaluation studies take longer than most people expect. Tools must be designed and tested, enumerators recruited and trained, approvals obtained and data cleaned before analysis can start. Build realistic time into the workplan and ring-fence the budget for all three rounds at proposal stage, so that the endline is not squeezed out when other costs rise.
How Melanel can help
Melanel designs the sampling and tools, manages fieldwork including enumerator recruitment and training and digital data collection, and delivers findings tied directly to your indicators. We hand over the dataset and the method, and where you wish, we build your team’s skills so that they can run the next round themselves.
