Blog › Guide
How to design a baseline survey for an NGO project
A baseline survey records the situation before a project starts. Later, you compare against it to show what changed. Get it wrong and every later report is weaker. Here is a practical way to get it right.
1. Start from your indicators
Do not write questions first. List the project's expected results and the indicators that will measure them. Every survey question should feed at least one indicator. If a question serves none, cut it.
2. Define who and where
Be exact about the target population: which communities, which households or people, which age groups. Decide whether you will compare with a group outside the project. A comparison group makes later claims about impact far stronger.
3. Choose a sensible sample
Decide how many respondents you need and how you will select them. Random selection within clear groups (for example, by community or by sex) reduces bias. Work out the sample size before fieldwork, not after, and allow for people you cannot reach.
4. Write clear questions
- Use simple words and one idea per question.
- Avoid leading questions that suggest an answer.
- Give answer options that do not overlap.
- Translate into the languages respondents use, then translate back to check meaning.
- Keep it short. Tired respondents give poor answers.
5. Pilot the tool
Test the questionnaire with a small group similar to your respondents. Watch for confusing questions, missing options and how long it takes. Fix the tool before full fieldwork.
6. Train enumerators and protect respondents
Train data collectors on every question, on consent and on how to stay neutral. Explain the purpose to respondents, ask for consent and keep personal data secure.
7. Check data as it comes in
Review data daily during fieldwork. Look for missing answers, impossible values and unusually fast interviews. Problems caught early can be fixed in the field.
8. Analyse and report the baseline values
Clean and code the data, then calculate a starting value for each indicator. Report the method, the sample and the limitations honestly. Store the tool and dataset so the endline survey can repeat the same questions in the same way.
Quick checklist
- Every question maps to an indicator
- Population and sample defined in advance
- Tool translated and piloted
- Enumerators trained; consent obtained
- Data checked daily
- Baseline values, method and limits documented
Need help with this? Message us on WhatsApp or see our services.