Surveys decision-makers trust: sampling, design and field quality

Household and facility surveys can provide information that routine systems cannot: coverage of services, out-of-pocket spending, quality of care, and the views of people who never reach a clinic. When they are well done, they shape national policy for years. When they are not, they waste money and mislead.

Credibility comes from decisions made long before the first interview.

Start with the questions

What decisions will this survey inform? List the indicators needed, the level at which they must be reliable (national, province, district) and the groups to be compared. Those answers drive sample size, cost and timeline. A survey designed to answer everything usually answers nothing well.

Get the sampling right

• Sampling frame. The list from which respondents are drawn should cover the population of interest as completely and as recently as possible.
• Sample size. Calculate it from the precision required for key indicators, allowing for design effects when using clusters and for non-response.
• Selection. Use clear, documented rules for choosing areas, households and respondents so interviewers have no discretion that can introduce bias.
• Weights. If units had unequal chances of selection, analysis must account for it.

Design the instrument carefully

Questions should be clear, neutral and translated into local languages with back-translation checks. Pre-testing with real respondents reveals confusing wording, long sections and awkward flows before they cost thousands of interviews. Keep the questionnaire as short as the purpose allows, because fatigue damages data quality.

Invest in the field team

Fieldwork quality rests on people. Select interviewers carefully, train them thoroughly with practice interviews, and supervise closely. Good practices include:

• daily review of completed questionnaires
• spot checks and back-checks by supervisors
• electronic data collection with built-in validation and GPS or time stamps
• monitoring of response rates and interview durations for each team

Ethics and consent

Respondents must understand the purpose, give informed consent and be assured of confidentiality. Sensitive topics need special training and safeguards.

Analysis and reporting

Document every decision. Report methods, response rates and limitations plainly, and present findings with confidence intervals, not only point estimates. Share data and findings in forms decision-makers can use.

Trust in a survey is earned through transparency at each step, long before the report is published.