Face-to-Face Surveys Kenya: Fieldwork, Sampling and Quality Guide

face-to-face surveys Kenya: evidence for better decisions

face-to-face surveys Kenya should begin with a specific organizational decision. Define who needs the evidence, the geography and population involved, the assumptions at risk, and what finding would change the course of action. For businesses, NGOs, development organizations and public institutions in Kenya, this keeps research focused on decisions rather than data volume.

Key takeaways

Start with the decision; match methods and sampling to the claim; build ethics and quality assurance into fieldwork; segment findings by meaningful groups; and separate evidence, interpretation and recommendations.

In practice, the approach should be proportionate to the decision, transparent about limitations and documented well enough for another reviewer to understand how the evidence was produced.

Evidence requirements

A strong brief identifies what is already known and what requires validation. Evidence may include demand, customer or beneficiary needs, competitors, pricing, access, distribution, service experience, stakeholder perspectives and operational constraints. Kenya is diverse, so findings should name the counties, towns, channels and populations they can credibly represent.

In practice, the approach should be proportionate to the decision, transparent about limitations and documented well enough for another reviewer to understand how the evidence was produced.

Research methods

Quantitative surveys support comparable measurement. Interviews and focus groups explain motivations and barriers. Observation and audits capture real-world execution. Desk research establishes context. Mixed methods can combine measurement and explanation, but every method needs a clear purpose and limitation.

In practice, the approach should be proportionate to the decision, transparent about limitations and documented well enough for another reviewer to understand how the evidence was produced.

Sampling

Define the target population, sampling frame, inclusion criteria and subgroup requirements before fieldwork. A larger sample cannot repair biased selection. If probability sampling is not feasible, report the limitation and avoid implying national representativeness.

In practice, the approach should be proportionate to the decision, transparent about limitations and documented well enough for another reviewer to understand how the evidence was produced.

Instrument design

Use clear, neutral and answerable questions. Avoid double-barrelled wording and undefined terms. Pilot questionnaires, discussion guides and scorecards for comprehension, timing, skip logic and operational feasibility. Document changes after piloting.

In practice, the approach should be proportionate to the decision, transparent about limitations and documented well enough for another reviewer to understand how the evidence was produced.

Quality assurance

Plan researcher training, supervisor review, logical checks, back-checks, duplicate detection and structured exception review as appropriate. Fieldwork deviations should be documented so decision-makers understand whether access, timing or respondent availability affected the evidence.

In practice, the approach should be proportionate to the decision, transparent about limitations and documented well enough for another reviewer to understand how the evidence was produced.

Ethics and privacy

Collect only information necessary for the stated purpose. Projects involving personal data should consider Kenya’s Data Protection Act, 2019 and current Office of the Data Protection Commissioner guidance. Define access, retention and reporting controls before collection begins.

In practice, the approach should be proportionate to the decision, transparent about limitations and documented well enough for another reviewer to understand how the evidence was produced.

Analysis

Move beyond an overall average. Analyse results by relevant geography, channel, segment or stakeholder group. Quantitative findings should show denominators and appropriate uncertainty; qualitative evidence should be coded into themes. Recommendations should be traceable to findings.

In practice, the approach should be proportionate to the decision, transparent about limitations and documented well enough for another reviewer to understand how the evidence was produced.

Decision framework

Common mistakes

Avoid starting with a questionnaire, using convenience samples for national claims, selecting providers only on price, collecting metrics without action owners, ignoring segment differences, changing definitions between waves without documentation, and presenting association as causation.

In practice, the approach should be proportionate to the decision, transparent about limitations and documented well enough for another reviewer to understand how the evidence was produced.

What this means for your organization

Write one sentence stating the decision the study must support. List assumptions that could make the decision fail and design evidence to test them. Decide who will use the report and what format they need. For repeat studies, establish a baseline and stable indicator definitions.

In practice, the approach should be proportionate to the decision, transparent about limitations and documented well enough for another reviewer to understand how the evidence was produced.

Choosing a research partner

Compare proposals on decision understanding, sampling logic, fieldwork assumptions, quality controls, analysis, privacy, limitations and useful deliverables. Ask providers to state what their design cannot conclude and how they will manage fieldwork deviations.

In practice, the approach should be proportionate to the decision, transparent about limitations and documented well enough for another reviewer to understand how the evidence was produced.

Frequently asked questions

How should a face-to-face surveys Kenya project begin?

With the decision, target population, geography and assumptions.

Which method is best?

The method should fit the question and required confidence.

How large should the sample be?

Size follows intended inference and subgroup analysis.

What should the report contain?

Methods, sample, limitations, findings, segmented analysis, recommendations and next steps.

Sources

• Kenya Data Protection Act, 2019
• ODPC Kenya
• KNBS