Choose the method based on the type of knowledge needed
Focus group discussions are useful for exploring shared norms, community experiences and areas of agreement or disagreement. Key informant interviews are better when the study needs institutional, technical or implementation knowledge from people with specific roles. In-depth interviews are useful for personal experiences, sensitive topics or detailed life histories.
Using all three methods does not automatically make a study stronger. Each method should have a clear analytical purpose and contribute evidence that another method cannot provide as effectively.
Sampling should seek information richness, not statistical representation
Qualitative sampling is generally purposive. Participants are selected because their experience, role or perspective can illuminate the research question. Diversity is often more important than large numbers: gender, age, livelihood, geography, programme exposure, institutional role or vulnerability may all shape experience.
The sampling rationale should explain which perspectives are needed and how the team will judge whether the evidence is sufficiently rich. Saturation can be useful, but it should not become a mechanical claim that every topic requires the same number of interviews.
Facilitator skill determines data depth
A discussion guide is only a framework. Skilled facilitators listen closely, probe for examples, separate personal experience from assumptions, invite quieter participants and manage dominant voices without turning the session into an interrogation.
For sensitive subjects, facilitator gender, language, age and familiarity with local norms may affect what participants are willing to discuss. These decisions should be considered during team formation, not after recruitment.
Language quality extends beyond translation
Qualitative meaning can be lost when local expressions are translated too literally. Field teams should agree on key concepts, use facilitators who understand the relevant context and document terms that do not have simple English equivalents.
When recordings are transcribed and translated, quality checks should focus on meaning, not only grammar. Analysts need enough context to distinguish what participants actually said from the translator’s interpretation.
Analysis should be systematic and traceable
Thematic analysis should move beyond selecting memorable quotations. Analysts should code or organise evidence systematically, compare patterns across participant groups, identify divergent cases and link interpretations back to the research questions.
Quotations are most useful when they illustrate a finding that has already been established through broader analysis. They should not stand alone as evidence for claims that the wider dataset does not support.
Ethics and safe participation are central
Researchers should consider privacy, group composition, recording consent, potential distress and the risk of disclosing information in front of others. In an FGD, confidentiality cannot be guaranteed in the same way as a private interview, so participants should understand the limits of the setting.
Sensitive topics may require separate groups, private IDIs, referral information or additional facilitator training. Methodological depth should never come at the expense of participant safety.
Group composition affects what people will say
FGDs work best when participants can speak without strong social pressure from others in the room. Mixing supervisors with staff, leaders with beneficiaries or older participants with younger participants may suppress disagreement. Gender, status, disability, livelihood and programme role can all influence whether a group feels safe enough for open discussion.
Homogeneous groups are not always required, but the composition should be intentional. Researchers should explain why participants are grouped together and what perspectives may still need to be explored through separate interviews.
Build a source-management system before analysis
Qualitative projects quickly generate recordings, notes, transcripts, translations and coding files. A naming convention should link each source to method, location, participant category and date or sequence while protecting identity. This makes it easier to track evidence during analysis and quality review.
Source management also supports accurate quotation. Analysts should be able to trace a quoted passage back to the relevant transcript or recording and confirm that the wording, translation and participant category are correct before publication.