Research methods · Expert interviews · Decision making
How do you use expert interviews without overstating the evidence?
Match experience to the research question, preserve the context of each answer, and trace corroboration to its original source before drawing a broader conclusion.
Recruit for the question you need to answer
Define the claim you want to investigate before choosing interviewees. Understanding a purchasing decision, a technical workflow and a service failure may require different kinds of direct experience. Ask what someone actually worked on, when they did it and which parts they could observe.
Use screening to establish relevance and identify gaps in the research plan. Treat a short screening answer as an initial lead that needs examination. A senior title can help explain someone's perspective while leaving their exposure to a particular tool or process uncertain.
Prepare open questions around concrete situations. GOV.UK's interview guidance recommends asking for real examples and using follow-up questions to understand them. For a workflow comparison, ask which alternatives the person used, under what conditions and what happened when the ordinary process broke down.
Keep the answer connected to its limits
Capture the distinction between something the participant experienced, something they heard from others and their interpretation of why it happened. Each can be useful when its basis remains visible. Follow up on broad statements with a specific example and the boundaries of the person's involvement.
Keep the researcher's interpretation separate from the participant's account. A discussion guide also needs its own status: a question that was prepared but never answered remains an open question. Do the same for topics the participant could not assess or chose not to discuss.
The Government Social Research qualitative-evaluation framework emphasizes context, the basis for conclusions and missing perspectives. Apply that discipline to a short interview summary: retain the qualification that matters, even when it makes the headline less convenient. Record disagreement where it changes the decision.
Trace agreement back to independent evidence
Before treating repeated claims as corroboration, identify their origins. An interview, its transcript and a slide summarizing it all trace to one conversation. Two interviewees might also be repeating the same announcement. Count the underlying observations and inspect what each actually adds.
Seek a different kind of evidence for the claims that matter most. A process description may be checked against a demonstration or an available work artifact. A market-wide claim needs evidence beyond one person's exposure. Choose the next check to address the specific uncertainty.
Finish with a bounded conclusion: what the interviews support, where perspectives differ and what remains unknown. Use that conclusion to guide the next decision or research step. The value of expert judgment increases when a reader can see how far it reaches.