Member Checking in Qualitative Research: How to Validate Findings With Participants

The short answer

Member checking (also called respondent validation or participant validation) means taking your data or your interpretation back to the people you studied and asking: does this match your experience? It is the classic way to test the credibility of qualitative findings. Lincoln and Guba, who set out the trustworthiness criteria most qualitative researchers still use, called it "the most crucial technique for establishing credibility" (Naturalistic Inquiry, 1985).

It is also one of the most debated techniques in qualitative research. A narrative review of 44 journal articles that used member checks concluded: "no evidence was found to support the view that member checks enhance research quality" (Thomas, Qualitative Research in Psychology, 2017). Both views are useful. Member checking will not prove your findings are true, but done well it catches misreadings, surfaces what you missed, and gives participants a voice in how they are represented.

This guide covers what member checking is, the main types, a step-by-step process, how to handle disagreement, and how to run a member check at a scale that used to be impractical.

What member checking is (and is not)

Member checking sits under credibility, the qualitative counterpart to internal validity. In Lincoln and Guba's framework it is one of several credibility techniques, alongside prolonged engagement, persistent observation, triangulation and peer debriefing. You return something to participants and invite a response:

• Raw data: a transcript or recording of their own interview, to check accuracy
• Interpreted data: a summary of what you think they said
• Synthesized findings: the themes you built across all participants, to see whether they recognise themselves in the whole

What it is not:

• Not a vote. If three participants disagree with a theme, the theme is not automatically wrong. Participants see their own experience; you see the pattern across 20 people.
• Not a substitute for analysis. Sending transcripts back for typo correction is a transcription check, not a credibility check.
• Not proof. Agreement can come from politeness, fatigue or not reading closely. Disagreement can come from a participant whose view changed since the interview.

Thomas (2017) draws the main line between approaches that return raw data and those that return analysed findings. The second is harder and far more informative.

The four main types of member checking

Synthesized member checking

The best-documented structured method is Synthesized Member Checking, developed by Linda Birt and colleagues (Qualitative Health Research, 2016) in a study with people diagnosed with melanoma. Several months after their interviews, participants received a document of synthesized findings with illustrative quotes and were invited to comment, add to or disagree with them. 56% (n=28) returned the document. Respondents were broadly similar in gender, age and melanoma type to the main sample, but younger participants were less likely to reply.

Two lessons from that study apply to any team:

1. Expect a little over half of participants to respond, at best, and plan for it.
2. The people who respond are not a random subset. Check who is missing before you treat silence as agreement.

Asynchronous and video member checks

Member checks used to rely on paper summaries or follow-up meetings. Schafer and Phillippi (International Journal of Qualitative Methods, 2025) describe an exemplar with 25 participants where findings were presented in a recorded video with slides and responses were collected through an online survey. They argue that asynchronous technology can raise engagement and equity in member checking, because participants respond when it suits them rather than when the researcher is free.

How to run a member check, step by step

Step 1: Decide what you are testing

Write one sentence: "We want participants to tell us whether ___." Common goals:

• Did we represent each person's experience accurately?
• Do the themes resonate with people who lived them?
• Is anything important missing?
• Are any of our interpretations wrong or harmful?

Your goal decides what you return. Accuracy needs transcripts or summaries. Resonance and gaps need synthesized findings.

Step 2: Plan it before the first interview

Tell participants at consent time that you may contact them again with findings. This sets expectations, lets people opt out early, and avoids a cold re-contact months later. Record how each person prefers to be contacted.

Step 3: Prepare findings people can actually read

Most member checks fail here. Academic theme names ("negotiating identity under uncertainty") mean little to participants. For each finding:

• Write it in plain language, one or two sentences
• Add one or two short, anonymised quotes
• Say who it applies to ("most people we spoke to", "people who switched from a competitor")
• Keep the full set to five to eight findings

Remove anything that could identify another participant. Synthesized findings mix many voices, so de-identification matters more here than anywhere else in the study.

Step 4: Ask structured and open questions about each finding

Give people a fast way to react and room to explain. For each finding:

1. A rating: "How well does this match your experience?" on a 1 to 5 scale
2. An open question: "What, if anything, is wrong or missing?"

End with one overall question: "Is there anything important about your experience that these findings do not capture?"

The rating lets you see patterns across responses. The open answer is where the real value is.

Step 5: Time it well

Birt and colleagues went back several months after the interviews. Longer gaps let participants reflect, but memories fade and situations change. In fast-moving product research, two to six weeks is usually a better balance. Whatever you choose, record it, because it affects how you read disagreement.

Step 6: Analyse responses as data

Treat member-check responses as a new, small dataset:

• Confirmations: note them, but weigh them lightly
• Corrections of fact: fix the finding
• Disagreements of interpretation: investigate (see below)
• New material: decide whether it extends a theme or warrants a new one

Step 7: Report what you did and what changed

Thomas (2017) found that published studies rarely say how member checks were done or what changed as a result. Close that gap in your own report: who you contacted, how many responded, what you asked, and which findings changed.

What to do when participants disagree

Disagreement is the most useful outcome of a member check, and the hardest to handle. Work through it in order:

1. Is it a factual error? You misheard, misquoted or misattributed. Fix it.
2. Is it about representation? The participant agrees with the pattern but not with being placed in it. Check your coding of their interview.
3. Is it a different interpretation? The participant sees the same events differently. Look across the dataset. If others support your reading, keep it and report the dissent. If the participant's reading fits the data better, revise.
4. Has their situation changed? A churned customer who has since returned may now remember things differently. Note the change; it may be a finding in itself.

Never quietly drop a theme because one person objected, and never ignore a disagreement because most people agreed. Report both.

Common mistakes

• Sending a 30-page report. Nobody reads it. Five to eight plain-language findings get real responses.
• Treating silence as agreement. Non-response is missing data, not consent.
• Only checking with friendly participants. The people most likely to disagree are often the ones you most need to hear from.
• Asking leading questions. "Does this capture your experience well?" invites a yes. Ask what is wrong or missing.
• Over-correcting. Participants are experts in their own experience, not in the pattern across the sample.
• Breaking confidentiality. A distinctive quote can identify someone to a colleague who also took part.

When member checking is worth it

Member checking costs time for you and for participants. It is most worth it when:

• Findings will drive a significant decision (pricing, a market entry, a product cut)
• You are studying a group you are not part of, where misreading is likely
• Participants are vulnerable or the topic is sensitive
• Stakeholders are sceptical and you need to show the findings hold up

It matters less for quick usability checks, or when you are triangulating across several methods anyway. In those cases triangulation and peer review may give you more for the same effort.

How Koji helps

A synthesized member check is a small follow-up study: you return findings to the same people and ask structured and open questions about each. That used to mean a second round of scheduling, emails and manual analysis. In Koji it is a second study on the same workflow as the first.

• Build the check as a short study. Create a new study for the member check. You can attach your findings summary as a context document while you build the brief, so the assistant drafts questions from it. Put each finding into the brief as a question.
• Mix structured and open questions. Use a scale question for "How well does this match your experience?" and an open-ended question for "What is wrong or missing?" Koji supports six question types (open-ended, scale, single choice, multiple choice, ranking and yes/no), and aggregates the structured answers into per-question charts automatically. See the structured questions guide.
• Probe disagreement automatically. When a participant says a finding does not fit, the AI interviewer asks follow-up questions, up to three per question depending on the depth you set. This is the part a paper or survey member check cannot do: you learn why someone disagrees, not just that they do.
• Go back to the same people. Send each original participant a personalized interview link so responses are tied to a known person, and you can see who has and has not replied. On Interviews and Enterprise plans you can also import a contact list.
• Let people respond when it suits them. Interviews are asynchronous, by text on every plan or by voice on Interviews and Enterprise. This matches the asynchronous approach Schafer and Phillippi recommend.
• See patterns across responses. Koji themes the open answers across all respondents, and you can ask questions of the full set of conversations with Ask AI. That shows which findings held up, which were challenged, and by whom.
• Pay only for useful responses. Each interview gets a quality score, and only interviews scoring 3 or more use credits or enter the report. A text interview is 1 credit, a voice interview 3.

Compared with a paper or email member check, you get structured ratings and probed open answers in one pass. Compared with follow-up interviews, you skip the scheduling. Neither replaces your judgement: deciding what a disagreement means is still the researcher's job.

Related Resources

• Qualitative Research Validity: Lincoln and Guba's trustworthiness criteria in practice
• Triangulation in Research: combining methods to strengthen credibility
• Thematic Analysis Guide: building the themes you will member-check
• Inter-Rater Reliability in Qualitative Research: checking coding consistency inside the team
• Structured Questions Guide: scale and open-ended questions for your member-check study
• Personalized Interview Links: inviting original participants back

Sources

• Lincoln, Y. S., & Guba, E. G. (1985). Naturalistic Inquiry. Sage.
• Thomas, D. R. (2017). Feedback from research participants: Are member checks useful in qualitative research? Qualitative Research in Psychology, 14(1), 23–41. https://doi.org/10.1080/14780887.2016.1219435
• Birt, L., Scott, S., Cavers, D., Campbell, C., & Walter, F. (2016). Member checking: A tool to enhance trustworthiness or merely a nod to validation? Qualitative Health Research, 26(13), 1802–1811. https://doi.org/10.1177/1049732316654870
• Schafer, R., & Phillippi, J. C. (2025). Updating and advancing member-checking methods: Use of video and asynchronous technology to optimize participant engagement. International Journal of Qualitative Methods, 24. https://doi.org/10.1177/16094069251315395