The Gold Standard of Anthropology | Exit Interviews, Nonresponse Reasons and the Ethics of Explaining Withdrawal

1. Begin with the right to leave without explaining

Back to contents

Withdrawal is a decision about participation, not an obligation to supply the project with a narrative. A reviewer may offer a reason, choose a broad category, speak through a representative or say nothing. The project should make those options explicit before problems arise. It should also explain what happens to comments, votes, recordings and identifiable quotations already collected. An exit interview is therefore a separate invitation requiring its own voluntary agreement. The person’s remaining payment, reimbursement, credit or access to services must not depend on accepting it. This distinction preserves dignity and improves the evidence: an answer given under pressure is not a reliable account of why someone left.

2. Distinguish absence, pause, nonresponse and withdrawal

Back to contents

A missed meeting is not automatically withdrawal. Neither is a delayed reply, a conflict-specific recusal or a temporary pause. Define operational states before analysing attrition: active participation, planned absence, pause with a review date, recusal from a named decision, unreachable after a stated contact sequence, and confirmed withdrawal. “Nonresponse” describes what the project observed; it does not reveal intention. A person who stops replying may be busy, unsafe, unwell, disconnected from the channel or deliberately ending contact. Keep the observed status separate from any inferred reason, and allow the person to correct the record later without having to re-enter the panel.

3. Design an exit invitation that can be declined

Back to contents

Use one calm invitation after the person has been told they may leave. Offer several routes: a short form, a confidential conversation, a support person, an anonymous channel or no response. State who will see the answer, how it will be summarised, whether quotations could be used and when the record will be deleted or de-identified. Avoid repeated “just checking” messages that function as refusal conversion. A single reminder may be appropriate when it was promised, but silence after the defined sequence should close the request. The script should thank the person, confirm outstanding payment and make clear that future opportunities are unaffected.

4. Separate prior contributions from the exit account

Back to contents

A reviewer may permit earlier contributions to remain while declining an exit interview, or may ask whether identifiable material can be removed. The answer depends on the original agreement, the project stage and what can still practicably be changed. Explain those limits without overstating legal certainty. Do not use the exit conversation to obtain retroactive permission for uses that were never discussed. Store the explanation with tighter access than ordinary attendance data, because it may contain health, employment, safeguarding or conflict information. Public reports should usually aggregate process lessons rather than narrate an identifiable person’s departure.

5. Code reasons without inventing motives

Back to contents

Create a small, pre-specified reason framework that allows multiple selections and an “another reason” option: time or care burden, accessibility, communication, payment, safety, conflict, disagreement with the project, changed circumstances and no reason provided. Keep verbatim notes separate from analytic codes. Two coders can review de-identified summaries when interpretation matters. Never convert “no reply” into “lost interest”, or a broad category into a psychological diagnosis. Report both known reasons and the amount of missing explanation. A transparent unknown category is better evidence than a complete-looking table built from guesses.

6. Worked example: three exits from a ten-person panel

Back to contents

A ten-person neighbourhood panel loses three reviewers. One emails that evening meetings conflict with shift work. One asks to leave and declines further questions. A third stops replying after two unpaid preparation requests. The project records one stated scheduling reason, one confirmed withdrawal with no reason, and one unresolved nonresponse. It does not label all three as disengaged. Payment records show that reimbursements were late for everyone, so the team clears arrears, shortens preparation, adds asynchronous review and offers daytime sessions. The report states that burden may have contributed to the pattern but does not claim to know the third person’s motive.

7. Analyse patterns without profiling individuals

Back to contents

The useful unit of inquiry is often the process pattern: when departures occurred, which tasks preceded them, whether payment or access failures clustered, and which roles became under-represented. Compare participation stages and modes, not personal worth. Small panels rarely support formal causal estimates, and subgroup percentages can identify people. Use minimum reporting thresholds, combine sensitive categories and involve community governance in deciding what may be shared. A pattern can justify a repair even when individual reasons remain private. Conversely, one vivid exit account should not be treated as the explanation for everyone who left.

8. Changed-context transfer check: a citizens’ jury

Back to contents

A citizens’ jury asks participants to deliberate on a transport plan. Two jurors withdraw after a hostile public meeting, while another becomes unreachable. The transferable method is to protect exit, distinguish known from unknown reasons and audit the conditions around the departures. The new context adds public visibility and possible intimidation, so confidentiality and safety planning matter more than obtaining a detailed interview. The jury should report how membership changed and whether the remaining group can still support the same representativeness claim. It may need to pause, recruit through the original route or narrow its conclusion rather than pressure former jurors to explain.

9. Know the limits and common misconceptions

Back to contents

An exit interview is not a neutral measurement instrument. Timing, interviewer identity, power, memory and fear of consequences shape what is said. A polite reason may protect relationships; several reasons may coexist; the final incident may not be the main cause. Nonresponse is not consent, proof of dissatisfaction or evidence that the person had nothing to add. High retention does not prove an ethical process if people felt unable to leave. Low retention does not by itself prove harm. Treat explanations as situated accounts, triangulate with workload and process records, and keep uncertainty visible.

10. Audit contact, consent, burden and repair

Back to contents

Retain the approved exit script, contact sequence, response channel, access controls, payment status, participation timeline, reason codes, de-identification decisions and actions taken. Check whether reminders exceeded the stated limit, whether staff with authority over payment conducted interviews, and whether sensitive accounts appeared in ordinary project files. Compare promised and actual workload before each departure. Record repairs and whether they reduced later burden. An independent community reviewer should be able to see how the project learned from attrition without being able to reconstruct private stories.

11. Frequently asked questions

Back to contents

Must someone give a reason to withdraw? No. May the project ask once? Yes, if the invitation is clearly optional and does not affect owed benefits. Is silence a refusal? It is nonresponse after the defined contact sequence; do not infer more. Can exit comments be quoted? Only with specific permission and careful assessment of identifiability. Should every reason be published? Usually not; aggregate process lessons are safer. Can a project contact someone again later? Only within the consented contact plan or after a clearly separate invitation that respects the earlier boundary.

12. The gold standard makes uncertainty ethical

Back to contents

A defensible attrition record does not fill every blank. It protects the right to leave, offers a low-pressure way to explain, honours prior agreements and separates observations from interpretations. It then asks what the project can repair: workload, payment, access, safety, communication or decision rules. The gold standard is not a perfect list of motives. It is an auditable account of who participated at each stage, what is known about changing participation, what remains unknown and how the process responded without making continued disclosure the price of departure.