Interviews produce accounts, not direct records of behavior
Aliases: interview account · self-report data · saying-doing gap
What it is
Interviews produce accounts of experience, meaning, reasons, and expectations, not direct behavioral records. An account can accurately express a person's understanding and identity presentation without establishing that an action occurred, how often it occurred, or whether the stated reason caused it.
Why it happens
Answers pass through memory retrieval, language, and social interaction. People compress repeated events, fill causal gaps, choose acceptable language, and adapt detail to the interviewer. Behavioral logs have other blind spots but measure a different object. Disagreement can reveal opportunity constraints, definitions, or recall rather than dishonesty.
Studying it
Classify claims as experience accounts, evaluations, explanations, or frequency statements. Pair claims requiring behavioral evidence with observation, telemetry, documents, or artifacts, aligning time window, task, and definitions. Preserve uncertainty and context rather than converting usually or I feel into facts.
Where it stops holding
When the object is meaning, attitude, norms, or identity, the account itself is central evidence. Observation cannot read motives and may alter behavior. Private, rare, or irreproducible activity may only be accessible through testimony, so the appropriate response is to calibrate the claim rather than dismiss interviews.
Applying it
- Specify whether each question needs accounts, behavior, or both.
- Report “participants said they had done” unless behavioral evidence supports “participants did.”
- Request concrete examples or artifacts for critical workflows with appropriate privacy permission.
- Explain saying–doing gaps through time, setting, and opportunity before attributing dishonesty.