survey-data-auditlisted

# Survey Data Audit The pass a careful survey team runs between soft launch and analysis: confirm the instrument captured what was registered, screen for automation and low-effort responding, reconcile the realized sample against quotas and vendor dispositions, and emit an appendix table plus a methods paragraph. Platform-agnostic; examples name Qualtrics-style fields because that is where the field names are standardized. ## When to use - Soft launch has landed (~50-150 completes per market) and a go/no-go read is due before full field. - Fielding is complete and the data-quality appendix has to exist before anyone touches the outcomes. - A vendor, reviewer, or PI asks what the study did about bots and AI respondents. Not for reshaping conjoint exports (that is `conjoint-cleaning`), and not for inventing exclusion rules a study should have registered (that is `pre-registration-writing`). ## Inputs Required: a response-level export, one row per submission, partials and screenouts included where the platform exports them. Optional, each unlocking a phase: - interaction paradata (per-page event counts, honeypot indicator, automation-surface flags) - platform fraud fields (reCAPTCHA score, duplicate/fraud identity scores) - geoip fields, IP, per-page timings, total duration - vendor disposition file (completes billed, screenouts, quota-fulls, terminates) - quota targets per cell - the pre-registration, PAP, or data-quality SOP - the instrument definition (QSF/JSON/codebo