Research

Records

Evidence and research records.

RecordsSelect study
Select a study to begin.
Reference & policy
DatabaseNetlify PostgreSQLDedicated research schema
Modeproduction-read-ready16 integrated production migrations applied
Runtime writesDisabledRead path active; write activation remains controlled
IdentitySeparatedRestricted linkage outside routine analysis

Record layers

Each layer has a purpose and a boundary.

Research records should connect cleanly without allowing convenience to collapse participant identity, evidence, analysis, and publication into one undifferentiated store.

01Study governance
  • study
  • retention policy
  • AI permission
  • access policy
02Restricted identity vaultRestricted
  • identity profile
  • participant linkage
03Consent and permissions
  • consent version
  • permissions
  • withdrawal
04Collection
  • instrument version
  • fieldwork session
  • session participant
05Evidence and provenance
  • evidence object
  • evidence version
  • checksum
  • storage reference
06Analysis
  • codebook
  • code
  • coding assignment
  • analytic memo
07Findings and review
  • finding
  • evidence link
  • challenge link
  • review event
08Audit trail
  • research action
  • review decision
  • release event

Persistence rules

The database records relationships and decisions, not just files.

The research schema is present in the managed production database and available to the governed read path. Browser and routine production writes remain disabled until explicit write activation is approved.

01

Participant codes, not names or email addresses, are the routine keys used across collection, evidence, analysis, and findings.

02

Direct identity is isolated in the identity vault and connected through an explicit restricted linkage record.

03

Binary evidence is referenced from approved storage; the database preserves provenance, version, checksum, sensitivity, and storage reference.

04

Source evidence is never overwritten by transcription, redaction, coding preparation, correction, or release transformations.

05

Code definitions are versioned so an analyst can determine which definition governed a coding decision.

06

A finding keeps separate links to supporting, challenging, contextual, and limiting evidence.

07

Analytical approval and external release are separate decisions with separate review evidence.

08

Machine-assisted outputs remain provisional unless a human researcher records the corresponding analytical or release decision.

Activation boundary

Production presence does not imply unrestricted writes.

The research schema is now present in the managed production database and the governed read path is active. Routine browser writes, restricted identity access, storage-provider changes, release actions, and other state-changing research operations remain separately controlled and require explicit activation and authorization.