Data Quality and Monitoring

Metricis separates automated validation from human review, queries, source verification, signatures, and locks. Passing one layer does not imply that every later layer is complete.

Automated validation

FormSpec rules can check required values, types, ranges, choices, branching dependencies, and cross-field conditions. Validation runs against the governing form version and records findings without silently changing the source response.

Missing and expected data

Schedules and bindings define what was expected for a participant and context. Review can distinguish not-yet-due, missed, intentionally skipped, withdrawn, not applicable, and technically failed work instead of treating every absence as the same kind of missingness.

Query workflow

Authorized researchers can raise a query against a record or field, assign or respond, resolve it, and retain the conversation and state history. A correction records reason-for-change and leaves the prior value reconstructable.

Source data verification

Risk-based SDV can target records or fields according to the monitoring plan. Verification records who reviewed what version and when. If the underlying data changes, the previous verification does not silently carry forward.

Signatures and locks

Signatures are attributable workflow events with meaning defined by role and context. Locks prevent ordinary mutation after an authorized transition; any exceptional correction follows a controlled, auditable path.

Audit reconstruction

An audit reconstruction should connect the research context, participant, definition/version, response state, values, actors, timestamps, validation findings, queries, verification, signatures, and export lineage. Technical history supports—but does not replace—the protocol's data-management and monitoring plans.

See Trust and governance, Forms and data capture, and Exports and analysis.