Data preparation and finalization

Resolve queries and derive datasets

Freeze a reproducible version after resolving data queries and documenting derivations. A frozen dataset may still have limitations and missing values; amendments require a new version and an audit trail.

Flow diagram

Data preparation and finalizationF1.S1Collect and validate source dataF1.S2Resolve discrepancies and queriesF1.S3Derive variables and datasetsF1.S4Review missingness and protocoldeviationsF1.S5Review and finalize dataF1.N1Query and decision logIssueSourceDecisionOwnerResolutionVersionF1.N2Dataset specificationDatasetPopulationVariablesDerivationsVersionChecksF1.S0 Start flowData preparation and finalizationF1.S999 End flowEnd flow

Steps

Collect and validate source data

Check source provenance, permitted access and coding conventions.

Resolve discrepancies and queries

Define ownership, adjudication and audit trails.

Derive variables and datasets

Version derivation rules and participant identifiers.

Review missingness and protocol deviations

Prespecify which issues affect each analysis.

Review and finalize data

Approve the analysis-ready version and reproducible programs.

Data preparation and finalization

End flow

Tables and supporting information

Query and decision log

Query and decision logIssueSourceDecisionOwnerResolutionVersion

Preserve decisions and source lineage.

Dataset specification

Dataset specificationDatasetPopulationVariablesDerivationsVersionChecks

Plan and track reproducible data preparation.

Sources