Practical guides · Page 11

Assign Visits to Analysis Windows with Explicit Tie Rules

Preserve actual assessment dates while resolving overlapping windows and repeated measurements reproducibly.

Separate weighted distribution summaries from interpolation rules and survey uncertainty.

Keep group variances and sample sizes separate when equal variance is not assumed.

Trace table, figure, and note callouts to the intended XML targets before extracting results.

Build article keys from identifier type and value when more than one identifier occurs in a record.

Resolve byline markers and preserve many-to-many author–institution relationships.

Track which included papers supplied each reference and count unique candidates correctly.

Calculate a binary Brier score from valid aggregate inputs and distinguish overall probability accuracy from calibration.

Audit bin boundaries, sample sizes, and aggregation before interpreting a probability calibration plot.

Use Shannon and Simpson summaries to describe category distributions without turning diversity into a clinical quality score.

Distinguish a credited group from its members and retain the published byline order during import.

Use Crossmark and the publisher record to identify updates affecting a saved article.