Metadata
The metadata records the type of each feature and the conditions under which each data set was collected or created, with the reasons for each choice.
What clause 7.4.2 asks
The metadata output records the data type of each feature and the reasons for selecting it, based on the anticipated outcome, the context of use and the data that is available. It also records the conditions under which each data set was collected or created. The clause gives a long list of conditions, from the source and jurisdiction to the collection method, consent, opt-out, timing, anonymization and the relation to other data sets. The metadata is a substantive record, not a schema description.
Why it matters
Metadata is where the assumptions behind a data set become visible. A feature that looks neutral can be a proxy for a protected attribute, and the reason for selecting it is the evidence that the team thought about that. The collection conditions matter because two data sets with the same columns can carry different biases depending on how, when and from whom they were collected. The clause asks for the reasons, not only the facts, so a later reader can test the judgement.
What good looks like
Record the feature type and the selection reason in one table, and the collection conditions in another, linked by data set. Use the clause list as a checklist and mark any condition that does not apply, with the reason. Review the metadata when a data set is refreshed or a feature is added. Keep the metadata with the map, so the population and the conditions are read together.
Evidence an assessor asks for
The assessor will pick a feature and ask why it was selected. A feature table that lists only the type invites the question of whether the feature was needed at all.
- Feature table Feature, data type, selection reason and the outcome it is expected to support
- Collection conditions The clause checklist for each data set, with not-applicable reasons
- Relation record How each data set relates to the others, including complementary or overlapping populations
- Refresh record The metadata review after a refresh or a feature change
Where it commonly fails
- Metadata is a schema dump with no reasons
- Collection conditions are recorded for the main data set only
- A refresh changes the data without updating the metadata
What to do before publication
Use the clause list as a checklist and record the not-applicable items. The silence is often where the assumption hides.
Read across
← 7.4.1 Decisions about data7.4.2.1 Collection and creation conditions →