Guide · about 1 min read
What Makes a Good Dataset?
Documentation, stable identifiers, sensible granularity, and clear licensing or sensitivity boundaries.
Documentation you should expect
Look for a data dictionary, update cadence, known issues log, and contact for questions. Strong data documentation reduces silent misuse.
Grain and identifiers
Clarify whether each row is an observation you can trust for analysis. Stable IDs across files make merges safer than matching on names alone.
Sensitivity and ethics
Good publishers document redaction rules and align with data ethics expectations. External references such as the NIST de-identification overview (opens in new tab) help teams discuss risk.
Related reading
Return to the Working with datasets hub for curated resources and glossary bridges.
Related glossary terms
Related guides
Curated external resources
- Data.gov datasets (opens in new tab)
Federal open data catalog for practice reading metadata and documentation.
- W3C Data on the Web Best Practices (opens in new tab)
Standards-oriented checklist for publishing usable, documented datasets.