Direct comparison
Data Catalog vs. Data Dictionary
Data catalogs index which datasets exist across a repository or org; data dictionaries document the fields within one dataset. Compare scope, tools, and use.
Side-by-side comparison
| Dimension | Data Catalog | Data Dictionary |
|---|---|---|
| Scope | Collection-level — many datasets across a repository/organization | Field-level — variables within a single dataset |
| Core question answered | Which datasets exist, and where? | What does each field/variable mean and permit? |
| Typical contents | Title, description, owner, location, format, access conditions, identifier (often a DOI) | Variable name, type, allowed values/labels, units, description, derivation logic |
| Common standards | W3C DCAT (Data Catalog Vocabulary), schema.org/Dataset, re3data for repository-level discovery | DDI Codebook, PREMIS (for preservation metadata specifically), discipline-specific codebooks |
| Primary user | Someone searching for a dataset they don't yet have | Someone who has the dataset and needs to interpret/reuse a specific field correctly |
| Typical tooling | Repository/catalog platforms — Dataverse, CKAN, DSpace, institutional data portals | Spreadsheets, README files, codebook-generation tools, discipline-specific metadata software |
| FAIR principle served | Primarily Findability (F) | Primarily Interoperability and Reusability (I, R) |
| Granularity of records | One record per dataset | One entry per field/variable within a dataset |
| Where it's usually specified | Repository/portal deposit metadata, DMP data-sharing section | DMP documentation-standards section; delivered alongside the dataset itself |
Common questions
FAQ
Do I need both a data catalog and a data dictionary?+
For a single small project, you may only produce a data dictionary — the dataset just isn't part of a larger indexed collection yet. Once that dataset is deposited in an institutional or discipline repository, the repository's own catalog record becomes the discovery layer, and your data dictionary remains the interpretation layer. At the institutional or organizational level, both are standard practice.
Is a data catalog the same as a repository?+
No. A repository is the storage/access system that holds the actual dataset files. A data catalog is the searchable index of metadata records describing what's in one or more repositories — a single catalog can span multiple repositories, and a repository can expose its own catalog interface.
Is a codebook the same as a data dictionary?+
They're closely related and often used interchangeably, but not identical — a codebook (particularly in survey/social-science research) typically also includes question wording and survey administration context, not just variable-level technical metadata. See CASRAI's Data Dictionary in Research Data Management guide for the fuller distinction.
Which one is required by a funder data management plan?+
Funder DMP requirements (e.g. NIH, NSF, UKRI) typically expect both in substance, even if not by these exact names: a plan for how the dataset will be made findable/deposited (catalog-level metadata) and a plan for how it will be documented so others can reuse it (data-dictionary-level detail). Neither term is usually mandated by name; the underlying obligations are.
