Self-assessment

Select the statement that best describes the documentation of one survey today, one per chapter. The result shows a level for each chapter and the chapters to read first. Answers are stored in this browser only.

1. Find How is the survey described at the study level?

Not yetA title and a report; no structured record.FoundationalA study record with abstract, dates, coverage, universe, and sampling, in a catalog.AI-readyThe record follows DDI Codebook in the World Bank schema, has a DOI, and is published with schema.org markup.AI-nativeThe record is kept current with versions, and its variables are searchable across surveys.

2. Produce How is documentation produced for a survey that has data files but no record?

Not yetBy hand, from scratch, when someone asks for the survey.FoundationalA dictionary is drafted from the data file and labelled files are imported into the editor; the backlog is listed.AI-readyQuestion text and code lists are recovered from questionnaires and codebooks with their source, and the backlog is ordered by use and gaps.AI-nativeDraft, import, recovery, and checks run as a pipeline on every new file before a curator sees it.

3. Variables How are the variables documented?

Not yetVariable names only, or labels for some variables.FoundationalEvery variable has a label, a type, value codes with labels, and missing-value codes.AI-readyThe dictionary is in the DDI-based schema with universes and question text, and it is checked before release.AI-nativeVariables carry concepts, derivations, and summary statistics, and the check runs on every release.

4. Concepts How are variables linked to concepts and classifications?

Not yetNot linked; classifications are described in the report, if at all.FoundationalThe questionnaire is published and the main classifications are named in the documentation.AI-readyA concept map links variables to classifications (ISCED, ISCO, ISIC) with versions, and question text is attached to each variable.AI-nativeConcepts are reused across surveys with harmonized names and a question bank.

5. Navigate How does a user find a variable among hundreds?

Not yetBy reading the data dictionary from top to bottom.FoundationalVariables are grouped into themes in the catalog and can be searched by keyword.AI-readyVariable search works by meaning and across surveys, and it is scored with a question set.AI-nativeThematic groups are generated and reviewed for every new survey, and the search is re-scored on each change.

6. Rounds How are differences between survey rounds documented?

Not yetThey are not; users compare variables by name.FoundationalA crosswalk between the latest two rounds is published with the record.AI-readyEvery pair of rounds has a crosswalk; recoded variables have correspondence tables; breaks are stated in the record.AI-nativeHarmonized variables with reusable definitions span the rounds, and a concept search returns each round with its comparability.

7. Fitness How can a user judge whether the data fit a purpose?

Not yetFrom the report, if it describes the design.FoundationalSampling, weights, response rates, and known issues are in the study record.AI-readyQuality information is in machine-readable fields, with versions and errata.AI-nativeQuality indicators are computed and published with each release, and changes are logged.

8. Access How are the access conditions stated and applied?

Not yetAccess is by request, with conditions decided case by case.FoundationalA written policy defines access tiers and the disclosure control for each.AI-readyEach record states its tier, licence, citation requirement, and restrictions in machine-readable fields.AI-nativeAutomated clients receive the conditions with the metadata, and only public use files are exposed to them.

0 of 8 answered

Interpretation of the result

• A Not yet or Foundational result in chapters 1 to 3 comes first: study-level and variable-level documentation are what every later chapter reads, and chapter 2 is where that documentation comes from when the files exist and the records do not.
• Chapter 8 can start at any time; the access policy is a one-day task that makes the rest safe to publish.
• Chapters 4 to 6 give the largest gains for users once chapter 3 is at AI-ready, because concept links, thematic groups, and crosswalks turn a long dictionary into something searchable and comparable.