find-data

name: find-data description: "Find and assess datasets for a research question. Use when Codex needs to identify candidate data sources, compare coverage and access constraints, evaluate whether data can support an empirical design, or turn a project idea into a concrete data-sourcing memo."

Find Data

Use this skill when the binding constraint is the dataset, not the estimator or writing.

Workflow

1. State the concept to be measured and the unit of analysis.
2. List candidate datasets.
3. Compare them on:
   • coverage
   • access cost
   • update frequency
   • variable fit
   • compatibility with the likely design
4. End with a ranked shortlist and the main caveats.

Output rules

• Distinguish public, restricted, and purchased data.
• Note measurement error and sample-selection risks.
• Do not confuse "exists" with "supports the design."

Load references as needed

• Read references/data-assessment.md for the comparison rubric and memo format.