Method
Data source hierarchy
Open Ordinal prioritizes official public datasets first, then recognized institutional repositories, then peer-reviewed reconstruction datasets when primary data is unavailable.
- Multilateral and national statistical sources (World Bank, IMF, AfDB, UN, national agencies)
- Institutional repositories with documented definitions and revision history
- Academic reconstruction series when gaps prevent direct comparison
Collection and normalization workflow
- Collect raw series and archive original source links
- Normalize units, date windows, and definitions before comparison
- Generate charts and summary tables from the cleaned dataset
- Document caveats, exclusions, and uncertainty in each entry
AI-assisted workflow
AI tools are used across drafting, data cleaning checks, and chart preparation. All prose is substantially rewritten by a human editor. Interpretation, conclusions, and fact-checking are human-led throughout.
Interpretation principles
- Data before narrative — we describe what the data shows before proposing what it means
- Definitions before comparisons — terms are established before figures are set against each other
- Limitations before conclusions — we state what the data cannot support before stating what it can
- No causal claims from correlation alone — association is noted, causation is not assumed
Corrections and versioning
Entries are dated snapshots. If an issue is found, a correction entry is published and linked from the original. Original entries are not silently rewritten.
Reproducibility and licensing
Code is licensed under MIT. Content and data are licensed under CC BY 4.0. Repository: github.com/openordinal/open-ordinal