Sources are part of the result
The current registry contains 1,120 versioned records. Every scored qualitative observation identifies its publisher, title and URL; missing data stays missing.
Source hierarchy
- Treaties, legislation, sanctions registers and official datasets.
- Government and intergovernmental statements or institutional records.
- Reputable secondary reporting only when a suitable primary record is unavailable.
Official bilateral evidence
Treaties, laws, sanctions registers, defence agreements, diplomatic statements and records from governments or intergovernmental organisations.
Reviewed when a relationship is added or a qualifying event occurs.UN General Assembly voting
Recorded-vote comparisons use the UN Dag Hammarskjöld Library dataset. Non-votes are excluded and years without comparable votes remain missing.
Annual snapshot.Open official source ↗UN Comtrade
Directional merchandise trade is imports plus exports reported by the source country. Reporting asymmetries and unavailable years remain visible.
Annual refresh.Open official source ↗UCDP conflict data
Interstate conflict coverage uses the UCDP/PRIO Armed Conflict Dataset alongside event-specific primary records.
Versioned annual release.Open official source ↗World Bank country indicators
Observed national context includes GDP, population, inflation, unemployment, life expectancy and other comparable indicators. These do not affect relationship scores.
Refreshed from the latest available source year.Open official source ↗IMF World Economic Outlook
Current-year GDP, growth, inflation and unemployment projections are displayed separately and always labelled as estimates.
April and October database releases.Open official source ↗What the atlas does not infer
No report is treated as zero merely because it is unavailable. Shared membership of an organisation does not automatically become a bilateral alliance. A strategic partnership label does not activate a defence commitment. National indicators and IMF forecasts provide country context only and never alter the relationship score.
Corrections and revisions
Official datasets can be revised and links can change. When evidence is corrected, the project preserves the methodological rule and records the new observation rather than tuning an impact to reproduce an earlier score. Suspected problems can be submitted through the correction channel.