Methodology
How figures are collected, what each measure means, and how the platform behaves when a value is missing.
Sources and licences
World Bank Open Data
CC BY 4.0Macro economic figures: gross domestic product, income per person, growth and population, with the observation year stored beside every value.
Research Organization Registry
CC0 1.0Institution registry: which education organizations exist, where they are, when they were founded and their website.
OpenAlex
CC0 1.0Research metrics: indexed works, citations received, two year mean citedness and index values for every catalogued institution.
How each measure is defined
- Observation year
- Countries are compared for one common observation year. Countries without an observation in that year are excluded and counted, never treated as zero, and a cohort that misses the completeness threshold is withheld.
- Publication totals
- Sum of the indexed works of every catalogued institution in the country. A work with authors at more than one institution is counted for each participating institution, so this total is wider than unique national output.
- Deduplicated output
- Indexed works whose authors include at least one author affiliated with an institution in that country, counted once per work.
- Citedness
- Mean citations per work for works published in the two calendar years preceding retrieval, as published in the citation index summary statistics. It is not a journal impact factor.
- Lifetime citations
- All citations received by all indexed works of that entity, divided by the number of those works. A lifetime figure that rises as citations accrue.
- Data quality
- verified means the figure came back from the provider on a recent retrieval. provisional means it is a provider figure that has not been rechecked. stale means the retrieval is older than the freshness window. unavailable means no observation exists.
Freshness, verification and corrections
Every comparison states the date of the observations behind it and the date the table was generated. A table rebuilt today from older observations reports both dates, so a fresh build of stale data is visible as such.
Each value carries a quality status and a retrieval date. Values retrieved for the current build are marked provisional until the next verification run confirms them. Values whose retrieval has aged past the freshness window are reported as stale rather than silently presented as current.
Coverage is published country by country, so a gap is visible before it affects a conclusion. Where a correction is needed, the corrected observation replaces the previous one and keeps its own retrieval date.
Check the coverage yourself
The coverage matrix lists what has been observed for every country and when it was retrieved.
Open the coverage matrix