Absent or unmatched? Missing affiliations in OpenAlex by country, year and field, compared with Scopus — data and code

This record contains the aggregate tables, figures and code behind the manuscript “Absent or unmatched? Missing affiliations in OpenAlex by country, year and field, compared with Scopus” (submitted to Scientometrics, 2026). The study classifies all 524,113,428 authorships of the 144,506,114 articles and reviews published since 2010 in the OpenAlex snapshot of 26 June 2026 into three states — matched to an institution, no affiliation string present, and string present but unmatched — and describes the two gaps by publication year, field, source type, language, author position and inferred country. A fixed-seed sample of 191,655 DOI-bearing works is compared with Scopus: indexing status for all sampled works (Scopus Search API) and complete affiliation lists for a stratified subsample of 17,337 works (Scopus Abstract Retrieval API). Country shares of world authorships are recomputed under three accounting rules to show the effect of the gap. Contents: data/openalex_population/ — authorship counts by state for year × field, language, source type, DOI presence, author position and source (whole population). data/openalex_population_by_source/ — the same cross-tabulated by year × source type, year × language, country × source type, year × DOI. data/country_inference/ — country × year and country × field tables of the two gaps with the country of unlinked authorships inferred from the author’s other works; coverage of the inference; country shares under three accounting rules. data/scopus_comparison/ — the full sampling frame (DOI, OpenAlex state, Scopus indexing status, subsample membership) and the stratum-weighted comparison tables (affiliation availability, country-set agreement, country-level recall, agreement by year and field). figures/ — Fig1–Fig6 as PNG and PDF. code/ — Python scripts for every step, from the snapshot scan to the figures, with a README giving the run order. MANIFEST.md — description of every file and column. Raw Scopus records are not included, in accordance with Elsevier’s API Service Agreement; the DOI list and stratum assignments allow the comparison to be reproduced with any Scopus API entitlement. The OpenAlex snapshot is public (CC0) and is not redistributed. The 23.3 million unmatched raw affiliation strings are deposited as a separate record linked below. Version 1.1.0 adds: country tables recomputed with matched authorships without an author identifier included in the denominators; a leave-one-out test of the country inference on matched authorships and a check of the inferred countries against Scopus country sets; country tables and accounting rules for journal-hosted, DOI-bearing works; tables by field × source type, publisher, and author-identifier presence; a recheck of a sample of unmatched authorships against the live OpenAlex API after the September 2026 matcher change; scripts 09, 10 and 11.