Publisher Gap dataset: metadata completeness rates for 48 fields across 14 scholarly publishers in Crossref and OpenAlex (2015-2025)

Consolidated dataset measuring metadata completeness for 835,664 journal articles from 14 publishers (2015-2025), comparing Crossref deposits against OpenAlex enrichment across 48 metadata fields (38 comparable, 10 Crossref-only). Publishers analyzed (Crossref member IDs): Elsevier (78), Springer Nature (297), Wiley (311), Taylor & Francis (301), SAGE (179), MDPI (1968), Frontiers (1965), PLOS (340), Hindawi (98), SciELO/FapUNIFESP (530), UNAM (9467), UAEM (11645), Universidad de Chile (3330), and Pontificia Universidad Católica de Chile (4571). This deposit contains two datasets: Dataset 1: Consolidated tidy dataset (`dataset.json`, `dataset.csv`): 7,392 observations (14 publishers × 48 fields × 11 years). Each observation represents one publisher × field × year combination with fill rates for both Crossref and OpenAlex, and the gap between them. Fields are classified by FAIR category (Findable Essential/Support, Accessible Essential/Support, Evaluation/CoARA), effort type (automatic vs. voluntary), and comparability. The JSON version includes the evaluation schema with field definitions, the API query URL templates used for data collection (Crossref REST API, OpenAlex API), and publisher metadata with verified Crossref member IDs. Dataset 2: Complete DOI list (`doi-list.json`, `doi-list.csv`): 835,664 journal article DOIs sampled via stratified quarterly sampling (Q1-Q4, 2015-2025). Each record includes publisher_id, doi, crossref_url, and openalex_url, enabling full reproducibility and verification against live APIs. The JSON version is grouped by publisher; the CSV version provides a flat table ready for R/Python/Excel. The dataset supports the study "Do publishers care about metadata?" and the interactive dashboard at https://metaudits.rijdho.org/publisher-gap/