Publication geography restatement

The 2021–2025 publication corpus contains 24,869 source articles and reviews. Under the geography rules dated 28 October 2026, 24,743 of them remain in the reviewed count of works with an Arkansas institutional affiliation, and 126 are removed or set aside as unresolved. 15 works have only unresolved candidate institution matches and no accepted match.

This count corrects a defined corpus. It is not a census of research physically produced in Arkansas.

How institutions were checked

All 100 institution control records were inspected against the 11 October 2026 source geography, current source metadata and lineages, and the exact institution mappings in the publication affiliations. Primary institutional and government sources were used to resolve selected conflicts. This inspection is not an independent validation of every record.

Current metadata were collected on 28 October 2026. They were used for checking only. Publication counts remain those from the 11 October 2026 source data and are not replaced with current API totals.

What was removed or reassigned

Three source geocodes are rejected. York District Hospital is in York, United Kingdom. Galera Therapeutics and Ocugen are in Malvern, Pennsylvania, and had been placed in Malvern, Arkansas. Removing these alone leaves 24,813 source records.

Further decisions at the level of individual works reject explicit entity mismatches, reassign three clear confusions between Arkansas units and institutions, and set aside unresolved names. A work is always kept if it has another valid Arkansas institutional match. Papers from BioStrategies LC that give explicit Jonesboro, Arkansas addresses are kept, and are handled separately from the Chicago organization's conflicting source metadata.

An out-of-state address is not treated as an automatic entity error. A genuine Walmart institutional affiliation, for example, still counts as an institutional affiliation in these counts even when the work took place elsewhere. The counts therefore describe institution-linked output, not physical research production within the state. Institution assignments and topic and subfield classifications are observations from the source data and classification models. They are not independently validated scientific facts.

Counts available

The recent-window counts table covers all seven broad ARA proxies, six delivery applications, four sensitivity definitions and the all-field cohort. These views overlap and should never be added together. The subfield counts table gives the 149 specialization fields shown previously, as counts.

Cells that earlier figures left blank for privacy, and the earlier display eligibility rules, are unchanged. No personal fields or affiliation strings are published.

Measures not reported

No national or peer comparison is reported for the corrected counts. No national or peer work ledger has been corrected in a compatible way, and there is no complete 2016–2020 ledger. National country codes also contain errors in foreign institution assignments. Citation-percentile indicators were not collected for the removed works.

For these reasons, no current national share, specialization, citation quality, peer rank or historical publication growth figures are given here. Missing values for these measures are not zero. Older figures remain available as historical evidence and are labeled as uncorrected.

Sources and reproducibility

All 24,869 work records are covered by raw source pages. For 94 of them, pages with full authorship lists replace earlier truncated pages.

Analytical inputs that are not published keep public work IDs and hashed addresses so the results can be reproduced. Published attachments contain only aggregates and institutional metadata. The restatement can be rebuilt offline from a copy of its inputs, without calling any network service, and its record of the build lists the source hashes, scope, geography rules and all output hashes.