Self-citation handling in the 2026 Stanford/Elsevier science-wide author databases
The 2026 release (version 9) of the Stanford/Elsevier science-wide author databases, widely referred to as the “top-2% list”, ranks scientists by a composite citation indicator (the c-score) computed both with and without self-citations. This page examines how self-citations enter the selection, using Table 1 of the release: 243,949 scientists on the career-long list and 249,310 on the single-year (2025) list. All figures are aggregates, and no individual is named.
Entry routes
A scientist is listed when any one of four conditions holds: a place in the top 100,000 by c-score, or in the top 2% of the primary subfield, each computed with or without self-citations. Accordingly, the self-citation-free figures that the release publishes do not remove anyone from the list. It is noted that the self-citation rule of the list is already broad: a citation is counted as a self-citation for every author of the cited paper whenever the citing paper shares at least one author with it.
Scientists who qualify through one ranking only
Career-long list
Single-year list (2025)
Scientists who qualify either way: 223,810 (career-long) and 224,013 (single-year). A list built only on the self-citation-free routes would drop the first group and keep the second.
Where entry through self-citations is most common
Share of listed scientists who qualify only with self-citations, by country
Career-long list; a selection of countries with 500 or more listed scientists, by country of the most recent institution. Greece highlighted.
The same share, by field
Career-long list; all 20 fields of the Science-Metrix classification. Engineering highlighted.
Civil, coastal and related subfields
Career-long list, by primary subfield. Greece overall: 60 of 1,243 listed scientists (4.8%).
| Subfield | Listed | Only with self-citations | Share | Median self-citation rate |
|---|---|---|---|---|
| Civil Engineering | 1,427 | 73 | 5.1% | 15.7% |
| Oceanography | 350 | 18 | 5.1% | 14.1% |
| Environmental Engineering | 1,439 | 66 | 4.6% | 12.4% |
| Geological & Geomatics Engineering | 1,724 | 69 | 4.0% | 12.8% |
| Building & Construction | 1,173 | 44 | 3.8% | 10.2% |
How much of “self-citation” comes from co-authors
On a single-authored paper, only its author can be the shared author, so the self-citation rate on single-authored papers approximates the author’s own self-citing; the gap to the overall rate approximates what co-authors add. For the fifth of authors with the largest teams, the overall rate is 11.1% against an own rate of 3.9%; for the fifth with the smallest teams, the two rates are 6.3% and 4.3%.
Median self-citation rate by typical team size
Authors grouped into fifths by hm-index divided by h-index, which falls as author lists grow. Covers the 102,137 listed authors with three or more single-authored papers and 200 or more citations to them. The approximation assumes that authors cite their single-authored and team papers at similar rates.
What the rule leaves behind
Citations per citing paper, with self-citations excluded
Career-long list, in tenths by self-citation rate (Spearman correlation 0.355 across all authors). The remaining citations of heavy self-citers arrive in bundles more often. Bundles from collaborators who are not authors of the cited paper would produce this pattern, although narrow research niches can produce it too.
Median drop of each c-score indicator when self-citations are excluded
Career-long list. The co-authorship-adjusted hm-index, highlighted, is among the least affected indicators.
Method and limits
- Entry routes are rebuilt from the overall and subfield ranks of each scientist, with and without self-citations. The rebuilt counts match the per-field counts of Table 2 of the release exactly, and the c-score recomputes from its six indicators to machine precision.
- Self-citations follow the rule of the list: a citing paper that shares at least one author with the cited paper is a self-citation for every author of the cited paper.
- Countries refer to the most recent institution of each scientist, as listed in the release.
- All results are associations within one dataset. Separating own, co-author and wider-network citations properly requires paper-level citation data; that analysis is in preparation.
Data and related work
Ioannidis, J.P.A. (2026). August 2026 data-update for “Updated science-wide author databases of standardized citation indicators”. Elsevier Data Repository, V9, DOI.
Ioannidis, J.P.A., Boyack, K.W. and Baas, J. (2020). Updated science-wide author databases of standardized citation indicators. PLOS Biology, 18 (10), e3000918, DOI.
Samaras, A.G. (2026). A declared percentage-based author-contribution system for weighted research metrics: A proposal. MetaArXiv preprint, DOI.
Samaras, A.G. (2026). dpac: declared percentage-based author-contribution metrics from OpenAlex publication records. Software, DOI; GitHub.