Beyond self-citation rates: A model-based approach to detecting anomalous journal self-citation

A journal self-citation rate is easy to read: it tells us what share of the citations received by a journal comes from that journal itself. The harder question is whether the count producing that share is unusual for the circumstances of the journal. We examine that question using 137,801 journal-years from 30,253 Scopus-indexed journals between 2021 and 2025. Expected self-citation is estimated with a negative-binomial model that includes external citation capacity, recent self-citable stock, external recognition, and year. Ten-fold cross-fitting is carried out at journal level, which keeps the evaluated journal out of the model used to produce its expectation. The observed count is then placed in the corresponding predictive distribution and summarized as the Journal Self-Citation Anomaly Score (JSCAS). Overall calibration is close, and the same is true across expected-count strata. The exception is the far-right tail, where extreme observations occur more often than the negative-binomial reference predicts. The conventional self-citation rate (SCR) and JSCAS are strongly related overall, but their upper tails overlap only partially. In journal-years receiving at least 500 citations, the JSCAS-only and SCR-only groups have similar observed self-citation counts but markedly different model-based expectations, illustrating how contextualization changes case selection. The longitudinal results add a different piece of information: a large departure in one year is not necessarily repeated. We therefore treat JSCAS as a complement to the conventional rate. Additional notePreprint / submitted manuscript. This version has been submitted to Journal of Informetrics and has not yet undergone peer review. The deposit includes the preprint manuscript and its Supporting Information (Tables S1–S8). The underlying Scopus-derived data are not publicly redistributed because they are subject to Elsevier licensing restrictions.

Authors

Institutions

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-21
DOI
https://doi.org/10.5281/zenodo.24424722
Primary Topic
scientometrics and bibliometrics research
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Beyond self-citation rates: A model-based approach to detecting anomalous journal self-citation

Rodrigo Sánchez Jiménez, Félix de Moya Anegón, Vicente P. Guerrero‐Bote, Pablo Guerrero-Castillo
Zenodo (CERN European Organization for Nuclear Research)
scientometrics and bibliometrics research
preprint

Beyond self-citation rates: A model-based approach to detecting anomalous journal self-citation

Rodrigo Sánchez Jiménez, Félix de Moya Anegón, Vicente P. Guerrero‐Bote, Pablo Guerrero-Castillo
preprint en

Abstract

A journal self-citation rate is easy to read: it tells us what share of the citations received by a journal comes from that journal itself. The harder question is whether the count producing that share is unusual for the circumstances of the journal. We examine that question using 137,801 journal-years from 30,253 Scopus-indexed journals between 2021 and 2025. Expected self-citation is estimated with a negative-binomial model that includes external citation capacity, recent self-citable stock, external recognition, and year. Ten-fold cross-fitting is carried out at journal level, which keeps the evaluated journal out of the model used to produce its expectation. The observed count is then placed in the corresponding predictive distribution and summarized as the Journal Self-Citation Anomaly Score (JSCAS). Overall calibration is close, and the same is true across expected-count strata. The exception is the far-right tail, where extreme observations occur more often than the negative-binomial reference predicts. The conventional self-citation rate (SCR) and JSCAS are strongly related overall, but their upper tails overlap only partially. In journal-years receiving at least 500 citations, the JSCAS-only and SCR-only groups have similar observed self-citation counts but markedly different model-based expectations, illustrating how contextualization changes case selection. The longitudinal results add a different piece of information: a large departure in one year is not necessarily repeated. We therefore treat JSCAS as a complement to the conventional rate. Additional notePreprint / submitted manuscript. This version has been submitted to Journal of Informetrics and has not yet undergone peer review. The deposit includes the preprint manuscript and its Supporting Information (Tables S1–S8). The underlying Scopus-derived data are not publicly redistributed because they are subject to Elsevier licensing restrictions.

Zenodo (CERN European Organization for Nuclear Research)
Universidad Complutense de Madrid (ES), SCImago Research Group (ES), Universidad de Extremadura (ES)
scientometrics and bibliometrics research
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon