Citation Cliques in Low Impact Journals

Abstract: This exploratory study examines how low-impact journals, defined through subject-normalized Eigenfactor percentiles, are associated with denser and more reciprocating patterns of author-to-author citations. Using Crossref records, we assign journals to broad subject areas, compute subject-specific Eigenfactor scores, propagate venue quality to works and authors, match authors in low- (Case) versus high-influence (Control) venues by subject and h5, and analyze citation edges for cohesion and anomalies. Across a 10% sample of 9,431 matched pairs, authors in low-impact venues exhibit significantly higher cohesion: 6.7x higher co-author citation rates and 4.7x higher reciprocity in the aggregate Case-Control comparison. A subject-aware hybrid detection pipeline flags 277 outliers with 93.5% Case purity; these outliers display an 11x clique-strength lift relative to non-outliers, revealing a stark "Two Worlds" segregation (r = 0.71) where low-impact venues operate as closed citation economies. The largest detected component (n = 23) displays a hub-and-spoke topology in which peripheral "Sycophants" funnel citations to central "Beneficiaries" through coordinated bursts, confirming a directed flow imbalance rather than reciprocal exchange among equals. Overall, cohesion, rather than broad asymmetry, accounts for the main Case-Control differences, suggesting that low-impact venues foster segregated, inward-looking citation economies that distort bibliometric indicators.

• The paper demonstrates that low-impact journals exhibit unusually high internal cohesion, with co-author citation rates 6.7× greater than typical norms.
• The study employs a hybrid detection pipeline using network metrics like clustering and reciprocity to identify distinct citation cliques across multiple disciplines.
• The findings challenge reliance on aggregate citation metrics, advocating for network-centric approaches to improve integrity in bibliometric evaluations.

Citation Cliques and Segregated Citation Economies in Low-Impact Journals

Introduction

The paper "Citation Cliques in Low Impact Journals" (2605.11930) provides a systematic investigation into the structural citation behaviors of authors publishing primarily in low-impact journals, with a focus on dense, internally referencing author networks—termed "citation cliques." The study operationalizes "low-impact" via subject-normalized Eigenfactor percentiles and focuses on analyzing the dynamics of author-to-author citation patterns, examining both cohesion (the density and reciprocity of citations within small author groups) and structural segregation from the broader research community. Methodologically, the work answers whether venue quality correlates with anomalous network patterns that may confound the interpretation of citation-based bibliometric indicators.

Methodological Framework

The researchers develop a subject-normalized, large-scale data pipeline anchored in Crossref metadata (2020–2024). Key methodological components include:

• Journal Subject Classification: Classification via LLM-based assignment to five broad subject areas, modeling after Scopus's ASJC taxonomy, enabling within-field normalization.
• Journal Impact Quantification: Computation of subject-specific Eigenfactor scores to establish venue impact percentiles, propagating impact from venues to individual works and author portfolios.
• Matched Case–Control Author Sampling: Authors with $\geq$70% of their output in bottom (Case) or top (Control) Eigenfactor quartile journals are pair-matched by subject and $h_5$ index, controlling for discipline and career productivity.
• Network Metrics and Outlier Detection: Citation networks constructed at the author–author granularity are analyzed using features like co-author citation rates, reciprocity, local clustering, k-core, eigenvector centrality, and Herfindahl-Hirschman indices. Outlier detection employs a hybrid of multivariate anomaly detection (Isolation Forest) and a domain-specific Cohesion Composite Score.

The study analyzes $9,431$ matched pairs across five scientific domains, with all code and derived data released openly.

Empirical Findings

Structural Cohesion and Citation Patterns

The analysis reveals that low-impact venues are associated with distinctly higher internal cohesion along multiple axes:

• Co-author Citation Rates: On average, authors in low-impact (Case) venues cite their co-authors $6.7\times$ more frequently, a result highly significant ($p \ll 10^{-40}$).
• Reciprocity and Clustering: Reciprocity rates are $4.7\times$ higher in the Case group. Clustering coefficients and local clique strength similarly exhibit multi-fold increases.
• Outgoing HHI and Tunnel Vision: Outgoing HHI, quantifying the concentration of citations directed at a narrow subset of authors, is $3.1\times$ higher among Cases.

This cohesion gap is a broad, population-level property not driven by outliers, as confirmed by permutation testing and meta-analytic effect sizes across subject domains.

Clique Detection and Anomaly Purity

Deploying the hybrid outlier detection pipeline—Isolation Forest with a cohesion-based filter—a set of 277 high-confidence anomalous author profiles is isolated, with 93.5% of these flagged authors concentrated in the Case group. These outliers demonstrate clique metrics—co-author citation rates and clique strength—over an order of magnitude higher than the non-anomalous baseline. Sensitivity analysis confirms the robustness of the hybrid approach relative to standard anomaly detectors.

Network Topology and Citation Flow

Network forensics uncover predominant hub-and-spoke topologies within the largest detected cliques. Peripheral "Sycophant" authors channel citation flows toward "Beneficiary" hubs, characterized by strong directionality and coordinated bursts rather than egalitarian exchange. These orchestrated bursts occur in short temporal windows, strengthening the evidence of strategic citation behavior.

Segregation and Assortativity

An extreme degree of network segregation is measured ($r=0.71$ assortativity, $Q=0.97$ modularity), substantiating a "Two Worlds" model in which authors in low-impact venues predominantly cite one another, forming insular citation economies largely disconnected from high-impact venues.

Cross-Subject and Archetype Analysis

Cohesion-based anomalies generalize across health, life, social, and physical sciences, though with field-specific signatures in temporal citation bursts or clustering. K-means clustering reveals three behavioral types—Central, Independent, and Solo—with Cases massively overrepresented in the Solo (extreme cohesion) archetype.

Implications

Theoretical Significance

The findings reinforce and extend existing models of "citation orchestration" and metric manipulation [Evdaimon2024], showing that clique behavior is not an isolated abnormality but a structural byproduct of venue-level incentives. The clear statistical separation of these groups—controlling for discipline and productivity—invalidates the naive equivalence of citation count or $h$-index as field-transcending measures of research influence.

The demonstration that cohesion—rather than broad asymmetry—dominates the difference between Cases and Controls distinguishes these patterns from generic collaboration or mere self-citation. Notably, clique behavior manifests as increasing citation segregation over time within these low-impact venues, suggesting an evolutionary process toward closed-loop citation economies.

Practical Implications

Bibliometric evaluation systems predicated solely on aggregate citation counts, $h_5$0-index, or naive impact factors are significantly vulnerable to distortion by cohesive, internally-referencing author groups. The hybrid detection pipeline—anchored in behavioral network features—outperforms generic or univariate statistical outlier detectors and demonstrates practical feasibility for large-scale bibliometric surveillance.

Flagging insular citation economies in venues with limited editorial oversight, and small-journal ecosystems, has downstream ramifications for research assessment frameworks, funding allocation, and international rankings. The identification of structural cliques provides actionable intelligence for publishers, indexers, and integrity offices seeking to audit manipulated citation practices.

Open Challenges and Future Work

The study's authors highlight several open issues:

• Generalizability and Robustness: Reliance on LLM-based subject classification and Crossref metadata, as well as exclusion of text-level citation motivation, present sources of potential misclassification.
• Cartel vs. Specialization: There remains inherent difficulty in distinguishing deliberate self-reinforcement from legitimate scholarly subnetworking or niche specialization.
• Temporal Evolution: The trend toward widening segregation indicates the need for extended temporal observation and potentially adaptive reweighting of bibliometric measures.
• Integration with Editorial Policy: Experimental audits and intervention studies could assess whether targeted editorial policies effectively disrupt echo-chamber citation economies.

Conclusion

This work provides quantitative evidence that low-impact journals foster statistically distinguishable, segregated citation cliques typified by high co-author citation rates, dense closure, and directional citation flows. The study establishes network cohesion—not citation volume per se—as the primary fingerprint of anomalous citation behavior in these venues. These insights compel the scientometric community to move beyond aggregate citation metrics, emphasizing network-centric diagnostics for integrity-preserving evaluation of scholarly influence.

---

Reference:

“Citation Cliques in Low Impact Journals” (2605.11930)