Crunching the numbers. Responsibly.
49.7% of links involve at least one definitive COI and a further 11.6% are possibly conflicted, with prevalence increasing over time
consistent link-level counts within a purposively selected cohort rather than all journal contributors
COIs cluster within a small, dense, and homogeneous elite cohort of 54 highly prolific scholars, predominantly North American (90%) and male (94.5%), who account for roughly half of all publications
direct cohort, publication, demographic, and link-level measurements support the descriptive concentration
Disproportionate publication gains during editorial service are observed only among editors embedded in reciprocal acceptance networks
descriptive within-person changes lack inferential testing and cannot exclude time-varying group differences
A comparable non-FT50 journal shows substantially lower COI rates, largely attributable to the same elite cohort
one same-field comparator shows a large descriptive difference but cannot isolate prestige from other journal conditions
reciprocal handling exhibits structural features consistent with a stratified and internally cohesive system of exchange, not coincidence
modularity, clustering, and k-core structure show cohesion, though no reported null model excludes coincidence
Derived from the full evaluation — not a separate score.
Strengths
The study reconstructs 946 author-editor links across two elite journals and quantifies definitive and possible conflicts under a stated reasonable-observer framework. This extends prior concentration research to editorial relationships themselves.
↳ Introduction; Data Collection and Network Construction; Analysis, Table 2
The paper combines a same-field comparator, within-person productivity comparisons, and reciprocity-network analysis. These analyses address disciplinary norms, editorial-role effects, and random dispersion from different angles.
↳ Robustness Checks 1–3; Figures 5–7
The paper explicitly distinguishes conflicts from misconduct, acknowledges the absence of rejected submissions, and identifies unresolved longitudinal dynamics. The Conclusion also disclaims intent, collusion, and individual misconduct.
↳ Limitations and Future Research; Conclusion
Limitations
The 49.7% estimate is calculated within 54 authors who each published at least ten papers, not across all 2,207 contributors. Selecting unusually prolific scholars may enrich network exposure, so the estimate is not demonstrably a lower bound for all editorial relationships.
↳ Data Collection and Network Construction; Analysis, Table 2
The primary citation comparison is non-significant, productivity changes lack inferential testing, and the network analysis reports no null-model comparison. These results support pattern detection more strongly than the proposed advantage mechanism.
↳ Analysis, citation comparison; Robustness Check 2, Table 4; Robustness Check 3, Figure 7
The Abstract describes systemic collective-action failure enabling recurrent insider advantage, although the dataset contains accepted papers and observed relationships rather than full editorial decisions or rejected submissions. The paper's later causal disclaimers do not fully neutralise that headline inference.
↳ Abstract; Limitations and Future Research
The paper's strongest contribution is the construction of a rare link-level audit with explicit COI categories, temporal windows, and dual coding. The same-field comparator and within-author or within-person analyses add useful triangulation, but the 54-author selection does not establish that prevalence is conservative for the wider contributor population. Secondary evidence is also less decisive than the main descriptive audit because citation significance depends on post hoc exclusions, productivity changes lack inferential tests, and network structure is not compared with a null model. The Abstract's systemic insider-advantage framing therefore carries more certainty than the accepted-paper observational design supports.
Nabu’s assessment, alongside the field’s view.
Are you an author of this paper?
Sound3.5
Confidence mediumThe audit makes a clearly delineated empirical advance by examining 946 author-editor links and distinguishing dyadic conflicts from prior work on institutional concentration. Its contribution remains bounded by two FT50 journals, one comparator, and a cohort of 54 prolific authors.
“we present the first large-scale audit of editorial COIs in two FT50 journals”
Explicit temporal rules, dual coding, within-author comparisons, and three robustness analyses provide a competent observational design. Methodological Rigour is limited by purposive prolific-author selection, visibility-dependent manual coding, missing inferential tests for productivity changes, and no reported network null model.
“All coding was conducted independently by two authors and reconciled through discussion”
The paper follows a traceable progression from governance theory through audit results and proposed reforms, with repeated disclaimers concerning intent and causality. The Abstract nevertheless states the systemic insider-advantage interpretation more decisively than accepted-paper metadata can establish.
“discretion and opacity enable recurrent insider advantage”
The paper relates the audit to editorial-integrity, publishing-governance, agency, and commons literatures and traces several limitations to narrower interpretations. Contextual balance is reduced because benign assignment mechanisms such as topical fit receive less sustained treatment than elite capture.
“we cannot conclusively disentangle these dynamics from broader processes”
Caveats4 of 4 checks
The principal prevalence counts reconcile, and no internal contradiction changes the central descriptive findings. Secondary outcome reporting warrants caution because citation significance emerges only after post hoc outlier exclusions and the productivity comparison lacks inferential testing.
Ethics approval, data-use constraints, coding procedures, and AI-assisted checks are disclosed. No explicit declaration of the study authors' own conflicts of interest appears in the supplied declarations material.
Flags: 1 declared / 5 total
80 of 80 checkable references verified
99 references in manuscript 19 have no canonical index record — counted, but not index-checkable 3 references confirmed by manual review
No retraction notice found in Retraction Watch.
Sources: Retraction Watch ✓
Where this paper’s evidence sits on the path from initial observation to real-world use.
The audit framework has been applied to actual journal metadata and is described sufficiently for replication. The wider governance package remains indicative and has not been piloted or evaluated in an editorial system.
“these should be understood as indicative rather than exhaustive”
AI-generated, human-governed. Something look off? Contact us to request a review.