The “collaboration premium,” higher citations for international research teams, is a foundational finding in scientometrics. However, existing studies conflate geographic diversity (where authors work) with diversity in authors' backgrounds, and leave unresolved whether the premium reflects superior novelty or wider audience reach. We address these gaps by analyzing 139,560 publications (2021–2025) across six policy-priority research domains (AI, Quantum, Space, Management, Sustainability, Wellbeing). Using a validated LLM-based method to infer the regional origin associated with author names, we separate Geographic Diversity (Single/Domestic/International Institution) from Regional-Origin Diversity (Single/Multiple Origins) and examine their associations with citation visibility at three levels: individual papers, national portfolios, and topic clusters. Three findings emerge. First, teams that are both internationally collaborative and diverse in regional origin are 1.6–2.8 times more likely than single-institution, single-origin teams to produce a Top 10% cited paper; the premium persists when teams of the same size are compared, and in five of six domains it is then statistically indistinguishable from the sum of the two single-dimension premiums. Second, at the country level, both dimensions correlate positively with citations, significantly so in five of six domains. Third, the associations are level-dependent: within topic clusters, regional-origin diversity is positively associated with citations in every domain, whereas between clusters the association is weaker and negative in sign in four domains, although not distinguishable from zero—a pattern consistent with a Simpson's paradox. These descriptive findings are consistent with an audience-based interpretation of the collaboration premium and suggest that policies aiming to enhance research visibility may benefit from attending both to international partnerships and to the diversity of the domestic research workforce.
Global university rankings increasingly influence higher education policy and institutional benchmarking, yet major ranking systems differ substantially in their indicators, weighting, normalization, and aggregation procedures. This raises questions about the robustness of conclusions drawn from a single league table. The issue is particularly relevant to Saudi Arabia, where the Human Capability Development Program under Saudi Vision 2030 establishes targets for increasing the number of Saudi universities among the global top 100 and top 200 and references QS, Times Higher Education (THE), and Shanghai Ranking. This study examines the consistency of QS, THE, and the Academic Ranking of World Universities (ARWU) in representing Saudi universities over the aligned 2022–2025 period. Cross-ranking coverage and positional agreement were assessed using Jaccard overlap, Spearman correlation, and Kendall correlation. Consensus rankings were constructed using Borda, median-rank, and Copeland aggregation, complemented by cross-ranking profile and Pareto assessment and sensitivity analyses under alternative weighting and rank-band specifications. QS and THE generally showed stronger positional agreement, reaching a Spearman correlation of 0.932 in 2025, whereas QS-ARWU agreement in the same year was weaker (ρ = 0.437). The three aggregation procedures produced broadly concordant consensus structures, although the leading position was method-sensitive in 2025: Borda ranked King Saud University first, whereas median-rank and Copeland ranked King Fahd University of Petroleum and Minerals first. Cross-ranking profile analysis further showed that King Abdulaziz University, King Saud University, and King Fahd University of Petroleum and Minerals were simultaneously non-dominated on the Pareto frontier in 2024 and 2025, indicating that a single consensus ordering can conceal differences across ranking-system outputs. Policy-threshold analysis likewise produced different Top-100 and Top-200 counts and institutional identities across the three systems. Overall, QS, THE, and ARWU provided related but non-equivalent representations of university standing. Combining cross-ranking comparison, rank aggregation, and cross-ranking profile assessment can therefore support more transparent monitoring and interpretation of ranking-based institutional and policy outcomes under Saudi Vision 2030.
IntroductionAcademic visibility is increasingly shaped by inclusion in major bibliometric databases, yet the extent to which these systems accurately represent scholarly contributions from small academic communities remains contested. This study examines how database selectivity, language hierarchies, and citation inequality influence the visibility of legal scholarship in a small-state context.MethodsUsing a comparative bibliometric analysis of Google Scholar, Scopus, and Web of Science, the study focuses on three leading law faculties in North Macedonia: Ss. Cyril and Methodius University in Skopje (UKIM), Goce Delcev University (UGD), and South East European University (SEEU).ResultsThe findings reveal a layered structure of academic visibility shaped by differences in database coverage and indexing practices. While Google Scholar captures a broad spectrum of scholarly output, Scopus and Web of Science provide increasingly selective representations, resulting in substantially lower levels of measurable research activity. Across the analyzed institutions, citation impact is unevenly distributed, with a small number of researchers accounting for a disproportionate share of total citations, although the precision of inequality estimates varies across institution-database combinations. Publication language emerges as an important factor associated with differences in citation visibility. English-language publications account for the dominant share of citations across the analyzed databases, whereas Macedonian- and Albanian-language outputs show substantially lower citation visibility despite their presence in overall publication volume.DiscussionThe findings suggest that international visibility is associated not only with research productivity, but also with linguistic and structural characteristics embedded within contemporary systems of scholarly communication. Academic visibility in small-state contexts is socially and institutionally constructed through the interaction of database inclusion, language dominance, and cumulative citation dynamics. The study contributes to debates on research evaluation and academic publishing by demonstrating how bibliometric systems may produce partial representations of scholarly activity, particularly in disciplines characterized by strong local and national orientations. The findings highlight the need for more context-sensitive approaches to assessing research impact in peripheral academic systems.
