Source-linked AI summary
Scaling rules in the science system: influence of field-specific citation characteristics on the impact of research groups
Anthony F. J. van Raan
TL;DR
The paper addresses limited understanding of how bibliometric indicators cohere within science as an interconnected system and how research-group size effects vary across fields. It represents science as a citation-density landscape and analyzes verified chemistry-group data while distinguishing performance levels. The analysis finds that size-related citation advantages depend on field citation density, with convergence at very large group sizes.
Problem
The paper examines limited knowledge about the mutual coherence and statistical properties of bibliometric indicators in an interconnected science system, including size effects across field citation densities.
Method
The study analyzes about 18,000 publications and 175,000 external citations from 157 chemistry groups using a citation-density landscape with FCSm as its coarse-scale parameter and JCSm as a finer attribute.
Results
Citation–size scaling is disadvantageous for high field-density groups (α = 0.84) but cumulatively advantageous for low field-density groups (α = 1.41), with impacts converging near P ~ 1,000.
Takeaways & Limitations
Field citation density is important for interpreting research-group impact and whether larger publication output dilutes average citations per publication.
Takeaways & Limitations
Research-group data are difficult to obtain externally, so the study relies on evaluation-study data based on strict verification.
Abstract
from arXiv · showhide
We propose a representation of science as a citation-density landscape and investigate scaling rules with the field-specific citation density as a main topological property. We focus on the size-dependence of several main bibliometric indicators for a large set of research groups while distinguishing between top-performance and lower performance groups. We demonstrate that this representation of the science system is particularly effective to understand the role and the interdependencies of the different bibliometric indicators and related topological properties of the landscape.
Introduction
The paper treats science as an interconnected system whose field-specific citation densities provide a landscape for studying bibliometric indicators and research-group performance. It examines how indicator size-dependence varies across citation-density regions and performance levels.
- Introduction: Bibliometric research has extensively constructed indicators, but their mutual coherence and statistical properties within science as an interconnected system remain little studied.
- Introduction: The study extends analysis of size-dependent cumulative advantage by comparing bibliometric indicators across field-specific citation densities and performance levels.
- Data, Indicators, Citation-Density Landscape: The dataset covers about 18,000 publications and 175,000 external citations from 157 chemistry groups at ten Dutch universities during 1991–2000.
- Data, Indicators, Citation-Density Landscape: The citation-density landscape treats FCSm as the coarse-scale topological parameter and JCSm as a finer, more group-specific attribute.
- Influence of field-specific citation density and journal impact: High field-density groups have citation–size exponent α = 0.84, whereas low field-density groups have α = 1.41, indicating disadvantage versus cumulative advantage, respectively.
- Influence of field-specific citation density and journal impact: At approximately P ~ 1,000, high- and low-density groups converge near CPP ~ 15 for top performers and CPP ~ 10 for lower performers.
CPPb ~ P+0.4
The paper derives a further relationship between average citations per publication in bottom and top field-density regions. This relationship follows from the preceding equations.
- CPPb ~ P+0.4: The preceding equations imply a relationship between CPPb and CPPt.
- CPPb ~ P+0.4: The derived comparison is presented as a first approximation.
CPPb ~ (CPPt)-2
The paper links research-group size to citations, citations per publication, field citation density, and journal impact across high- and low-density fields and performance groups. Larger groups in low-density fields gain cumulative citation advantage, while high-density groups show dilution and lower journal impact as size increases.
- Larger publication counts dilute average CPP in high-density fields but not in low-density fields.This pattern holds across the reported field-density comparison.
- Low-density groups gain CPP with size regardless of performance, while top- and lower-performance groups have similar size dependence.High-performance groups retain higher CPP, but their size dependence does not differ much from lower-performance groups.
- FCSm remains nearly unchanged with size in high-density regions but increases with size in low-density regions, with no significant performance-group difference.The low-density pattern is described as expansion toward regions with higher field citation density.
- Larger groups in high-density fields have lower average JCSm, whereas larger groups in low-density fields have higher average JCSm.The high-density decline is particularly associated with high-performance groups; lower-performance groups remain in lower-impact journals.
- α = 1.50 cumulative advantage appears only for low-field-density groups with low average journal impact.High average journal impact fields show no cumulative advantage with size.
- The landscape model treats FCSm as the basic scaling relation with size from which citation-count scaling can be deduced.The paper frames FCSm as a coarse-scale topological property and JCSm as finer-scale tuning within it.